RTMNU · Second Year B.Tech Computer Science & Engineering (AI & ML) · NEP 2020
Machine Learning
BCSEAI&ML4T13 · Semester IV · 3 credits · College 30 + university 70
Unit 1: Introduction to Machine Learning
8 hoursComponents of learning; geometric, probabilistic and logic models, grouping and grading; designing a learning system; supervised, unsupervised and reinforcement learning, perspectives and issues; version spaces, PAC learning and the VC dimension.
Examined in the 70-mark end-semester paper
- What machine learning is
- Models of machine learning
- Types of learning
- Hypothesis space and inductive bias
- Computational learning theory and PAC learning
- Sample complexity and VC dimension
Unit 2: Supervised and Unsupervised Learning
7 hoursID3 and classification and regression trees; linear, multiple linear and logistic regression; the perceptron and multilayer perceptron; linear and non-linear support vector machines and kernel functions; k-nearest neighbours; k-means and k-modes clustering.
Examined in the 70-mark end-semester paper
- Decision tree learning and overfitting
- Simple and multiple linear regression
- Logistic regression
- Perceptron and multilayer networks
- Support vector machines
- K-nearest neighbours
- K-means and choosing k
- K-medoids and K-prototype
Unit 3: Ensemble and Probabilistic Learning
7 hoursModel combination schemes, voting and error-correcting output codes; bagging and random forests, AdaBoost and stacking; Gaussian mixture models and the EM algorithm, information criteria; nearest-neighbour smoothing, KD-trees and distance measures.
Examined in the 70-mark end-semester paper
- Voting ensembles
- Error-correcting output codes
- Bagging and random forests
- Boosting and stacking
- Gaussian mixture models
- KD-trees and efficient nearest-neighbour search
- Distance measures: Euclidean, Manhattan, Minkowski and Hamming
Unit 4: Reinforcement Learning and Evaluating Hypotheses
7 hoursThe learning task and Q-learning; non-deterministic rewards and actions; temporal-difference learning and its relation to dynamic programming; active reinforcement learning; bandit algorithms with UCB.
Examined in the 70-mark end-semester paper
- Elements of reinforcement learning
- Q-learning
- Temporal difference learning
- Dynamic programming for MDPs
- Multi-armed bandits
- Exploration strategies
Unit 5: RL Algorithms
7 hoursMonte Carlo, TD and TD(λ) algorithms, dynamic programming and eligibility traces; Bellman optimality, full RL and MDPs; on-policy and off-policy updates; case studies of RL.
Examined in the 70-mark end-semester paper
- Monte Carlo prediction
- Monte Carlo control
- Eligibility traces and TD(λ)
- Markov decision processes
- Bellman equation
- SARSA
- Importance sampling
Unit contents are summarised from the official RTMNU syllabus . AiCanCode.org is not affiliated with Rashtrasant Tukadoji Maharaj Nagpur University (RTMNU).