Mumbai University · Final Year Computer Engineering · Rev-2019 'C' Scheme
Machine Learning
CSC701 · Semester VII · 3 credits · Internal tests 20 (average of two) + end-semester 80
End-semester paper (3 hours): Six questions of 20 marks; you answer four. Each module's share of the paper follows its lecture hours.
Module 1: Introduction to Machine Learning
4 hoursTypes and issues of ML; applications; steps in developing an ML application; training and generalisation error; over- and underfitting; the bias–variance trade-off.
Examined in the 80-mark end-semester paper (about 10% of it, by lecture hours); the class tests cover what has been taught by then
Module 2: Learning with Regression and Trees
9 hoursLinear, multivariate and logistic regression; decision trees with the Gini index; CART; confusion matrix, kappa, sensitivity, specificity, precision, recall, F-measure and ROC.
Examined in the 80-mark end-semester paper (about 23% of it, by lecture hours); the class tests cover what has been taught by then
- Linear and polynomial regression
- Logistic regression
- Decision trees and CART
- Confusion matrix, precision, recall and F1
- ROC curves and the kappa statistic
Module 3: Ensemble Learning
6 hoursEnsembles; k-fold cross-validation; boosting, stumping and XGBoost; bagging, subagging and random forests; combining classifiers.
Examined in the 80-mark end-semester paper (about 15% of it, by lecture hours); the class tests cover what has been taught by then
- Ensemble learning and voting
- Cross-validation and micro and macro averaging
- Boosting and stacking
- XGBoost
- Bagging and random forests
Module 4: Learning with Classification
8 hoursConstrained optimisation and the optimal boundary; margins and support vectors; SVM as a quadratic program; linear and non-linear SVM and the kernel trick; support vector regression; multiclass classification.
Examined in the 80-mark end-semester paper (about 21% of it, by lecture hours); the class tests cover what has been taught by then
- Support vector machines and kernels
- SVM as constrained optimisation
- Tree-based and support vector regression
- Multiclass classification
Module 5: Learning with Clustering
7 hoursDistance metrics and clustering approaches; minimal-spanning-tree clustering; expectation maximisation; DBSCAN.
Examined in the 80-mark end-semester paper (about 18% of it, by lecture hours); the class tests cover what has been taught by then
- Clustering and hierarchical clustering
- Clustering with minimal spanning trees
- Expectation maximisation
- DBSCAN and Gaussian mixture models
Module 6: Dimensionality Reduction
5 hoursPCA, LDA and singular value decomposition.
Examined in the 80-mark end-semester paper (about 13% of it, by lecture hours); the class tests cover what has been taught by then
- PCA and LDA
- Singular value decomposition
Module contents are summarised from the official Mumbai University syllabus . AiCanCode.org is not affiliated with the University of Mumbai.