OCI Identity and Access Management

Listen to Lois Houston and Nikita Abraham, along with Senior Principal Product Manager Wes Prichard, as they explore the five core components of OCI AI services: language, speech, vision, document understanding, and anomaly detection, to help you make better sense of all that unstructured data around you.

Oracle MyLearn: https://mylearn.oracle.com/ou/learning-path/become-an-oci-ai-foundations-associate-2023/127177

Oracle University Learning Community: https://education.oracle.com/ou-community

LinkedIn: https://www.linkedin.com/showcase/oracle-university/

X (formerly Twitter): https://twitter.com/Oracle_Edu

Special thanks to Arijit Ghosh, David Wright, Himanshu Raj, and the OU Studio Team for helping us create this episode.

--------------------------------------------------------

Episode Transcript:

00:00

Welcome to the Oracle University Podcast, the first stop on your cloud journey. During this series of informative podcasts, we’ll bring you foundational training on the most popular
Oracle technologies. Let’s get started!

00:26

Nikita: Welcome to the Oracle University Podcast! I’m Nikita Abraham, Principal Technical Editor with Oracle University, and with me is Lois Houston, Director of Innovation Programs.

Lois: Hi there! In our last episode, we spoke about OCI AI Portfolio, including AI and ML services, and the OCI AI infrastructure.

Nikita: Yeah, and in today’s episode, we’re going to continue down a similar path and take a closer look at OCI AI services.

00:55

Lois: With us today is Senior Principal Product Manager, Wes Prichard. Hi Wes! It’s lovely to have you here with us. Hemant gave us a broad overview of the various OCI AI services last week, but we’re really hoping to get into each of them with you. So, let’s jump right in and start with the OCI Language service. What can you tell us about it?

Wes: OCI Language analyzes unstructured text for you. It provides models trained on industry data to perform language analysis with no data science experience needed. 

01:27

Nikita: What kind of big things can it do?

Wes: It has five main capabilities. First, it detects the language of the text. It recognizes 75 languages, from Afrikaans to Welsh. 
It identifies entities, things like names, places, dates, emails, currency, organizations, phone numbers--14 types in all. It identifies the sentiment of the text, and not just one sentiment for the entire block of text, but the different sentiments for different aspects. 

01:56

Nikita: What do you mean by that, Wes?

Wes: So let's say you read a restaurant review that said, the food was great, but the service sucked. You'll get food with a positive sentiment and service with a negative sentiment. And it also analyzes the sentiment for every sentence. 

Lois: Ah, that’s smart. Ok, so we covered three capabilities. What else?

Wes: It identifies key phrases in the text that represent the important ideas or subjects. And it classifies the general topic of the text from a list of 600 categories and subcategories. 

02:27

Lois: Ok, and then there’s the OCI Speech service... 

Wes: OCI Speech is very straightforward. It locks the data in audio tracks by converting speech to text. Developers can use Oracle's time-tested acoustic language models to provide highly accurate transcription for audio or video files across multiple languages. 

OCI Speech automatically transcribes audio and video files into text using advanced deep learning techniques. There's no data science experience required. It processes data directly in object storage. And it generates timestamped, grammatically accurate transcriptions. 

03:01

Nikita: What are some of the main features of OCI Speech?

Wes: OCI Speech supports multiple languages, specifically English, Spanish, and Portuguese, with more coming in the future. It has batching support where multiple files can be submitted with a single call. It has blazing fast processing. It can transcribe hours of audio in less than 10 minutes. It does this by chunking up your audio into smaller segments, and transcribing each segment, and then joining them all back together into a single file. It provides a confidence score, both per word and per transcription. It punctuates transcriptions to make the text more readable and to allow downstream systems to process the text with less friction. 

And it has SRT file support. 

03:45

Lois: SRT? What’s that?

Wes: SRT is the most popular closed caption output file format. And with this SRT support, users can add closed captions to their video. OCI Speech makes transcribed text more readable to resemble how humans write. This is called normalization. And the service will normalize things like addresses, times, numbers, URLs, and more. 

It also does profanity filtering, where it can either remove, mask, or tag profanity and output text, where removing replaces the word with asterisks, and masking does the same thing, but it retains the first letter, and tagging will leave the word in place, but it provides tagging in the output data. 

04:29

Nikita: And what about OCI Vision? What are its capabilities?

Wes: Vision is a computed vision service that works on images, and it provides two main capabilities-- image analysis and document AI. Image analysis analyzes photographic images. Object detection is the feature that detects objects inside an image using a bounding box and assigning a label to each object with an accuracy percentage. Object detection also locates and extracts text that appears in the scene, like on a sign. 
Image classification will assign classification labels to the image by identifying the major features in the scene. One of the most powerful capabilities of image analysis is that, in addition to pretrained models, users can retrain the models with their own unique data to fit their specific needs. 

05:20

Lois: So object detection and image classification are features of image analysis. I think I got it! So then what’s document AI? 
Wes: It's used for working with document images. You can use it to understand PDFs or document image types, like JPEG, PNG, and Tiff, or photographs containing textual information. 

05:40

Lois: And what are its most important features?

Wes: The features of document AI are text recognition, also known as OCR or optical character recognition. 
And this extracts text from images, including non-trivial scenarios, like handwritten texts, plus tilted, shaded, or rotated documents. Document classification classifies documents into 10 different types based on visual appearance, high-level features, and extracted keywords. This is useful when you need to process a document, based on its classification, like an invoice, a receipt, or a resume. 

Language detection analyzes the visual features of text to determine the language rather than relying on the text itself. Table extraction identifies tables in docs and extracts their content in tabular form. Key value extraction finds values for 13 common fields and line items in receipts, things like merchant name and transaction date. 

06:41

Want to get the inside scoop on Oracle University? Head over to the Oracle University Learning Community. Attend exclusive events. Read up on the latest news. Get first-hand access to new products. Read the OU Learning Blog. Participate in Challenges. And stay up-to-date with upcoming certification opportunities.

Visit mylearn.oracle.com to get started. 

07:06

Nikita: Welcome back! Wes, I want to ask you about OCI Anomaly Detection. We discussed it a bit last week and it seems like such an intelligent and efficient service.

Wes: Oracle Cloud Infrastructure Anomaly Detection identifies anomalies in time series data. Equipment sensors generate time series data, but all kinds of business metrics are also time-based. The unique feature of this service is that it finds anomalies, not just in a single signal, but across many signals at once. That's important because machines often generate multiple signals at once and the signals are often related. 

07:42

Nikita: Ok you need to give us an example of this!

Wes: Think of a pump that has an output pressure, a flow rate, an RPM, and an electrical current draw. When a pump's going to fail, anomalies may appear across several of those signals but at different times. OCI Anomaly Detection helps you to identify anomalies in a multivariate data set by taking advantage of the interrelationship among signals. 

The service contains algorithms for both multi-signal, as in multivariate, single signal, as in univariate anomaly detection, and it automatically determines which algorithm to use based on the training data provided. The multivariate algorithm is called MSET-2, which stands for Multivariate State Estimation technique, and it's unique to Oracle. 

08:28

Lois: And the 2?

Wes: The 2 in the name refers to the patented enhancements by Oracle labs that automatically identify and fix data quality issues resulting in fewer false alarms and more accurate results. 
Now unlike some of the other AI services, OCI Anomaly Detection is always trained on the customer's data. It's trained using actual historical data with no anomalies, and there can be as many different trained models as needed for different sets of signals. 

08:57

Nikita: So where would one use a service like this?

Wes: One of the most obvious applications of this service is for predictive maintenance. Early warning of a problem provides the opportunity to deploy maintenance resources and schedule downtime to minimize disruption to the business. 

09:12

Lois: How would you train an OCI Anomaly Detection model?

Wes: It's a simple four-step process to prepare a model that can be used for anomaly detection. The first step is to obtain training data from the system to be monitored. The data must contain no anomalies and should cover the normal range of values that would be experienced in a full business cycle. 
Second, the training data file is uploaded to an object storage bucket. 

Third, a data set is created for the training data. So a data set in this context is an object in the OCI Anomaly Detection service to manage data used for training and testing models. 

And fourth, the model is trained. A wizard in the user interface steps the user through the required inputs, such as the training data set and some training parameters like the target false alarm probability. 

10:02

Lois: How would this service know about the data and whether the trained model is univariate or multivariate?

Wes: When training OCI Anomaly Detection models, the user does not need to specify whether the intended model is for multivariate or univariate data. It does this detection automatically. 

For example, if a model is trained with 10 signals and 5 of those signals are determined to be correlated enough for multivariate anomaly detection, it will create an internal multivariate model for those signals. If the other five signals are not correlated with each other, it will create an internal univariate model for each one. 

From the user's perspective, the result will be a single OCI anomaly detection model for the 10 signals. But internally, the signals are treated differently based on the training. A user can also train a model on a single signal and it will result in a univariate model. 

10:55

Lois: What does this OCI Anomaly Detection model training entail? How does it ensure that it does not have any false alarms?

Wes: Training a model requires a single data file with no anomalies that should cover a complete business cycle, which means it should represent all the normal variations in the signal. During training, OCI Anomaly Detection will use a portion of the data for training and another portion for automated testing. The fraction used for each is specified when the model is trained. 
When model training is complete, it's best practice to do another test of the model with a data set containing anomalies to see if the anomalies are detected and if there are any false alarms. Based on the outcome, the user may want to retrain the model and specify a different false alarm probability, also called F-A-P or FAP. The FAP is the probability that the model would produce a false alarm. The false alarm probability can be thought of as the sensitivity of the model. The lower the false alarm probability, the less likelihood of it reporting a false alarm, but the less sensitive it will be to detecting anomalies. Selecting the right FAP is a business decision based on the need for sensitive detections balanced by the ability to tolerate false alarms. 

Once a model has been trained and the user is satisfied with its detection performance, it can then be used for inferencing. 

12:23

Nikita: Inferencing? Is that what I think it is? 

Wes: New data is submitted to the model and OCI Anomaly Detection will respond with anomalies that are detected. The input data must contain the same signals that the model was trained on. So, for example, if the model was trained on signals A, B, C, and D, then for detection inferencing, the same four signals must be provided. No more, no less.

12:46

Lois: Where can I find the features of OCI Anomaly Detection that you mentioned? 

Wes: The training and inferencing features of OCI Anomaly Detection can be accessed through the OCI console. However, a human-driven interface is not efficient for most business scenarios. 

