SPPU · Final Year Computer Engineering · 2019 Course (Revised)
Machine Learning
410242 · Semester VII · 3 credits · In-semester 30 + end-semester 70
Shares notes with the third-year course of the same name; the final-year units add their own topics.
Unit 1: Introduction to Machine Learning
7 hoursML vs traditional programming, ML vs AI vs data science; supervised, unsupervised, semi-supervised and reinforcement learning; geometric, probabilistic, logical, grouping and grading, parametric and non-parametric models; data formats, learnability, statistical learning.
Marks: 15 in the in-semester exam
Unit 2: Feature Engineering
7 hoursNormalisation, scaling, standardisation and missing values; dimensionality reduction and PCA; kernel PCA and local binary patterns; sequential forward and backward selection; statistical features; multidimensional scaling and matrix factorisation.
Marks: 15 in the in-semester exam
- Feature engineering
- PCA and LDA
- Kernel PCA and local binary patterns
- Feature selection: sequential forward and backward
- Multidimensional scaling and matrix factorisation
Unit 3: Supervised Learning: Regression
6 hoursBias, variance, generalisation, under- and overfitting; linear, lasso and ridge regression; gradient descent; MAE, RMSE and R².
Marks: 17–18 in the end-semester exam
Unit 4: Supervised Learning: Classification
8 hoursKNN and SVM; bagging, boosting, random forest and AdaBoost; binary vs multiclass, balanced and imbalanced, one-vs-one and one-vs-all; accuracy, precision, recall, F-score, cross-validation, micro- and macro-averaging.
Marks: 17–18 in the end-semester exam
Unit 5: Unsupervised Learning
7 hoursK-means, K-medoids, hierarchical and density-based clustering, spectral clustering; outlier analysis with isolation forest and local outlier factor; the elbow method, extrinsic and intrinsic evaluation.
Marks: 17–18 in the end-semester exam
- K-means and K-medoids
- Clustering and hierarchical clustering
- DBSCAN and Gaussian mixture models
- Spectral clustering
- Outlier detection: isolation forest and LOF
- Evaluating clusters: elbow, extrinsic and intrinsic measures
Unit 6: Introduction to Neural Networks
7 hoursSingle-layer networks, the multilayer perceptron, backpropagation, functional link networks, radial basis function networks, activation functions; introduction to RNNs and CNNs.
Marks: 17–18 in the end-semester exam
- Perceptron and multilayer perceptron
- Backpropagation learning
- Functional link and radial basis function networks
- Activation functions
- Introduction to CNNs and RNNs
Unit contents are summarised from the official SPPU syllabus . AiCanCode.org is not affiliated with Savitribai Phule Pune University.