Generative AI

 ChatGPT is developed by company OpenAI. This product became the first fastest growing internet product hitting 100 Million users in just 2 months. 

It was just a research project to ensure AGI (Artificial General Intelligence) is achieved for the benefits of humanity. 

ChatGPT refers to chat enabled version of Generative Pretrained Transformer model.

The company OpenAI is working with Microsoft since 2016 using Azure Cloud to train and run AI models. 

OpenAI APIs are hosted on Azure servers but it is maintained by OpenAI only. 

Microsoft has its own version of OpenAI api called AzureOpenAI Api. 

Machine Learning

Machine learning is a subset of AI that applies mathematics to large datasets to find trends and patterns while mapping inputs to outputs. 

Such as, given these features of a home, age, location, and the number of bedrooms as input, what will the selling price be? This means computers can produce output without being explicitly programmed by a developer or software engineer. 

Model

The mappings uncovered between the inputs and outputs are stored in a mathematical model, simply called the Model. 

Implementing Machine Learning project has two options:

1. Train a model from Scratch

2. User a pretrained Model

Pretrained Model sources:

1. ModelZoo.com

2. AWS Marketplace 

3. Hugging Face

Train a custom model (Creating model from Scratch)

1. Using Python

2. Using R

3. Using Java

To work with Phython, the environment is Jupyter Notebook

Machine Learning Framework Libraries:

1. scikit-learn 

2. TensorFlow

3. MXNet 

4. PyTorch

5. Keras

Day2 - Artificial Intelligence

NumPy - A library in Python for basic operations

Pandas - Based on Numpy, it is another library

Installed Jupyter Notebook

MatplotLib: A library

Seaborn

Cufflinks: is a Python library that simplifies interactive plotting. It allows you to create interactive charts directly from Pandas DataFrames or Series.

Kaggle is a fantastic platform for accessing high-quality datasets to practice and develop your skills in machine learning, data science, and artificial intelligence. 

Kaggle is one of the largest global hosting platforms used by data scientists and machine learning enthusiasts globally. 

Founded in 2010 and acquired by Google in 2017, Kaggle always supported a diverse set of tools and resources to enable learning, collaboration, and innovation.

Kaggle is best known for its competition platform, through which it hosts users to face real-world problems. Companies, researchers, and organizations pose competition. These challenges include easy tasks in predictive modelling as well as highly complex machine learning problems, oftentimes with large sums of money on the line. Besides the competitions, there are also several thousands of publicly available datasets in Kaggle, encompassing various domains including finance, healthcare, and social sciences among others. 

The importance of these datasets to experimentation, learning, and community-driven project contributions cannot be overemphasized.

Introduction to Artificial Intelligence (AI)

Artificial intelligence (AI) refers to the computer systems capable of performing complex tasks that historically only a human could do, such as reasoning, making decisions, or solving problems. 

It involves creating machines that can think like humans and imitate their actions.

Types of AI

1. Narrow AI (Weak AI):

   • Designed to perform a single task.
   • Example: Virtual assistants like Siri or Alexa.

2. General AI (Strong AI):

   • Has human-like cognitive abilities across multiple domains (still theoretical).
   • Example: A machine capable of understanding, reasoning, and learning any task a human can do.

3. Super AI:

   • Surpasses human intelligence (currently a concept).

Key concepts:

1. Machine Learning: focuses on developing systems that learn or improve performance, based on the data. Example: Predicting weather or stock prices.

2. Deep Learning: A subset of ML using neural networks to mimic the human brain. Example: Image and speech recognition.

3. Neural Networks: Enables machines to understand and generate human language. Example: Chatbots and language translators.

4. Datasets: Datasets are large collections of information that AI systems use to learn.

5. LLM

6. NLP

7. Computer Vision: Focuses on enabling machines to interpret visual data. Example: Facial recognition systems.

Machine Learning:

1. Supervised Learning: is a learning process where algorithm learns from labeled data.

Tools for Supervised Learning:

Python Libraries:

• scikit-learn.
• TensorFlow.
• PyTorch.

Platforms:

• Azure ML, AWS SageMaker, Google AI.

2. Unsupervised Learning 

3. Reinforcement Learning 

AWS Lambda

What is Serverless : 

 Pay for what you use

Service automatically scales up or down accordingly. 

AWS Lambda allows upload the code to Amazon aws without the need of the server. 

AWS Lambda is a Function as a service (FaaS) which is a serverless architecture that developers can use to write custom backend functions and deploy the function code directly to the cloud infrastructure

It falls under the Compute services in AWS. 

Process:

1. Create a lambda function

2. Upload to AWS 

3. Test and trigger

Lambda execution models:

1. Synchronous 

2. Asynchronous

3. Pull stream based

AWS Certified Cloud Practitioner Exam CLF-C02

Cloud Concepts - 24%

Security and Compliance - 30%

Cloud Technology and Services - 34%

Billing, Pricing and Support - 12%

AWS Storage

 1. S3 (Simple Storage Service) - File storage

2. Glacier

3. CloundFront

4. EBS (Elastic Block Stoarge) - Block storage

5. Storage Gateway

6. Snow Family

7. Databases

S3 Storage Class:

It is all about Object Storage through buckets. Objects could be a file or any chunk of data. AWS stores them and makes available to at least 3 available zones. 

Supports Encryption.

Ways to get data into S3 buckets:

1. Through API

2. AWS Direct Connect

3. Storage Gateway

4. Kinesis Firehose

5. Transfer Acceleration

6. Snowball, snowball edge and snowmobile

S3 Concepts:

1. Bucket

2. Regions

3. Objects

4. Keys

5. Object URLs

6. Eventual Consistency

7. Prefixes and Delimiters

Operations on S3:

1. Creating and deleting buckets

2. Reading objects

3. Writing objects

4. Deleting objects