In most cases, automating the detection of anomalies through software is preferred to be able to process hundreds or thousands of signals using many trained models. The service provides multiple software interfaces for this purpose. 
Each trained model is accessible through a REST API and an HTTP endpoint. Additionally, programming language-specific SDKs are available for multiple languages, including Python. Using the Python SDK, data scientists can work with OCI Anomaly Detection for both training and inferencing in an OCI Data Science notebook. 

13:37

Nikita: How can a data scientist take advantage of these capabilities? 

Wes: Well, you can write code against the REST API or use any of the various language SDKs. But for data scientists working in OCI Data Science, it makes sense to use Python. 

13:51

Lois: That’s exciting! What does it take to use the Python SDK in a notebook… to be able to use the AI services?

Wes: You can use a Notebook session in OCI Data Science to invoke the SDK for any of the AI services. 

This might be useful to generate new features for a custom model or simply as a way to consume the service using a familiar Python interface. But before you can invoke the SDK, you have to prepare the data science notebook session by supplying it with an API Signing Key. 

Signing Key is unique to a particular user and tenancy and authenticates that user to OCI when invoking the SDK. So therefore, you want to make sure you safeguard your Signing Key and never share it with another user. 

14:34

Nikita: And where would I get my API Signing Key?

Wes: You can obtain an API Signing Key from your user profile in the OCI Console. Then you save that key as a file to your local machine. 

The API Signing Key also provides commands to be added to a config file that the SDK expects to find in the environment, where the SDK code is executing. The config file then references the key file. Once these files are prepared on your local machine, you can upload them to the Notebook session, where you will execute SDK code for the AI service. 
The API Signing Key and config file can be reused with any of your notebook sessions, and the same files also work for all of the AI services. So, the files only need to be created once for each user and tenancy combination. 

15:27

Lois: Thank you so much, Wes, for this really insightful discussion. To learn more about the topics covered today, you can visit mylearn.oracle.com and search for the Oracle Cloud Infrastructure AI Foundations course.

Nikita: And remember, that course prepares you for the Oracle Cloud Infrastructure AI Foundations Associate certification that you can take for free! So, don’t wait too long to check it out. Join us next week for another episode of the Oracle University Podcast. Until then, this is Nikita Abraham…

Lois Houston: And Lois Houston, signing off!

16:03

That’s all for this episode of the Oracle University Podcast. If you enjoyed listening, please click Subscribe to get all the latest episodes. We’d also love it if you would take a moment to rate and review us on your podcast app. See you again on the next episode of the Oracle University Podcast.

Oracle has been actively focusing on bringing AI to the enterprise at every layer of its tech stack, be it SaaS apps, AI services, infrastructure, or data.

In this episode, hosts Lois Houston and Nikita Abraham, along with senior instructors Hemant Gahankari and Himanshu Raj, discuss OCI AI and Machine Learning services. They also go over some key OCI Data Science concepts and responsible AI principles.

Oracle MyLearn: https://mylearn.oracle.com/ou/learning-path/become-an-oci-ai-foundations-associate-2023/127177

Oracle University Learning Community: https://education.oracle.com/ou-community

LinkedIn: https://www.linkedin.com/showcase/oracle-university/

X (formerly Twitter): https://twitter.com/Oracle_Edu

Special thanks to Arijit Ghosh, David Wright, Himanshu Raj, and the OU Studio Team for helping us create this episode.

-------------------------------------------------------

Episode Transcript:

00:00

Welcome to the Oracle University Podcast, the first stop on your cloud journey. During this series of informative podcasts, we’ll bring you foundational training on the most popular
Oracle technologies. Let’s get started!

00:26

Lois: Welcome to the Oracle University Podcast! I’m Lois Houston, Director of Innovation Programs with Oracle University, and with me is Nikita Abraham, Principal Technical Editor.

Nikita: Hey everyone! In our last episode, we dove into Generative AI and Language Learning Models. 

Lois: Yeah, that was an interesting one. But today, we’re going to discuss the AI and machine learning services offered by Oracle Cloud Infrastructure, and we’ll look at the OCI AI infrastructure.

Nikita: I’m also going to try and squeeze in a couple of questions on a topic I’m really keen about, which is responsible AI. To take us through all of this, we have two of our colleagues, Hemant Gahankari and Himanshu Raj. Hemant is a Senior Principal OCI Instructor and Himanshu is a Senior Instructor on AI/ML. So, let’s get started!

01:16

Lois: Hi Hemant! We’re so excited to have you here! We know that Oracle has really been focusing on bringing AI to the enterprise at every layer of our stack. 

Hemant: It all begins with data and infrastructure layers. OCI AI services consume data, and AI services, in turn, are consumed by applications. 

This approach involves extensive investment from infrastructure to SaaS applications. Generative AI and massive scale models are the more recent steps. Oracle AI is the portfolio of cloud services for helping organizations use the data they may have for the business-specific uses. 

Business applications consume AI and ML services. The foundation of AI services and ML services is data. AI services contain pre-built models for specific uses. Some of the AI services are pre-trained, and some can be additionally trained by the customer with their own data. 

AI services can be consumed by calling the API for the service, passing in the data to be processed, and the service returns a result. There is no infrastructure to be managed for using AI services. 

02:37

Nikita: How do I access OCI AI services?

Hemant: OCI AI services provide multiple methods for access. The most common method is the OCI Console. The OCI Console provides an easy to use, browser-based interface that enables access to notebook sessions and all the features of all the data science, as well as AI services. 

The REST API provides access to service functionality but requires programming expertise. And API reference is provided in the product documentation. OCI also provides programming language SDKs for Java, Python, TypeScript, JavaScript, .Net, Go, and Ruby. The command line interface provides both quick access and full functionality without the need for scripting. 

03:31

Lois: Hemant, what are the types of OCI AI services that are available? 

Hemant: OCI AI services is a collection of services with pre-built machine learning models that make it easier for developers to build a variety of business applications. The models can also be custom trained for more accurate business results. The different services provided are digital assistant, language, vision, speech, document understanding, anomaly detection. 

04:03

Lois: I know we’re going to talk about them in more detail in the next episode, but can you introduce us to OCI Language, Vision, and Speech?

Hemant: OCI Language allows you to perform sophisticated text analysis at scale. Using the pre-trained and custom models, you can process unstructured text to extract insights without data science expertise. Pre-trained models include language detection, sentiment analysis, key phrase extraction, text classification, named entity recognition, and personal identifiable information detection. 

Custom models can be trained for named entity recognition and text classification with domain-specific data sets. In text translation, natural machine translation is used to translate text across numerous languages. 

Using OCI Vision, you can upload images to detect and classify objects in them. Pre-trained models and custom models are supported. In image analysis, pre-trained models perform object detection, image classification, and optical character recognition. In image analysis, custom models can perform custom object detection by detecting the location of custom objects in an image and providing a bounding box. 
The OCI Speech service is used to convert media files to readable texts that's stored in JSON and SRT format. Speech enables you to easily convert media files containing human speech into highly exact text transcriptions. 

05:52

Nikita: That’s great. And what about document understanding and anomaly detection?

Hemant: Using OCI document understanding, you can upload documents to detect and classify text and objects in them. You can process individual files or batches of documents. In OCR, document understanding can detect and recognize text in a document. In text extraction, document understanding provides the word level and line level text, and the bounding box, coordinates of where the text is found. 

In key value extraction, document understanding extracts a predefined list of key value pairs of information from receipts, invoices, passports, and driver IDs. In table extraction, document understanding extracts content in tabular format, maintaining the row and column relationship of cells. In document classification, the document understanding classifies documents into different types. 

The OCI Anomaly Detection service is a service that analyzes large volume of multivariate or univariate time series data. The Anomaly Detection service increases the reliability of businesses by monitoring their critical assets and detecting anomalies early with high precision. Anomaly Detection is the identification of rare items, events, or observations in data that differ significantly from the expectation. 

07:34

Nikita: Where is Anomaly Detection most useful?

Hemant: The Anomaly Detection service is designed to help with analyzing large amounts of data and identifying the anomalies at the earliest possible time with maximum accuracy. Different sectors, such as utility, oil and gas, transportation, manufacturing, telecommunications, banking, and insurance use Anomaly Detection service for their day-to-day activities. 

08:02

Lois: Ok.. and the first OCI AI service you mentioned was digital assistant…

Hemant: Oracle Digital Assistant is a platform that allows you to create and deploy digital assistants, which are AI driven interfaces that help users accomplish a variety of tasks with natural language conversations. When a user engages with the Digital Assistant, the Digital Assistant evaluates the user input and routes the conversation to and from the appropriate skills. 
Digital Assistant greets the user upon access. Upon user requests, list what it can do and provide entry points into the given skills. It routes explicit user requests to the appropriate skills. And it also handles interruptions to flows and disambiguation. It also handles requests to exit the bot. 

09:00

Nikita: Excellent! Let’s bring Himanshu in to tell us about machine learning services. Hi Himanshu! Let’s talk about OCI Data Science. Can you tell us a bit about it?

Himanshu: OCI Data Science is the cloud service focused on serving the data scientist throughout the full machine learning life cycle with support for Python and open source. 

The service has many features, such as model catalog, projects, JupyterLab notebook, model deployment, model training, management, model explanation, open source libraries, and AutoML. 

09:35
Lois: Himanshu, what are the core principles of OCI Data Science? 

Himanshu: There are three core principles of OCI Data Science. The first one, accelerated. The first principle is about accelerating the work of the individual data scientist. OCI Data Science provides data scientists with open source libraries along with easy access to a range of compute power without having to manage any infrastructure. It also includes Oracle's own library to help streamline many aspects of their work. 
The second principle is collaborative. It goes beyond an individual data scientist’s productivity to enable data science teams to work together. This is done through the sharing of assets, reducing duplicative work, and putting reproducibility and auditability of models for collaboration and risk management. 

Third is enterprise grade. That means it's integrated with all the OCI Security and access protocols. The underlying infrastructure is fully managed. The customer does not have to think about provisioning compute and storage. And the service handles all the maintenance, patching, and upgrades so user can focus on solving business problems with data science. 

10:50

Nikita: Let’s drill down into the specifics of OCI Data Science. So far, we know it’s cloud service to rapidly build, train, deploy, and manage machine learning models. But who can use it? Where is it? And how is it used?

