Machine Learning · Priority List

For when time is short. In recent MAKAUT papers, a few topics carry most of the marks.

# Reading Guide

• Read Priority 1 first, then Priority 2. Inside each, topics are in study order.
• For everything else, read the ⚡ Quick recall box of each topic for the 1-mark questions.
• Full topic list → ML

# Priority 1 · 12 Topics, 50% of Marks

⏱ about 1 h 3 min.

• 0.1Introduction to Machine Learning: Definition, AI vs ML vs DL★★ PYQ ⏱ 8 min
• 1.3K-Nearest Neighbours (KNN)★★★ ⏱ 5 min
• 1.4Decision Trees: ID3, Entropy & Information Gain★★ ⏱ 6 min
• 1.7Naive Bayes Classifier★★ ⏱ 5 min
• 2.2K-means Clustering★★★ ⏱ 6 min
• 2.4Dimensionality Reduction & the Curse of Dimensionality★★★ ⏱ 4 min
• 2.5Feature Selection★★ ⏱ 4 min
• 2.9Matrix Factorization & Matrix Completion★★ ⏱ 6 min
• 2.10Generative Models: Mixture Models (GMM & EM)★★ ⏱ 5 min
• 3.1Confusion Matrix & Classification Metrics★★★ ⏱ 5 min
• 3.3Overfitting, Underfitting & the Bias–Variance Trade-off★★★ ⏱ 5 min
• 3.7Bagging & Bootstrapping★★★ ⏱ 4 min

# Priority 2 · 16 More Topics, Up to 88%

⏱ about 1 h 16 min.

• 0.2Types of Machine Learning★★★ PYQ ⏱ 4 min
• 0.4Parametric vs Non-parametric Models★★★ PYQ ⏱ 3 min
• 1.1Supervised Learning: Classification vs Regression★★ ⏱ 3 min
• 1.8Linear Regression★★★ ⏱ 6 min
• 1.9Logistic Regression★★★ ⏱ 4 min
• 2.1Clustering: Concepts, Requirements & Types★ ⏱ 6 min
• 2.6Principal Component Analysis (PCA)★★★ ⏱ 5 min
• 2.8Recommender Systems: Content-Based & Collaborative Filtering★★ PYQ ⏱ 5 min
• 3.4Model Selection & Cross-Validation★★ ⏱ 5 min
• 3.6Ensemble Learning: Principle & Types★★★ ⏱ 4 min
• 3.8Boosting & Bagging vs Boosting★★★ ⏱ 4 min
• 3.9Random Forest★★ ⏱ 4 min
• 4.1Sparse Modeling & Estimation★★ ⏱ 6 min
• 4.3Neural Network Fundamentals: Perceptron, Activation Functions & Backpropagation★★★ ⏱ 6 min
• 4.4Deep Learning: Concepts, ML vs DL & Applications★★★ ⏱ 5 min
• 5.4Reinforcement Learning★★★ ⏱ 6 min

# Everything Else · 22 Topics

The remaining 12% of marks is spread thinly across these topics, mostly as 1-mark questions.