Artificial Intelligence (AI) & Machine Learning (ML) Notes

What is Artificial Intelligence (AI)?

Artificial Intelligence (AI) is the field of computer science that focuses on creating systems capable of performing tasks that normally require human intelligence.

Examples:

• Chatbots
• Self-driving cars
• Recommendation systems
• Voice assistants
• Medical diagnosis systems

AI Overview

Common AI Tasks

---

What is Machine Learning (ML)?

Machine Learning is a subset of AI that allows computers to learn patterns from data without being explicitly programmed.

Instead of writing:

pythonCopy
1if email.contains("win money"):
2    spam = True

ML learns automatically:

textCopy
1Input Data → Training → Model → Prediction

Machine Learning Workflow

---

AI vs ML

Relationship

textCopy
1Artificial Intelligence
2│
3├── Machine Learning
4│     │
5│     ├── Deep Learning
6│     │      ├── CNN
7│     │      ├── RNN
8│     │      └── Transformers

---

Types of Machine Learning

1. Supervised Learning

Uses labeled data.

Example:

textCopy
1Input: House Features
2Output: House Price

Algorithms:

• Linear Regression
• Logistic Regression
• Random Forest
• XGBoost

Supervised Learning

---

2. Unsupervised Learning

Uses unlabeled data.

Goal:

• Find hidden patterns
• Group similar data

Algorithms:

• K-Means
• DBSCAN
• PCA

Example:

textCopy
1Customer Data
2       ↓
3   Clustering
4       ↓
5Customer Groups

Unsupervised Learning

---

3. Reinforcement Learning

Agent learns by rewards and penalties.

Example:

• Chess AI
• Self-driving cars
• Game AI

Reinforcement Learning

---

Deep Learning

Deep Learning is a subset of ML that uses neural networks with many layers.

textCopy
1Input Layer
2     ↓
3Hidden Layers
4     ↓
5Output Layer

Neural Network Structure

---

ANN (Artificial Neural Network)

Inspired by the human brain.

ANN Structure

textCopy
1Input
2  ↓
3Hidden Layer
4  ↓
5Output

Use Cases:

• Classification
• Prediction
• Pattern Recognition

---

CNN (Convolutional Neural Network)

Best for images.

Applications:

• Face Detection
• Medical Imaging
• Object Detection

CNN Visualization

How CNN Works:

textCopy
1Image
2 ↓
3Convolution
4 ↓
5Pooling
6 ↓
7Fully Connected Layer
8 ↓
9Prediction

---

RNN (Recurrent Neural Network)

Designed for sequential data.

Applications:

• Language Translation
• Speech Recognition
• Time Series Forecasting

RNN Visualization

---

Transformers

Modern architecture powering today's LLMs.

Used in:

• ChatGPT
• Claude
• Gemini
• Translation Systems

Transformer Architecture

Advantages:

• Parallel Processing
• Long Context Understanding
• Better Accuracy
• Scalable Training

---

Training Process of an AI Model

textCopy
1Collect Data
2      ↓
3Clean Data
4      ↓
5Tokenization
6      ↓
7Training
8      ↓
9Validation
10      ↓
11Testing
12      ↓
13Deployment

AI Training Pipeline

---

Popular AI Frameworks

---

Real-World AI Applications

Healthcare

• Disease Detection
• Medical Imaging
• Drug Discovery

Finance

• Fraud Detection
• Risk Analysis
• Algorithmic Trading

Education

• Personalized Learning
• AI Tutors

Technology

• Search Engines
• Chatbots
• Virtual Assistants

AI Applications

---

Quick Summary

textCopy
1AI
2 └── ML
3      └── Deep Learning
4            ├── ANN
5            ├── CNN
6            ├── RNN
7            └── Transformers

• AI = Making machines intelligent.
• ML = Learning patterns from data.
• Deep Learning = Neural networks with many layers.
• CNN = Best for images.
• RNN = Best for sequences.
• Transformers = Best for modern language models and LLMs.

These notes provide a strong foundation before learning advanced topics such as attention mechanisms, embeddings, tokenization, vector databases, RAG, fine-tuning, LoRA, quantization, and LLM training.