Himanshu: It serves data scientists and data science teams throughout the full machine learning life cycle. 

Users work in a familiar JupyterLab notebook interface, where they write Python code. And how it is used? So users preserve their models in the model catalog and deploy their models to a managed infrastructure. 

11:25

Lois: Walk us through some of the key terminology that’s used.

Himanshu: Some of the important product terminology of OCI Data Science are projects. The projects are containers that enable data science teams to organize their work. They represent collaborative work spaces for organizing and documenting data science assets, such as notebook sessions and models. 

Note that tenancy can have as many projects as needed without limits. Now, this notebook session is where the data scientists work. Notebook sessions provide a JupyterLab environment with pre-installed open source libraries and the ability to add others. Notebook sessions are interactive coding environment for building and training models. 

Notebook sessions run in a managed infrastructure and the user can select CPU or GPU, the compute shape, and amount of storage without having to do any manual provisioning. The other important feature is Conda environment. It's an open source environment and package management system and was created for Python programs. 

12:33

Nikita: What is a Conda environment used for?

Himanshu: It is used in the service to quickly install, run, and update packages and their dependencies. Conda easily creates, saves, loads, and switches between environments in your notebooks sessions.

12:46

Nikita: Earlier, you spoke about the support for Python in OCI Data Science. Is there a dedicated library?

Himanshu: Oracle's Accelerated Data Science ADS SDK is a Python library that is included as part of OCI Data Science. 
ADS has many functions and objects that automate or simplify the steps in the data science workflow, including connecting to data, exploring, and visualizing data. Training a model with AutoML, evaluating models, and explaining models. In addition, ADS provides a simple interface to access the data science service mode model catalog and other OCI services, including object storage. 

13:24

Lois: I also hear a lot about models. What are models?

Himanshu: Models define a mathematical representation of your data and business process. You create models in notebooks, sessions, inside projects. 

13:36

Lois: What are some other important terminologies related to models?

Himanshu: The next terminology is model catalog. The model catalog is a place to store, track, share, and manage models. 
The model catalog is a centralized and managed repository of model artifacts. A stored model includes metadata about the provenance of the model, including Git-related information and the script. Our notebook used to push the model to the catalog. Models stored in the model catalog can be shared across members of a team, and they can be loaded back into a notebook session. 

The next one is model deployments. Model deployments allow you to deploy models stored in the model catalog as HTTP endpoints on managed infrastructure. 

14:24

Lois: So, how do you operationalize these models?

Himanshu: Deploying machine learning models as web applications, HTTP API endpoints, serving predictions in real time is the most common way to operationalize models. HTTP endpoints or the API endpoints are flexible and can serve requests for the model predictions. Data science jobs enable you to define and run a repeatable machine learning tasks on fully managed infrastructure. 

Nikita: Thanks for that, Himanshu. 

14:57

Did you know that Oracle University offers free courses on Oracle Cloud Infrastructure? You’ll find training on everything from cloud computing, database, and security, artificial intelligence, and machine learning, all free to subscribers. So, what are you waiting for? Pick a topic, leverage the Oracle University Learning Community to ask questions, and then sit for your certification.

Visit mylearn.oracle.com to get started. 

15:25

Nikita: Welcome back! The Oracle AI Stack consists of AI services and machine learning services, and these services are built using AI infrastructure. So, let’s move on to that. Hemant, what are the components of OCI AI Infrastructure?
Hemant: OCI AI Infrastructure is mainly composed of GPU-based instances. Instances can be virtual machines or bare metal machines. High performance cluster networking that allows instances to communicate to each other. Super clusters are a massive network of GPU instances with multiple petabytes per second of bandwidth. And a variety of fully managed storage options from a single byte to exabytes without upfront provisioning are also available. 

16:14

Lois: Can we explore each of these components a little more? First, tell us, why do we need GPUs?

Hemant: ML and AI needs lots of repetitive computations to be made on huge amounts of data. Parallel computing on GPUs is designed for many processes at the same time. A GPU is a piece of hardware that is incredibly good in performing computations. 
GPU has thousands of lightweight cores, all working on their share of data in parallel. This gives them the ability to crunch through extremely large data set at tremendous speed. 

16:54

Nikita: And what are the GPU instances offered by OCI?

Hemant: GPU instances are ideally suited for model training and inference. Bare metal and virtual machine compute instances powered by NVIDIA GPUs H100, A100, A10, and V100 are made available by OCI. 

17:14

Nikita: So how do we choose what to train from these different GPU options? 

Hemant: For large scale AI training, data analytics, and high performance computing, bare metal instances BM 8 X NVIDIA H100 and BM 8 X NVIDIA A100 can be used. 

These provide up to nine times faster AI training and 30 times higher acceleration for AI inferencing. The other bare metal and virtual machines are used for small AI training, inference, streaming, gaming, and virtual desktop infrastructure. 

17:53

Lois: And why would someone choose the OCI AI stack over its counterparts?

Hemant: Oracle offers all the features and is the most cost effective option when compared to its counterparts. 

For example, BM GPU 4.8 version 2 instance costs just $4 per hour and is used by many customers. 

Superclusters are a massive network with multiple petabytes per second of bandwidth. It can scale up to 4,096 OCI bare metal instances with 32,768 GPUs. 

We also have a choice of bare metal A100 or H100 GPU instances, and we can select a variety of storage options, like object store, or block store, or even file system. For networking speeds, we can reach 1,600 GB per second with A100 GPUs and 3,200 GB per second with H100 GPUs. 

With OCI storage, we can select local SSD up to four NVMe drives, block storage up to 32 terabytes per volume, object storage up to 10 terabytes per object, file systems up to eight exabyte per file system. OCI File system employs five replicated storage located in different fault domains to provide redundancy for resilient data protection. 

HPC file systems, such as BeeGFS and many others are also offered. OCI HPC file systems are available on Oracle Cloud Marketplace and make it easy to deploy a variety of high performance file servers. 

19:50

Lois: I think a discussion on AI would be incomplete if we don’t talk about responsible AI. We’re using AI more and more every day, but can we actually trust it?

Hemant: For us to trust AI, it must be driven by ethics that guide us as well.

Nikita: And do we have some principles that guide the use of AI?
Hemant: AI should be lawful, complying with all applicable laws and regulations. AI should be ethical, that is it should ensure adherence to ethical principles and values that we uphold as humans. And AI should be robust, both from a technical and social perspective. Because even with the good intentions, AI systems can cause unintentional harm.

AI systems do not operate in a lawless world. A number of legally binding rules at national and international level apply or are relevant to the development, deployment, and use of AI systems today. The law not only prohibits certain actions but also enables others, like protecting rights of minorities or protecting environment. Besides horizontally applicable rules, various domain-specific rules exist that apply to particular AI applications. For instance, the medical device regulation in the health care sector. 

In AI context, equality entails that the systems’ operations cannot generate unfairly biased outputs. And while we adopt AI, citizens right should also be protected. 

21:30

Lois: Ok, but how do we derive AI ethics from these?

Hemant: There are three main principles. 
AI should be used to help humans and allow for oversight. It should never cause physical or social harm. Decisions taken by AI should be transparent and fair, and also should be explainable. AI that follows the AI ethical principles is responsible AI. 

So if we map the AI ethical principles to responsible AI requirements, these will be like, AI systems should follow human-centric design principles and leave meaningful opportunity for human choice. This means securing human oversight. AI systems and environments in which they operate must be safe and secure, they must be technically robust, and should not be open to malicious use. 

The development, and deployment, and use of AI systems must be fair, ensuring equal and just distribution of both benefits and costs. AI should be free from unfair bias and discrimination. Decisions taken by AI to the extent possible should be explainable to those directly and indirectly affected. 

23:01

Nikita: This is all great, but what does a typical responsible AI implementation process look like? 

Hemant: First, a governance needs to be put in place. Second, develop a set of policies and procedures to be followed. And once implemented, ensure compliance by regular monitoring and evaluation. 

Lois: And this is all managed by developers?

Hemant: Typical roles that are involved in the implementation cycles are developers, deployers, and end users of the AI. 

23:35

Nikita: Can we talk about AI specifically in health care? How do we ensure that there is fairness and no bias?

Hemant: AI systems are only as good as the data that they are trained on. If that data is predominantly from one gender or racial group, the AI systems might not perform as well on data from other groups. 

24:00

Lois: Yeah, and there’s also the issue of ensuring transparency, right?

Hemant: AI systems often make decisions based on complex algorithms that are difficult for humans to understand. As a result, patients and health care providers can have difficulty trusting the decisions made by the AI. AI systems must be regularly evaluated to ensure that they are performing as intended and not causing harm to patients. 

24:29

Nikita: Thank you, Hemant and Himanshu, for this really insightful session. If you’re interested in learning more about the topics we discussed today, head on over to mylearn.oracle.com and search for the Oracle Cloud Infrastructure AI Foundations course. 

Lois: That’s right, Niki. You’ll find demos that you watch as well as skill checks that you can attempt to better your understanding. In our next episode, we’ll get into the OCI AI Services we discussed today and talk about them in more detail. Until then, this is Lois Houston…

Nikita: And Nikita Abraham, signing off!

25:05

That’s all for this episode of the Oracle University Podcast. If you enjoyed listening, please click Subscribe to get all the latest episodes. We’d also love it if you would take a moment to rate and review us on your podcast app. See you again on the next episode of the Oracle University Podcast.

In this week’s episode, Lois Houston and Nikita Abraham, along with Senior Instructor Himanshu Raj, take you through the extraordinary capabilities of Generative AI, a subset of deep learning that doesn’t make predictions but rather creates its own content.

They also explore the workings of Large Language Models.

Oracle MyLearn: https://mylearn.oracle.com/ou/learning-path/become-an-oci-ai-foundations-associate-2023/127177

Oracle University Learning Community: https://education.oracle.com/ou-community

LinkedIn: https://www.linkedin.com/showcase/oracle-university/

X (formerly Twitter): https://twitter.com/Oracle_Edu

Special thanks to Arijit Ghosh, David Wright, and the OU Studio Team for helping us create this episode.

--------------------------------------------------------

Episode Transcript:

00:00

Welcome to the Oracle University Podcast, the first stop on your cloud journey. During this series of informative podcasts, we’ll bring you foundational training on the most popular Oracle technologies. Let’s get started!

