9.1 Machine Learning Overview
Three main paradigms of machine learning:
| Type | Description | Examples |
|---|---|---|
| Supervised | Learns from labeled examples | Classification, regression |
| Unsupervised | Finds patterns in unlabeled data | Clustering, dimensionality reduction |
| Reinforcement | Learns through rewards/penalties | Game playing, robotics |
9.2 Supervised Learning
Classification
Predicts discrete class labels:
- Decision Trees
- Naive Bayes
- Support Vector Machines (SVM)
- Neural Networks
Regression
Predicts continuous values:
- Linear Regression
- Polynomial Regression
- Regression Trees
9.3 Unsupervised Learning
Clustering
Groups similar data points:
- K-means
- Hierarchical Clustering
- DBSCAN
Dimensionality Reduction
Reduces number of variables:
- Principal Component Analysis (PCA)
- t-SNE
9.4 Reinforcement Learning
Key components:
- Agent: The learner/decision maker
- Environment: World the agent interacts with
- Actions: What the agent can do
- Reward: Feedback from environment
- Policy: Strategy the agent follows
Algorithms: Q-learning, Deep Q Networks (DQN), Policy Gradients
9.5 Frequently Asked Exam Questions
- Compare supervised vs unsupervised learning with examples.
- Explain the reinforcement learning framework.
- What is the bias-variance tradeoff?
- Describe how Q-learning works.
- When would you use PCA vs t-SNE?