Summary of "AIF-C01 Module 1.1 - Introduction to AI/ML"

Summary of AIF-C01 Module 1.1 - Introduction to AI/ML

This module provides a foundational overview of Artificial Intelligence (AI) and Machine Learning (ML), explaining key concepts, terminologies, and their interrelationships, along with practical examples and data types used in AI/ML training.

Main Ideas and Concepts

1. Introduction to Artificial Intelligence (AI)

AI is a branch of computer science focused on simulating human intelligence in machines.
It enables machines to perform tasks that typically require human intelligence such as speech recognition, problem-solving, and decision-making.
Example: Voice assistants like Amazon Alexa, Apple Siri, Microsoft Cortana, Google Assistant, and Samsung Bixby are AI-enabled devices.

2. What is Machine Learning (ML)?

ML is a subset of AI focused on creating algorithms that allow computers to learn from data and improve performance without explicit programming.
Unlike traditional programming, ML systems learn patterns from data to make predictions or decisions.
Example: Netflix or Amazon Prime recommendations based on user viewing history.
ML is powerful for pattern recognition, fraud detection, and forecasting trends.

3. How AI and ML Work Together

AI models are trained using datasets (e.g., images of dogs, cats, cows).
Two main steps:
- Model Training: Feeding labeled data to the model to learn patterns (classification is one example).
- Inferencing: Using the trained model to predict or classify new, unseen data.
Example: Recognizing a dog in a new image despite similarities to a cat.

4. Neural Networks and Deep Learning

Neural Networks mimic the human brain structure with layers of interconnected nodes (neurons).
Layers include input, multiple hidden layers, and output.
Deep Learning is a subset of ML that uses neural networks to perform complex tasks like image and speech recognition.
Example: Tesla’s self-driving cars use deep learning to adapt to dynamic road conditions by processing sensor data continuously.

5. Generative AI (GenAI)

A subset of deep learning focused on creating new content (text, images, videos, source code).
Unlike traditional AI that predicts from existing data, GenAI creates unique, human-like content.
Example: ChatGPT (based on GPT-4), DALL·E for image generation.
GenAI uses Foundation Models which are large, pre-trained models on massive datasets.
Foundation models can be fine-tuned for specific organizational needs.
Large Language Models (LLMs) are a type of foundation model specialized in language tasks (e.g., GPT-4, Google Bard).

6. Prompts in Generative AI

A prompt is the input text or instruction guiding the AI to generate responses.
Prompts can be questions, statements, or commands.
Generative AI is nondeterministic: the same prompt may produce different but contextually similar outputs each time.

7. Image Generation Using Diffusion Models

Example: Stable Diffusion model.
Training involves gradually adding noise to an image until it becomes pure noise.
Generation starts with pure noise and progressively adds details to form a final image based on the prompt.

8. Types of Training Data

Key Characteristics: Diversity (to avoid bias) and Quality (garbage in, garbage out).
Labelled Data: Data tagged with correct answers (used in supervised learning).
Unlabelled Data: Data without tags, used in unsupervised learning where the model finds patterns on its own.
Structured Data: Organized in rows and columns with consistent schema (e.g., spreadsheets, SQL databases).
Unstructured Data: No predefined format, includes text, images, audio, video (used in NLP, image recognition).

9. Training Data Sets for ML Models

Data is split into three sets:
- Training Set (60-80%): Used to train the model.
- Validation Set (10-20%): Used to tune hyperparameters and prevent overfitting.
- Test Set (10-20%): Used to evaluate model performance on unseen data.

Detailed Methodologies and Processes

AI and ML Relationship
- AI simulates human intelligence.
- ML is a technique within AI to learn from data.
Model Training and Inferencing Workflow
- Collect training data (images, text, etc.).
- Train model using ML algorithms (e.g., classification).
- Use trained model for inferencing on new data.
Neural Network Structure
- Input layer receives data.
- Multiple hidden layers process data.
- Output layer produces prediction or classification.
- Layers self-adjust weights based on input data.
Deep Learning Characteristics
- Uses multiple neural network layers.
- Learns complex features automatically.
- Continuously adapts and improves predictions.
Generative AI Workflow
- Train foundation model on massive datasets.
- Use prompt to generate new content.
- Outputs are probabilistic and nondeterministic.
Diffusion Model for Image Generation
- Training: Add noise to images until pure noise.
- Generation: Start with noise and add details to create image.
Training Data Types and Usage
- Labelled data for supervised learning.
- Unlabelled data for unsupervised learning.
- Structured data for numerical/categorical tasks.
- Unstructured data for complex content analysis.
Data Split for Model Development
- Training set for learning.
- Validation set for tuning.
- Test set for unbiased evaluation.

Speakers / Sources Featured

Primary Speaker / Instructor: Unnamed instructor guiding through AWS Certified Cloud Practitioner AI/ML boot camp.
References to AI Systems and Tools:
- Amazon Alexa (AI voice assistant example).
- ChatGPT (OpenAI’s large language model).
- Tesla (deep learning in self-driving cars).
- Various AI assistants: Apple Siri, Microsoft Cortana, Google Assistant, Samsung Bixby.
Mentioned Foundation Models and Companies:
- OpenAI (GPT-4, DALL·E).
- Amazon (Titan).
- Meta (LLaMA).
- Anthropic (Claude).
- Google (Bard, Gemini).
Model Example: Stable Diffusion (image generation diffusion model).

This summary captures the essence of the module’s content, explaining foundational AI/ML concepts, their applications, data handling, and emerging technologies like generative AI with clear examples and methodologies.