00:26

Lois: Hello and welcome to the Oracle University Podcast. I’m Lois Houston, Director of Innovation Programs with Oracle University, and with me is Nikita Abraham, Principal 
Technical Editor. 

Nikita: Hi everyone! In our last episode, we went over the basics of deep learning. Today, we’ll look at generative AI and large language models, and discuss how they work. To help us with that, we have Himanshu Raj, Senior Instructor on AI/ML. So, let’s jump right in. Hi Himanshu, what is generative AI? 

01:00

Himanshu: Generative AI refers to a type of AI that can create new content. It is a subset of deep learning, where the models are trained not to make predictions but rather to generate output on their own. 

Think of generative AI as an artist who looks at a lot of paintings and learns the patterns and styles present in them. Once it has learned these patterns, it can generate new paintings that resembles what it learned.

01:27

Lois: Let's take an example to understand this better. Suppose we want to train a generative AI model to draw a dog. How would we achieve this?

Himanshu: You would start by giving it a lot of pictures of dogs to learn from. The AI does not know anything about what a dog looks like. But by looking at these pictures, it starts to figure out common patterns and features, like dogs often have pointy ears, narrow faces, whiskers, etc. You can then ask it to draw a new picture of a dog. 

The AI will use the patterns it learned to generate a picture that hopefully looks like a dog. But remember, the AI is not copying any of the pictures it has seen before but creating a new image based on the patterns it has learned. This is the basic idea behind generative AI. In practice, the process involves a lot of complex maths and computation, and there are different techniques and architectures that can be used, such as variational autoencoders (VAs) and Generative Adversarial Networks (GANs). 

02:27

Nikita: Himanshu, where is generative AI used in the real world?

Himanshu: Generative AI models have a wide variety of applications across numerous domains. For the image generation, generative models like GANs are used to generate realistic images. They can be used for tasks, like creating artwork, synthesizing images of human faces, or transforming sketches into photorealistic images. 

For text generation, large language models like GPT 3, which are generative in nature, can create human-like text. This has applications in content creation, like writing articles, generating ideas, and again, conversational AI, like chat bots, customer service agents. They are also used in programming for code generation and debugging, and much more. 

For music generation, generative AI models can also be used. They create new pieces of music after being trained on a specific style or collection of tunes. A famous example is OpenAI's MuseNet.

03:21

Lois: You mentioned large language models in the context of text-based generative AI. So, let’s talk a little more about it. Himanshu, what exactly are large language models?

Himanshu: LLMs are a type of artificial intelligence models built to understand, generate, and process human language at a massive scale. They were primarily designed for sequence to sequence tasks such as machine translation, where an input sequence is transformed into an output sequence. 

LLMs can be used to translate text from one language to another. For example, an LLM could be used to translate English text into French. To do this job, LLM is trained on a massive data set of text and code which allows it to learn the patterns and relationships that exist between different languages. The LLM translates, “How are you?” from English to French, “Comment allez-vous?” 

It can also answer questions like, what is the capital of France? And it would answer the capital of France is Paris. And it will write an essay on a given topic. For example, write an essay on French Revolution, and it will come up with a response like with a title and introduction.

04:33

Lois: And how do LLMs actually work?

Himanshu: So, LLM models are typically based on deep learning architectures such as transformers. They are also trained on vast amount of text data to learn language patterns and relationships, again, with a massive number of parameters usually in order of millions or even billions. LLMs have also the ability to comprehend and understand natural language text at a semantic level. They can grasp context, infer meaning, and identify relationships between words and phrases. 

05:05

Nikita: What are the most important factors for a large language model?

Himanshu: Model size and parameters are crucial aspects of large language models and other deep learning models. They significantly impact the model’s capabilities, performance, and resource requirement. So, what is model size? The model size refers to the amount of memory required to store the model's parameter and other data structures. Larger model sizes generally led to better performance as they can capture more complex patterns and representation from the data. 

The parameters are the numerical values of the model that change as it learns to minimize the model's error on the given task. In the context of LLMs, parameters refer to the weights and biases of the model's transformer layers. Parameters are usually measured in terms of millions or billions. For example, GPT-3, one of the largest LLMs to date, has 175 billion parameters making it extremely powerful in language understanding and generation. 

Tokens represent the individual units into which a piece of text is divided during the processing by the model. In natural language, tokens are usually words, subwords, or characters. Some models have a maximum token limit that they can process and longer text can may require truncation or splitting. Again, balancing model size, parameters, and token handling is crucial when working with LLMs. 

06:29

Nikita: But what’s so great about LLMs?

Himanshu: Large language models can understand and interpret human language more accurately and contextually. They can comprehend complex sentence structures, nuances, and word meanings, enabling them to provide more accurate and relevant responses to user queries. This model can generate human-like text that is coherent and contextually appropriate. This capability is valuable for context creation, automated writing, and generating personalized response in applications like chatbots and virtual assistants. They can perform a variety of tasks. 

Large language models are very versatile and adaptable to various industries. They can be customized to excel in applications such as language translation, sentiment analysis, code generation, and much more. LLMs can handle multiple languages making them valuable for cross-lingual tasks like translation, sentiment analysis, and understanding diverse global content. 

Large language models can be again, fine-tuned for a specific task using a minimal amount of domain data. The efficiency of LLMs usually grows with more data and parameters.

07:34

Lois: You mentioned the “sequence to sequence tasks” earlier. Can you explain the concept in simple terms for us?

Himanshu: Understanding language is difficult for computers and AI systems. The reason being that words often have meanings based on context. Consider a sentence such as Jane threw the frisbee, and her dog fetched it. 

In this sentence, there are a few things that relate to each other. Jane is doing the throwing. The dog is doing the fetching. And it refers to the frisbee. Suppose we are looking at the word “it” in the sentence. As a human, we understand easily that “it” refers to the frisbee. But for a machine, it can be tricky.

The goal in sequence problems is to find patterns, dependencies, or relationships within the data and make predictions, classification, or generate new sequences based on that understanding.

08:27

Lois: And where are sequence models mostly used?

Himanshu: Some common example of sequence models includes natural language processing, which we call NLP, tasks such as machine translation, text generation sentiment analysis, language modeling involve dealing with sequences of words or characters. 

Speech recognition. Converting audio signals into text, involves working with sequences of phonemes or subword units to recognize spoken words. Music generation. Generating new music involves modeling musical sequences, nodes, and rhythms to create original compositions. 

Gesture recognition. Sequences of motion or hand gestures are used to interpret human movements for applications, such as sign language recognition or gesture-based interfaces. Time series analysis. In fields such as finance, economics, weather forecasting, and signal processing, time series data is used to predict future values, detect anomalies, and understand patterns in temporal data.

09:35

The Oracle University Learning Community is an excellent place to collaborate and learn with Oracle experts and fellow learners. Grow your skills, inspire innovation, and celebrate your successes. All your activities, from liking a post to answering questions and sharing with others, will help you earn a valuable reputation, badges, and ranks to be recognized in the community.

Visit mylearn.oracle.com to get started. 

10:03

Nikita: Welcome back! Himanshu, what would be the best way to solve those sequence problems you mentioned? Let’s use the same sentence, “Jane threw the frisbee, and her dog fetched it” as an example.

Himanshu: The solution is transformers. It's like model has a bird's eye view of the entire sentence and can see how all the words relate to each other. This allows it to understand the sentence as a whole instead of just a series of individual words. Transformers with their self-attention mechanism can look at all the words in the sentence at the same time and understand how they relate to each other. 

For example, transformer can simultaneously understand the connections between Jane and dog even though they are far apart in the sentence.

10:52

Nikita: But how?

Himanshu: The answer is attention, which adds context to the text. Attention would notice dog comes after frisbee, fetched comes after dog, and it comes after fetched. 

Transformer does not look at it in isolation. Instead, it also pays attention to all the other words in the sentence at the same time. But considering all these connections, the model can figure out that “it” likely refers to the frisbee. 

The most famous current models that are emerging in natural language processing tasks consist of dozens of transformers or some of their variants, for example, GPT or Bert.

11:32

Lois: I was looking at the AI Foundations course on MyLearn and came across the terms “prompt engineering” and “fine tuning.” Can you shed some light on them?

Himanshu: A prompt is the input or initial text provided to the model to elicit a specific response or behavior. So, this is something which you write or ask to a language model. Now, what is prompt engineering? So prompt engineering is the process of designing and formulating specific instructions or queries to interact with a large language model effectively. 
In the context of large language models, such as GPT 3 or Burt, prompts are the input text or questions given to the model to generate responses or perform specific tasks. 

The goal of prompt engineering is to ensure that the language model understands the user's intent correctly and provide accurate and relevant responses.

12:26

Nikita: That sounds easy enough, but fine tuning seems a bit more complex. Can you explain it with an example?

Himanshu: Imagine you have a versatile recipe robot named chef bot. Suppose that chef bot is designed to create delicious recipes for any dish you desire. 

Chef bot recognizes the prompt as a request for a pizza recipe, and it knows exactly what to do.

However, if you want chef bot to be an expert in a particular type of cuisine, such as Italian dishes, you fine-tune chef bot for Italian cuisine by immersing it in a culinary crash course filled with Italian cookbooks, traditional Italian recipes, and even Italian cooking shows. 

During this process, chef bot becomes more specialized in creating authentic Italian recipes, and this option is called fine tuning. LLMs are general purpose models that are pre-trained on large data sets but are often fine-tuned to address specific use cases. 

When you combine prompt engineering and fine tuning, and you get a culinary wizard in chef bot, a recipe robot that is not only great at understanding specific dish requests but also capable of following a specific dish requests and even mastering the art of cooking in a particular culinary style.

13:47

Lois: Great! Now that we’ve spoken about all the major components, can you walk us through the life cycle of a large language model?

Himanshu: The life cycle of a Large Language Model, LLM, involves several stages, from its initial pre-training to its deployment and ongoing refinement. 

The first of this lifecycle is pre-training. The LLM is initially pre-trained on a large corpus of text data from the internet. During pre-training, the model learns grammar, facts, reasoning abilities, and general language understanding. The model predicts the next word in a sentence given the previous words, which helps it capture relationships between words and the structure of language. 

The second phase is fine tuning initialization. After pre-training, the model's weights are initialized, and it's ready for task-specific fine tuning. Fine tuning can involve supervised learning on labeled data for specific tasks, such as sentiment analysis, translation, or text generation. 

