A4.1.1 Types of Machine Learning and Their Applications in the Real World Notes

What is Machine Learning?

1. Machine learning is a type of artificial intelligence where computers learn from data and improve their performance over time without being explicitly programmed.
2. Algorithms detect patterns, make decisions, and predict outcomes based on previous examples.

Supervised Learning

1. Supervised learning involves training a model on a labeled dataset, where each input is paired with a corresponding output.
2. Types:
   1. Classification: Predicts discrete categories (e.g., spam or not spam).
   2. Regression: Predicts continuous values (e.g., house prices).

Unsupervised Learning

1. Unsupervised learning deals with unlabeled data, aiming to find hidden patterns or structures.
2. Types:
   1. Clustering: Groups similar data points (e.g., customer segmentation).
   2. Association rule mining (finding item relationships)
   3. Dimensionality Reduction: Reduces data complexity while preserving important features (e.g., PCA).

Reinforcement Learning

1. Reinforcement learning involves an agent learning to make decisions by interacting with an environment to maximize cumulative reward.
2. Key Concepts:
   1. Exploration vs. Exploitation: Balancing trying new actions and using known strategies.
   2. Policy: A strategy that defines the agent's actions.

Deep Learning

1. Deep learning is a subset of machine learning that uses artificial neural networks with multiple layers to automatically extract and learn hierarchical features from large, complex data sets.
2. Architectures:
   1. Convolutional Neural Networks (CNNs): Ideal for image processing.
   2. Recurrent Neural Networks (RNNs): Designed for sequential data like text.
   3. Transformers – modern Natural Language Processing (NLP) and vision models

Transfer Learning

1. Transfer learning involves reusing a pre-trained model on a new, related task, with the aim to reduce the time and data needed for training a new model.
2. How it works:
   1. Use a pre-trained model (e.g. on a general dataset)
   2. Fine-tune it on a smaller, specific dataset
3. Benefits:
   1. Reduces training time.
   2. Improves performance with limited data.

Real-World Applications of Machine Learning

Market Basket Analysis

1. Description: Identifies associations between products in transaction data.
2. Algorithm: Apriori (Unsupervised Learning).

Medical Imaging Diagnostics

1. Description: Uses deep learning models like CNNs to analyze medical images.
2. Learning Paradigms:
   1. Supervised Learning: Classifies images into diagnostic categories.
   2. Transfer Learning: Fine-tunes pre-trained models for specific medical tasks.

Natural Language Processing (NLP)

1. Description: Involves processing and understanding human language.
2. Techniques:
   1. Deep Learning: Uses transformers for tasks like translation and sentiment analysis.
   2. Transfer Learning: Models like BERT are pre-trained on large text datasets and fine-tuned for specific tasks.

Object Detection and Classification

1. Description: Identifies and classifies objects within images.
2. Learning Paradigms:
   1. Supervised Learning: Trains models on labeled images.
   2. Deep Learning: Uses CNNs to recognize patterns and features.
3. Used in:
   1. Autonomous vehicles
   2. Face recognition
   3. Surveillance and retail

Robotics Navigation

1. Description: Enables robots to navigate and interact with their environment.
2. Learning Paradigms:
   1. Reinforcement Learning: Teaches robots to make decisions through trial and error.
   2. Supervised Learning: Used for tasks like obstacle recognition.

Sentiment Analysis

1. Description: Analyzes text data to determine sentiment (positive, negative, neutral).
2. Learning Paradigms:
   1. Supervised Learning: Trains models on labeled datasets.
   2. Deep Learning: Uses LSTMs and transformers to capture contextual nuances.
   3. Transfer Learning: Adapts pre-trained models to specific domains.

Summary

1. Supervised Learning: Ideal for tasks with labeled data, such as medical imaging and sentiment analysis.
2. Unsupervised Learning: Used for discovering patterns in unlabeled data, like market basket analysis.
3. Reinforcement Learning: Suited for decision-making in dynamic environments, such as robotics navigation.
4. Deep Learning: Powers complex tasks like image recognition and NLP through neural networks.
5. Transfer Learning: Enhances model performance by leveraging pre-trained models for new tasks.