The model is fine-tuned on specific tasks using a smaller domain-specific data set. The weights from pre-training are updated based on the new data, making the model task aware and specialized. The next phase of the LLM life cycle is prompt engineering. So this phase craft effective prompts to guide the model's behavior in generating specific responses. 

Different prompt formulations, instructions, or context can be used to shape the output. 

15:13

Nikita: Ok… we’re with you so far. What’s next?

Himanshu: The next phase is evaluation and iteration. So models are evaluated using various metrics to access their performance on specific tasks. Iterative refinement involves adjusting model parameters, prompts, and fine tuning strategies to improve results. 

So as a part of this step, you also do few shot and one shot inference. If needed, you further fine tune the model with a small number of examples. Basically, few shot or a single example, one shot for new tasks or scenarios. 

Also, you do the bias mitigation and consider the ethical concerns. These biases and ethical concerns may arise in models output. You need to implement measures to ensure fairness in inclusivity and responsible use. 

16:07

Himanshu: The next phase in LLM life cycle is deployment. Once the model has been fine-tuned and evaluated, it is deployed for real world applications. Deployed models can perform tasks, such as text generation, translation, summarization, and much more. You also perform monitoring and maintenance in this phase. 

So you continuously monitor the model's performance and output to ensure it aligns with desired outcomes. You also periodically update and retrain the model to incorporate new data and to adapt to evolving language patterns. This overall life cycle can also consist of a feedback loop, whether you gather feedbacks from users and incorporate it into the model’s improvement process. 

You use this feedback to further refine prompts, fine tuning, and overall model behavior. RLHF, which is Reinforcement Learning with Human Feedback, is a very good example of this feedback loop. You also research and innovate as a part of this life cycle, where you continue to research and develop new techniques to enhance the model capability and address different challenges associated with it.

17:19

Nikita: As we’re talking about the LLM life cycle, I see that fine tuning is not only about making an LLM task specific. So, what are some other reasons you would fine tune an LLM model?
Himanshu: The first one is task-specific adaptation. Pre-trained language models are trained on extensive and diverse data sets and have good general language understanding. They excel in language generation and comprehension tasks, though the broad understanding of language may not lead to optimal performance in specific task. 

These models are not task specific. So the solution is fine tuning. The fine tuning process customizes the pre-trained models for a specific task by further training on task-specific data to adapt the model's knowledge. 

The second reason is domain-specific vocabulary. Pre-trained models might lack knowledge of specific words and phrases essential for certain tasks in fields, such as legal, medical, finance, and technical domains. This can limit their performance when applied to domain-specific data. 

Fine tuning enables the model to adapt and learn domain-specific words and phrases. These words could be, again, from different domains. 

18:35

Himanshu: The third reason to fine tune is efficiency and resource utilization. So fine tuning is computationally efficient compared to training from scratch. 

Fine tuning reuses the knowledge from pre-trained models, saving time and resources. Fine tuning requires fewer iterations to achieve task-specific competence. Shorter training cycles expedite the model development process. It conserves computational resources, such as GPU memory and processing power. 

Fine tuning is efficient in quicker model deployment. It has faster time to production for real world applications. Fine tuning is, again, a scalable enabling adaptation to various tasks with the same base model, which further reduce resource demands, and it leads to cost saving for research and development. 
The fourth reason to fine tune is of ethical concerns. Pre-trained models learns from diverse data. And those potentially inherit different biases. Fine tune might not completely eliminate biases. But careful curation of task specific data ensures avoiding biased or harmful vocabulary. The responsible uses of domain-specific terms promotes ethical AI applications. 

19:53

Lois: Thank you so much, Himanshu, for spending time with us. We had such a great time learning from you. If you want to learn more about the topics discussed today, head over to mylearn.oracle.com and get started on our free AI Foundations course.

Nikita: Yeah, we even have a detailed walkthrough of the architecture of transformers that you might want to check out. Join us next week for a discussion on the OCI AI Portfolio. Until then, this is Nikita Abraham…

Lois: And Lois Houston signing off!

20:24

That’s all for this episode of the Oracle University Podcast. If you enjoyed listening, please click Subscribe to get all the latest episodes. We’d also love it if you would take a moment to rate and review us on your podcast app. See you again on the next episode of the Oracle University Podcast.

Did you know that the concept of deep learning goes way back to the 1950s? However, it is only in recent years that this technology has created a tremendous amount of buzz (and for good reason!). A subset of machine learning, deep learning is inspired by the structure of the human brain, making it fascinating to learn about.

In this episode, Lois Houston and Nikita Abraham interview Senior Principal OCI Instructor Hemant Gahankari about deep learning concepts, including how Convolution Neural Networks work, and help you get your deep learning basics right.

Oracle MyLearn: https://mylearn.oracle.com/

Oracle University Learning Community: https://education.oracle.com/ou-community

LinkedIn: https://www.linkedin.com/showcase/oracle-university/

X (formerly Twitter): https://twitter.com/Oracle_Edu

Special thanks to Arijit Ghosh, David Wright, Himanshu Raj, and the OU Studio Team for helping us create this episode.

--------------------------------------------------------

Episode Transcript:

00:00

Welcome to the Oracle University Podcast, the first stop on your cloud journey. During this series of informative podcasts, we’ll bring you foundational training on the most popular Oracle technologies. Let’s get started!

00:26

Lois: Hello and welcome to the Oracle University Podcast. I’m Lois Houston, Director of Innovation Programs with Oracle University, and with me is Nikita Abraham, Principal Technical Editor.

Nikita: Hi everyone! Last week, we covered the new MySQL HeatWave Implementation Associate certification. So do go check out that episode if it interests you.

Lois: That was a really interesting discussion for sure. Today, we’re going to focus on the basics of deep learning with our Senior Principal OCI Instructor, Hemant Gahankari.

00:58

Nikita: Hi Hemant! Thanks for being with us today. So, to get started, what is deep learning?

Hemant: Deep learning is a subset of machine learning that focuses on training Artificial Neural Networks to solve a task at hand. Say, for example, image classification. A very important quality of the ANN is that it can process raw data like pixels of an image and extract patterns from it. These patterns are treated as features to predict the outcomes. 

Let us say we have a set of handwritten images of digits 0 to 9. As we know, everyone writes the digits in a slightly different way. So how do we train a machine to identify a handwritten digit? For this, we use ANN. 

ANN accepts image pixels as inputs, extracts patterns like edges and curves and so on, and correlates these patterns to predict an outcome. That is what digit does the image has in this case. 

02:04

Lois: Ok, so what you’re saying is given a bunch of pixels, ANN is able to process pixel data, learn an internal representation of the data, and predict outcomes. That’s so cool! So, why do we need deep learning?

Hemant: We need to specify features while we train machine learning algorithm. With deep learning, features are automatically extracted from the data. Internal representation of features and their combinations is built to predict outcomes by deep learning algorithms. This may not be feasible manually. 
Deep learning algorithms can make use of parallel computations. For this, usually data is split into small batches and process parallelly. So these algorithms can process large amount of data in a short time to learn the features and their combinations. This leads to scalability and performance. In short, deep learning complements machine learning algorithms for complex data for which features cannot be described easily. 

03:13

Nikita: What can you tell us about the origins of deep learning?

Hemant: Some of the deep learning concepts like artificial neuron, perceptron, and multilayer perceptron existed as early as 1950s. One of the most important concept of using backpropagation for training ANN came in 1980s. 

In 1990s, convolutional neural network were also introduced for image analysis task. Starting 2000, GPUs were introduced. And 2010 onwards, GPUs became cheaper and widely available. This fueled the widespread adoption of deep learning uses like computer vision, natural language processing, speech recognition, text translation, and so on. 

In 2012, major networks like AlexNet and Deep-Q Network were built. 2016 onward, generative use cases of the deep learning also started to come up. Today, we have widely adopted deep learning for a variety of use cases, including large language models and many other types of generative models. 

04:29

Lois: Hemant, what are various applications of deep learning algorithms? 

Hemant: Deep learning algorithms are targeted at a variety of data and applications. For data, we have images, videos, text, and audio. For images, applications can be image classification, object detection, and so on. For textual data, applications are to translate the text or detect a sentiment of a text. For audio, the applications can be music generation, speech to text, and so on. 

05:08

Lois: It's important that we select the right deep learning algorithm based on the data and application, right? So how do we do that? 

Hemant: For image task like image classification, object detection, image segmentation, or facial recognition, CNN is a suitable architecture. For text, we have a choice of the latest transformers or LSTM or even RNN. For generative tasks like text summarization, question answering, transformers is a good choice. For generating images, text to image generation, transformers, GANs, or diffusion models are available choice.

05:51

Nikita: Let’s dive a little deeper into Artificial Neural Networks. Can you tell us more about them, Hemant?
Hemant: Artificial Neural Networks are inspired by the human brain. They are made up of interconnected nodes called as neurons. 

Nikita: And how are inputs processed by a neuron? 

Hemant: In ANN, we assign weights to the connection between neurons. Weighted inputs are added up. And if the sum crosses a specified threshold, the neuron is fired. And the outputs of a layer of neuron become an input to another layer. 

06:27

Lois: Hemant, tell us about the building blocks of ANN so we understand this better.

Hemant: So first, building block is layers. We have input layer, output layer, and multiple hidden layers. The input layer and output layer are mandatory. And the hidden layers are optional. The second unit is neurons. Neurons are computational units, which accept an input and produce an output. 

Weights determine the strength of connection between neurons. So the connection could be between input and a neuron, or it could be between a neuron and another neuron. Activation functions work on the weighted sum of inputs to a neuron and produce an output. Additional input to the neuron that allows a certain degree of flexibility is called as a bias. 

07:27

Nikita: I think we’ve got the components of ANN straight but maybe you should give us an example. You mentioned this example earlier…of needing to train ANN to recognize handwritten digits from images. How would we go about that?
Hemant: For that, we have to collect a large number of digit images, and we need to train ANN using these images. 
So, in this case, the images consist of 28 by 28 pixels which act as input layer. For the output, we have neurons-- 10 neurons which represent digits 0 to 9. And we have multiple hidden layers. So, in this case, we have two hidden layers which are consisting of 16 neurons each. 

The hidden layers are responsible for capturing the internal representation of the raw image data. And the output layer is responsible for producing the desired outcomes. So, in this case, the desired outcome is the prediction of whether the digit is 0 or 1 or up to digit 9. 

So how do we train this particular ANN? So the first thing we use the backpropagation algorithm. During training, we show an image to the ANN. Let us say it is an image of digit 2. So we expect output neuron for digit 2 to fire. But in real, let us say output neuron of a digit 6 fired. 

09:12

Lois: So, then, what do we do? 

Hemant: We know that there is an error. So to correct an error, we adjust the weights of the connection between neurons based on a calculation, which we call as backpropagation algorithm. By showing thousands of images and adjusting the weights iteratively, ANN is able to predict correct outcome for most of the input images. This process of adjusting weights through backpropagation is called as model training. 

09:48

Do you have an idea for a new course or learning opportunity? We’d love to hear it! Visit the Oracle University Learning Community and share your thoughts with us on the Idea Incubator. Your suggestion could find a place in future development projects! Visit mylearn.oracle.com to get started. 

10:09

Nikita: Welcome back! Let’s move on to CNN. Hemant, what is a Convolutional Neural Network? 

Hemant: CNN is a type of deep learning model specifically designed for processing and analyzing grid-like data, such as images and videos. In the ANN, the input image is converted to a single dimensional array and given as an input to the network.  
But that does not work well with the image data because image data is inherently two dimensional. CNN works better with two dimensional data. The role of the CNN is to reduce the image into a form, which is easier to process and without losing features, which are critical for getting a good prediction. 

10:53

Lois: A CNN has different layers, right? Could you tell us a bit about them? 

Hemant: The first one is input layer. Input layer is followed by feature extraction layers, which is a combination and repetition of multiple feature extraction layers, including convolutional layer with ReLu activation and a pooling layer. 

And this is followed by a classification layer. These are the fully connected output layers, where the classification occurs as output classes. The feature extraction layers play a vital role in image classification.  

11:33

Nikita: Can you explain these layers with an example?

Hemant: Let us say we have a robot to inspect a house and tell us what type of a house it is. It uses many tools for this purpose. The first tool is a blueprint detector. It scans different parts of the house, like walls, floors, or windows, and looks for specific patterns or features. 

The second tool is a pattern highlighter. This tool marks areas detected by the blueprint detector. The next tool is a summarizer. It tries to capture the most significant features of every room. The next tool is house expert, which looks at all the highlighted patterns and features, and tries to understand the house. 

The next tool is a guess maker. It assigns probabilities to the different possible house types. And finally, the quality checker randomly checks different parts of the analysis to make sure that the robot doesn't rely too much on any single piece of information. 

12:40

Nikita: Ok, so how are you mapping these to the feature extraction layers? 

Hemant: Similar to blueprint detector, we have a convolutional layer. This layer applies convolutional operations to the input image using small filters known as kernels. 

Each filter slides across the input image to detect specific features, such as edges, corners, or textures. Similar to pattern highlighter, we have a activation function. The activation function allows the network to learn more complex and non-linear relationships in the data. Pooling layer is similar to room summarizer. 

Pooling helps reduce the spatial dimensions of the feature maps generated by the convolutional layers. Similar to house expert, we have a fully connected layer, which is responsible for making final predictions or classifications based on the learned features. Softmax layer converts the output of the last fully connected layers into probability scores. 

The class with the highest probability is the predicted class. This is similar to the guess maker. And finally, we have the dropout layer. This layer is a regularization technique used to prevent overfitting in the network. This has the same role as that of a quality checker. 

14:05

Lois: Do CNNs have any limitations that we need to be aware of?

Hemant: Training CNNs on large data sets can be computationally expensive and time consuming. CNNs are susceptible to overfitting, especially when the training data is limited or imbalanced. CNNs are considered black box models making it difficult to interpret. 

And CNNs can be sensitive to small changes in the input leading to unstable predictions. 

14:33

Nikita: And what are the top applications of CNN?
Hemant: One of the most widely used applications of CNNs is image classification. For example, classifying whether an image contains a specific object, say cat or a dog. 

CNNs are used for object detection tasks. The goal here is to draw bounding boxes around objects in an image. CNNs can perform pixel level segmentation, where each pixel in the image is labeled to represent different objects or regions. CNNs are employed for face recognition tasks as well, identifying and verifying individuals based on facial features. 

CNNs are widely used in medical image analysis, helping with tasks like tumor detection, diagnosis, and classification of various medical conditions. CNNs play an important role in the development of self-driving cars, helping them to recognize and understand the road traffic signs, pedestrians, and other vehicles. And CNNs are applied in analyzing satellite images and remote sensing data for tasks, such as land cover classification and environmental monitoring. 

15:50

Nikita: Hemant, let’s talk about sequence models. What are they and what are they used for?

Hemant: Sequence models are used to solve problems, where the input data is in the form of sequences. The sequences are ordered lists of data points or events. 

The goal in sequence models is to find patterns and dependencies within the data and make predictions, classifications, or even generate new sequences. 

16:17

Lois: Can you give us some examples of sequence models? 

Hemant: Some common examples of the sequence models are in natural language processing, deep learning models are used for tasks, such as machine translation, sentiment analysis, or text generation. In speech recognition, deep learning models are used to convert a recorded audio into text. 

In deep learning models, can generate new music or create original compositions. Even sequences of hand gestures are interpreted by deep learning models for applications like sign language recognition. In fields like finance or weather prediction, time series data is used to predict future values. 

17:03

Nikita: Which deep learning models can be used to work with sequence data? 

Hemant: Recurrent Neural Networks, abbreviated as RNNs, are a class of neural network architectures specifically designed to handle sequential data. Unlike traditional feedforward neural network, RNNs have a feedback loop that allows information to persist across different timesteps. 

The key features of RNN is their ability to maintain an internal state often referred to as a hidden state or memory, which is updated as the network processes each element in the input sequence. The hidden state is then used as input to the network for the next time step, allowing the model to capture dependencies and patterns in the data that are spread across time. 

17:58

Nikita: Are there various types of RNNs?

Hemant: There are different types of RNN architecture based on application. 

One of them is one to one. This is like feed forward neural network and is not suited for sequential data. A one to many model produces multiple output values for one input value. Music generation or sequence generation are some applications using this architecture. 

A many to one model produces one output value after receiving multiple input values. Example is sentiment analysis based on the review. Many to many model produces multiple output values for multiple input values. Examples are machine translation and named entity recognition. 

RNN does not perform that well when it comes to capturing long term dependencies. This is due to the vanishing gradients problem, which is overcome by using LSTM model. 

19:07

Lois: Another acronym. What is LSTM, Hemant?

Hemant: Long Short-Term memory, abbreviated as LSTM, works by using a specialized memory cell and a gating mechanisms to capture long term dependencies in the sequential data. 
The key idea behind LSTM is to selectively remember or forget information over time, enabling the model to maintain relevant information over long sequences, which helps overcome the vanishing gradients problem. 

19:40

Nikita: Can you take us, step-by-step, through the working of LSTM? 

Hemant: At each timestep, the LSTM takes an input vector representing the current data point in the sequence. The LSTM also receives the previous hidden state and cell state. These represent what the LSTM has remembered and forgotten up to the current point in the sequence. 

The core of the LSTM lies in its gating mechanisms, which include three gates: the input gate, the forget gate, and the output gate. These gates are like the filters that control the flow of information within the LSTM cell. The input gate decides what new information from the current input should be added to the memory cell. 

The forget gate determines what information in the current memory cell should be discarded or forgotten. The output gate regulates how much of the current memory cell should be exposed as the output of the current time step. Using the information from the input gate and forget gate, the LSTM updates its cell state. The LSTM then uses the output gate to produce the current hidden state, which becomes the output of the LSTM for the next time step. 

21:12

Lois: Thank you, Hemant, for joining us in this episode of the Oracle University Podcast. I learned so much today. If you want to learn more about deep learning, visit mylearn.oracle.com and search for the Oracle Cloud Infrastructure AI Foundations course. And remember, the AI Foundations course and certification are free. So why not get started now?

Nikita: Right, Lois. In our next episode, we will discuss generative AI and language learning models. Until then, this is Nikita Abraham…

Lois: And Lois Houston signing off!

21:45

That’s all for this episode of the Oracle University Podcast. If you enjoyed listening, please click Subscribe to get all the latest episodes. We’d also love it if you would take a moment to rate
and review us on your podcast app. See you again on the next episode of the Oracle University Podcast.

What is MySQL HeatWave? How do I get certified in it? Where do I start?

Listen to Lois Houston and Nikita Abraham, along with MySQL Developer Scott Stroz, answer all these questions and more on this week's episode of the Oracle University Podcast.

MySQL Document Store: https://oracleuniversitypodcast.libsyn.com/mysql-document-store

Oracle MyLearn: https://mylearn.oracle.com/

Oracle University Learning Community: https://education.oracle.com/ou-community

LinkedIn: https://www.linkedin.com/showcase/oracle-university/

X (formerly Twitter): https://twitter.com/Oracle_Edu

Special thanks to Arijit Ghosh, David Wright, and the OU Studio Team for helping us create this episode.

--------------------------------------------------------

Episode Transcript:

00:00

Welcome to the Oracle University Podcast, the first stop on your cloud journey. During this 
series of informative podcasts, we’ll bring you foundational training on the most popular 
Oracle technologies. Let’s get started!

00:26

Nikita: Welcome to the Oracle University Podcast! I’m Nikita Abraham, Principal Technical Editor with Oracle University, and with me is Lois Houston, Director of Innovation Programs.

Lois: Hi there! For the last two weeks, we’ve been having really exciting discussions on everything AI. We covered the basics of artificial intelligence and machine learning, and we’re taking a short break from that today to talk about the new MySQL HeatWave Implementation Associate Certification with MySQL Developer Advocate Scott Stroz.
00:59
Nikita: You may remember Scott from an episode last year where he came on to discuss MySQL Document Store. We’ll post the link to that episode in the show notes so you can listen to it if you haven’t already.

Lois: Hi Scott! Thanks for joining us again. Before diving into the certification, tell us, what is MySQL HeatWave? 

01:19

Scott: Hi Lois, Hi Niki. I’m so glad to be back. So, MySQL HeatWave Database Service is a fully managed database that is capable of running transactional and analytic queries in a single database instance. This can be done across data warehouses and data lakes. We get all the benefits of analytic queries without the latency and potential security issues of performing standard extract, transform, and load, or ETL, operations. Some other MySQL HeatWave database service features are automated system updates and database backups, high availability, in-database machine learning with AutoML, MySQL Autopilot for managing instance provisioning, and enhanced data security. 

HeatWave is the only cloud database service running MySQL that is built, managed, and supported by the MySQL Engineering team.

02:14

Lois: And where can I find MySQL HeatWave?

Scott: MySQL HeatWave is only available in the cloud. MySQL HeatWave instances can be provisioned in Oracle Cloud Infrastructure or OCI, Amazon Web Services (AWS), and Microsoft Azure. Now, some features though are only available in Oracle Cloud, such as access to MySQL Document Store.

02:36

Nikita: Scott, you said MySQL HeatWave runs transactional and analytic queries in a single instance. Can you elaborate on that?

Scott: Sure, Niki. So, MySQL HeatWave allows developers, database administrators, and data analysts to run transactional queries (OLTP) and analytic queries (OLAP). 

OLTP, or online transaction processing, allows for real-time execution of database transactions. A transaction is any kind of insertion, deletion, update, or query of data. Most DBAs and developers work with this kind of processing in their day-to-day activities.
 
OLAP, or online analytical processing, is one way to handle multi-dimensional analytical queries typically used for reporting or data analytics. OLTP system data must typically be exported, aggregated, and imported into an OLAP system. This procedure is called ETL as I mentioned – extract, transform, and load. With large datasets, ETL processes can take a long time to complete, so analytic data could be “old” by the time it is available in an OLAP system. There is also an increased security risk in moving the data to an external source.

03:56

Scott: MySQL HeatWave eliminates the need for time-consuming ETL processes. We can actually get real-time analytics from our data since HeatWave allows for OLTP and OLAP in a single instance. I should note, this also includes analytic from JSON data that may be stored in the database.

Another advantage is that applications can use MySQL HeatWave without changing any of the application code. Developers only need to point their applications at the MySQL HeatWave databases. MySQL HeatWave is fully compatible with on-premise MySQL instances, which can allow for a seamless transition to the cloud.

And one other thing. When MySQL HeatWave has OLAP features enabled, MySQL can determine what type of query is being executed and route it to either the normal database system or the in-memory database.

04:52

Lois: That’s so cool! And what about the other features you mentioned, Scott? Automated updates and backups, high availability…

Scott: Right, Lois. But before that, I want to tell you about the in-memory query accelerator. MySQL HeatWave offers a massively parallel, in-memory hybrid columnar query processing engine. It provides high performance by utilizing algorithms for distributed query processing. And this query processing in MySQL HeatWave is optimized for cloud environments. 

MySQL HeatWave can be configured to automatically apply system updates, so you will always have the latest and greatest version of MySQL.

Then, we have automated backups. By this, I mean MySQL HeatWave can be configured to provide automated backups with point-in-time recovery to ensure data can be restored to a particular date and time. MySQL HeatWave also allows us to define a retention plan for our database backups, that means how long we keep the backups before they are deleted.

High availability with MySQL HeatWave allows for more consistent uptime. When using high availability, MySQL HeatWave instances can be provisioned across multiple availability domains, providing automatic failover for when the primary node becomes unavailable. All availability domains within a region are physically separated from each other to mitigate the possibility of a single point of failure.

06:14

Scott: We also have MySQL Lakehouse. Lakehouse allows for the querying of data stored in object storage in various formats. This can be CSV, Parquet, Avro, or an export format from other database systems. And basically, we point Lakehouse at data stored in Oracle Cloud, and once it’s ingested, the data can be queried just like any other data in a database. Lakehouse supports querying data up to half a petabyte in size using the HeatWave engine. And this allows users to take advantage of HeatWave for non-MySQL workloads.

MySQL AutoPilot is a part of MySQL HeatWave and can be used to predict the number of HeatWave nodes a system will need and automatically provision them as part of a cluster. AutoPilot has features that can handle automatic thread pooling and database shape predicting. A “shape” is one of the many different CPU, memory, and ethernet traffic configurations available for MySQL HeatWave.

MySQL HeatWave includes some advanced security features such as asymmetric encryption and automated data masking at query execution.

As you can see, there are a lot of features covered under the HeatWave umbrella!
07:31

Did you know that Oracle University offers free courses on Oracle Cloud Infrastructure? You’ll find training on everything from cloud computing, database, and security to artificial intelligence and machine learning, all free to subscribers. So, what are you waiting for? Pick a topic, leverage the Oracle University Learning Community to ask questions, and then sit for your certification. Visit mylearn.oracle.com to get started. 
08:02

Nikita: Welcome back! Now coming to the certification, who can actually take this exam, Scott?

Scott: The MySQL HeatWave Implementation Associate Certification Exam is designed specifically for administrators and data scientists who want to provision, configure, and manage MySQL HeatWave for transactions, analytics, machine learning, and Lakehouse.

08:22

Nikita: Can someone who’s just graduated, say an engineering graduate interested in data analytics, take this certification? Are there any prerequisites? What are the career prospects for them?

Scott: There are no mandatory prerequisites, but anyone who wants to take the exam should have experience with MySQL HeatWave and other aspects of OCI, such as virtual cloud networks and identity and security processes. Also, the learning path on MyLearn will be extremely helpful when preparing for the exam, but you are not required to complete the learning path before registering for the exam.

The exam focuses more on getting MySQL HeatWave running (and keeping it running) than accessing the data. That doesn’t mean it is not helpful for someone interested in data analytics. I think it can be helpful for data analysts to understand how the system providing the data functions, even if it is at just a high level. It is also possible that data analysts might be responsible for setting up their own systems and importing and managing their own data.

09:23

Lois: And how do I get started if I want to get certified on MySQL HeatWave?

Scott: So, you’ll first need to go to mylearn.oracle.com and look for the “Become a MySQL HeatWave Implementation Associate” learning path. The learning path consists of over 10 hours of training across 8 different courses. 

These courses include “Getting Started with MySQL HeatWave Database Service,” which offers an introduction to some Oracle Cloud functionality such as security and networking, as well as showing one way to connect to a MySQL HeatWave instance. Another course demonstrates how to configure MySQL instances and copy that configuration to other instances. Other courses cover how to migrate data into MySQL HeatWave, set up and manage high availability, and configure HeatWave for OLAP.

You’ll find labs where you can perform hands-on activities, student and activity guides, and skill checks to test yourself along the way. And there’s also the option to Ask the Instructor if you have any questions you need answers to. You can also access the Oracle University Learning Community and discuss topics with others on the same journey. The learning path includes a practice exam to check your readiness to pass the certification exam.

10:33

Lois: Yeah, and remember, access to the entire learning path is free so there’s nothing stopping you from getting started right away. Now Scott, what does the certification test you on?

Scott: The MySQL HeatWave Implementation exam, which is an associate-level exam, covers various topics. It will validate your ability to identify key features and benefits of MySQL HeatWave and describe the MySQL HeatWave architecture; identify Virtual Cloud Network (VCN) requirements and the different methods of connecting to a MySQL HeatWave instance; manage the automatic backup process and restore database systems from these backups; configure and manage read replicas and inbound replication channels; import data into MySQL HeatWave; configure and manage high availability and clustering of MySQL HeatWave instances.

I know this seems like a lot of different topics. That is why we recommend anyone interested in the exam follow the learning path. It will help make sure you have the exposure to all the topics that are covered by the exam.

11:35

Lois: Tell us more about the certification process itself.

Scott: While the courses we already talked about are valuable when preparing for the exam, nothing is better than hands-on experience. We recommend that candidates have hands-on experience with MySQL HeatWave with real-world implementations. The format of the exam is Multiple Choice. It is 90 minutes long and consists of 65 questions. When you’ve taken the recommended training and feel ready to take the certification exam, you need to purchase the exam and register for it. You go through the section on things to do before the exam and the exam policies, and then all that’s left to do is schedule the date and time of the exam according to when is convenient for you.

12:16

Nikita: And once you’ve finished the exam?

Scott: When you’re done your score will be displayed on the screen when you finish the exam. You will also receive an email indicating whether you passed or failed. You can view your exam results and full score report in Oracle CertView, Oracle’s certification portal. From CertView, you can download and print your eCertificate and even share your newly earned badge on places like Facebook, Twitter, and LinkedIn.

12:38

Lois: And for how long does the certification remain valid, Scott?

Scott: There is no expiration date for the exam, so the certification will remain valid for as long as the material that is covered remains relevant. 

12:49

Nikita: What’s the next step for me after I get this certification? What other training can I take?

Scott: So, because this exam is an associate level exam, it is kind of a stepping stone along a person’s MySQL training. I do not know if there are plans for a professional level exam for HeatWave, but Oracle University has several other training programs that are MySQL-specific. There are learning paths to help prepare for the MySQL Database Administrator and MySQL Database Developer exams. As with the HeatWave learning paths, the learning paths for these exams include video tutorials, hands-on activities, skill checks, and practice exams.

13:27

Lois: I think you’ve told us everything we need to know about this certification, Scott. Are there any parting words you might have?

Scott: We know that the whole process of training and getting certified may seem daunting, but we’ve really tried to simplify things for you with the “Become a MySQL HeatWave Implementation Associate” learning path. It not only prepares you for the exam but also gives you experience with features of MySQL HeatWave that will surely be valuable in your career.

13:51

Lois: Thanks so much, Scott, for joining us today.

Nikita: Yeah, we’ve had a great time with you.

Scott: Thanks for having me.

Lois: Next week, we’ll get back to our focus on AI with a discussion on deep learning. Until then, this is Lois Houston…

Nikita: And Nikita Abraham, signing off.

14:07

That’s all for this episode of the Oracle University Podcast. If you enjoyed listening, please click
Subscribe to get all the latest episodes. We’d also love it if you would take a moment to rate
and review us on your podcast app. See you again on the next episode of the Oracle University 
Podcast.

Does machine learning feel like too convoluted a topic? Not anymore!

Listen to hosts Lois Houston and Nikita Abraham, along with Senior Principal OCI Instructor Hemant Gahankari, talk about foundational machine learning concepts and dive into how supervised learning, unsupervised learning, and reinforcement learning work.

Oracle MyLearn: https://mylearn.oracle.com/

Oracle University Learning Community: https://education.oracle.com/ou-community

LinkedIn: https://www.linkedin.com/showcase/oracle-university/

X (formerly Twitter): https://twitter.com/Oracle_Edu

Special thanks to Arijit Ghosh, David Wright, Himanshu Raj, and the OU Studio Team for helping us create this episode.

---------------------------------------------------------

Episode Transcript:

00:00

Welcome to the Oracle University Podcast, the first stop on your cloud journey. During this 

series of informative podcasts, we’ll bring you foundational training on the most popular 

Oracle technologies. Let’s get started! 

00:26

Lois: Hello and welcome to the Oracle University Podcast. I’m Lois Houston, Director of Innovation Programs with Oracle University, and with me is Nikita Abraham, Principal 

Technical Editor.

Nikita: Hi everyone! Last week, we went through the basics of artificial intelligence and we’re going to take it a step further today by talking about some foundational machine learning concepts. After that, we’ll discuss the three main types of machine learning models: supervised learning, unsupervised learning, and reinforcement learning.

00:57

Lois: Hemant Gahankari, a Senior Principal OCI Instructor, joins us for this episode. Hi Hemant! Let’s dive right in. What is machine learning? How does it work?

Hemant: Machine learning is a subset of artificial intelligence that focuses on creating computer systems that can learn and predict outcomes from given examples without being explicitly programmed. It is powered by algorithms that incorporate intelligence into machines by automatically learning from a set of examples usually provided as data.

01:34

Nikita: Give us a few examples of machine learning… so we can see what it can do for us.

Hemant: Machine learning is used by all of us in our day-to-day life.

When we shop online, we get product recommendations based on our preferences and our shopping history. This is powered by machine learning.

We are notified about movies recommendations based on our viewing history and choices of other similar viewers. This too is driven by machine learning.

While browsing emails, we are warned of a spam mail because machine learning classifies whether the mail is spam or not based on its content. In the increasingly popular self-driving cars, machine learning is responsible for taking the car to its destination.

02:24

Lois: So, how does machine learning actually work?

Hemant: Let us say we have a computer and we need to teach the computer to differentiate between a cat and a dog. We do this by describing features of a cat or a dog.

Dogs and cats have distinguishing features. For example, the body color, texture, eye color are some of the defining features which can be used to differentiate a cat from a dog. These are collectively called as input data.

We also provide a corresponding output, which is called as a label, which can be a dog or a cat in this case. By describing a specific set of features, we can say that it is a cat or a dog.

Machine learning model is first trained with the data set. Training data set consists of a set of features and output labels, and is given as an input to the machine learning model.

During the process of training, machine learning model learns the relation between input features and corresponding output labels from the provided data. Once the model learns from the data, we have a trained model.

Once the model is trained, it can be used for inference. Inference is a process of getting a prediction by giving a data point. In this example, we input features of a cat or a dog, and the trained model predicts the output that is a cat or a dog label.

The types of machine learning models depend on whether we have a labeled output or not.

04:08

Nikita: Oh, there are different types of machine learning models?

Hemant: In general, there are three types of machine learning approaches.

In supervised machine learning, labeled data is used to train the model. Model learns the relation between features and labels.

Unsupervised learning is generally used to understand relationships within a data set. Labels are not used or are not available.

Reinforcement learning uses algorithms that learn from outcomes to make decisions or choices.

04:45

Lois: Ok…supervised learning, unsupervised learning, and reinforcement learning. Where do we use each of these machine learning models?

Hemant: Some of the popular applications of supervised machine learning are disease detection, weather forecasting, stock price prediction, spam detection, and credit scoring. For example, in disease detection, the patient data is input to a machine learning model, and machine learning model predicts if a patient is suffering from a disease or not.

For unsupervised machine learning, some of the most common real-time applications are to detect fraudulent transactions, customer segmentation, outlier detection, and targeted marketing campaigns. So for example, given the transaction data, we can look for patterns that lead to fraudulent transactions.

Most popular among reinforcement learning applications are automated robots, autonomous driving cars, and playing games.

05:51

Nikita: I want to get into how each type of machine learning works. Can we start with supervised learning?

Hemant: Supervised learning is a machine learning model that learns from labeled data. The model learns the mapping between the input and the output.

As a house price predictor model, we input house size in square feet and model predicts the price of a house. Suppose we need to develop a machine learning model for detecting cancer, the input to the model would be the person's medical details, the output would be whether the tumor is malignant or not.

06:29

Lois: So, that mapping between the input and output is fundamental in supervised learning.

Hemant: Supervised learning is similar to a teacher teaching student. The model is trained with the past outcomes and it learns the relationship or mapping between the input and output.

In supervised machine learning model, the outputs can be either categorical or continuous. When the output is continuous, we use regression. And when the output is categorical, we use classification.

07:05

Lois: We want to keep this discussion at a high level, so we’re not going to get into regression and classification. But if you want to learn more about these concepts and look at some demonstrations, visit mylearn.oracle.com.

Nikita: Yeah, look for the Oracle Cloud Infrastructure AI Foundations course and you’ll find a lot of resources that you can make use of.

07:30

The Oracle University Learning Community is an excellent place to collaborate and learn with Oracle experts and fellow learners. Grow your skills, inspire innovation, and celebrate your successes. All your activities, from liking a post to answering questions and sharing with others, will help you earn a valuable reputation, badges, and ranks to be recognized in the community.

Visit mylearn.oracle.com to get started. 

07:58

Nikita: Welcome back! So that was supervised machine learning. What about unsupervised machine learning, Hemant?

Hemant: Unsupervised machine learning is a type of machine learning where there are no labeled outputs. The algorithm learns the patterns and relationships in the data and groups similar data items. In unsupervised machine learning, the patterns in the data are explored explicitly without being told what to look for.

For example, if you give a set of different-colored LEGO pieces to a child and ask to sort it, it may the LEGO pieces based on any patterns they observe. It could be based on same color or same size or same type. Similarly, in unsupervised learning, we group unlabeled data sets.

One more example could be-- say, imagine you have a basket of various fruits-- say, apples, bananas, and oranges-- and your task is to group these fruits based on their similarities. You observe that some fruits are round and red, while others are elongated and yellow. Without being told explicitly, you decide to group the round and red fruits together as one cluster and the elongated and yellow fruits as another cluster. There you go. You have just performed an unsupervised learning task.

09:21

Lois: Where is unsupervised machine learning used? Can you take us through some use cases?

Hemant: The first use case of unsupervised machine learning is market segmentation. In market segmentation, one example is providing the purchasing details of an online shop to a clustering algorithm. Based on the items purchased and purchasing behavior, the clustering algorithm can identify customers based on the similarity between the products purchased. For example, customers with a particular age group who buy protein diet products can be shown an advertisement of sports-related products.

The second use case is on outlier analysis. One typical example for outlier analysis is to provide credit card purchase data for clustering. Fraudulent transactions can be detected by a bank by using outliers. In some transaction, amounts are too high or recurring. It signifies an outlier.

The third use case is recommendation systems. An example for recommendation systems is to provide users' movie viewing history as input to a clustering algorithm. It clusters users based on the type or rating of movies they have watched. The output helps to provide personalized movie recommendations to users. The same applies for music recommendations also.

10:53

Lois: And finally, Hemant, let’s talk about reinforcement learning.

Hemant: Reinforcement learning is like teaching a dog new tricks. You reward it when it does something right, and over time, it learns to perform these actions to get more rewards. Reinforcement learning is a type of Machine Learning that enables an agent to learn from its interaction with the environment, while receiving feedback in the form of rewards or penalties without any labeled data.

Reinforcement learning is more prevalent in our daily lives than we might realize. The development of self-driving cars and autonomous drones rely heavily on reinforcement learning to make real time decisions based on sensor data, traffic conditions, and safety considerations.

Many video games, virtual reality experiences, and interactive entertainment use reinforcement learning to create intelligent and challenging computer-controlled opponents. The AI characters in games learn from player interactions and become more difficult to beat as the game progresses.

12:05

Nikita: Hemant, take us through some of the terminology that’s used with reinforcement learning.

Hemant: Let us say we want to train a self-driving car to drive on a road and reach its destination. For this, it would need to learn how to steer the car based on what it sees in front through a camera. Car and its intelligence to steer on the road is called as an agent.

More formally, agent is a learner or decision maker that interacts with the environment, takes actions, and learns from the feedback received. Environment, in this case, is the road and its surroundings with which the car interacts. More formally, environment is the external system with which the agent interacts. It is the world or context in which the agent operates and receives feedback for its actions.

What we see through a camera in front of a car at a moment is a state. State is a representation of the current situation or configuration of the environment at a particular time. It contains the necessary information for the agent to make decisions. The actions in this example are to drive left, or right, or keep straight. Actions are a set of possible moves or decisions that the agent can take in a given state.

Actions have an impact on the environment and influence future states. After driving through the road many times, the car learns what action to take when it views a road through the camera. This learning is a policy. Formally, policy is a strategy or mapping that the agent uses to decide which action to take in a given state. It defines the agent's behavior and determines how it selects actions.

13:52

Lois: Ok. Say we’re talking about the training loop of reinforcement learning in the context of training a dog to learn tricks. We want it to pick up a ball, roll, sit…

Hemant: Here the dog is an agent, and the place it receives training is the environment. While training the dog, you provide a positive reward signal if the dog picks it right and a warning or punishment if the dog does not pick up a trick. In due course, the dog gets trained by the positive rewards or negative punishments.

The same tactics are applied to train a machine in the reinforcement learning. For machines, the policy is the brain of our agent. It is a function that tells what actions to take when in a given state. The goal of reinforcement learning algorithm is to find a policy that will yield a lot of rewards for the agent if the agent follows that policy referred to as the optimal policy.

Through a process of learning from experiences and feedback, the agent becomes more proficient at making good decisions and accomplishing tasks. This process continues until eventually we end up with the optimal policy. The optimal policy is learned through training by using algorithms like Deep Q Learning or Q Learning.

15:19

Nikita: So through multiple training iterations, it gets better. That’s fantastic. Thanks, Hemant, for joining us today. We’ve learned so much from you.

Lois: Remember, the course and certification are free, so if you’re interested, make sure you log in to mylearn.oracle.com and get going. Join us next week for another episode of the Oracle University Podcast. Until then, I’m Lois Houston…

Nikita: And Nikita Abraham signing off!

15:48

That’s all for this episode of the Oracle University Podcast. If you enjoyed listening, please click Subscribe to get all the latest episodes. We’d also love it if you would take a moment to rate and review us on your podcast app. See you again on the next episode of the Oracle University Podcast.