DBATU · Final Year B.Tech Information Technology · Non-NEP
Machine Learning
BTITOE704B · Elective II · Semester VII · 3 credits · Continuous assessment 20 + mid-semester 20 + end-semester 60
One of the Elective II options.
Unit 1: Introduction and Evaluating Hypotheses
Types of learning; well-posed learning problems and designing a learning system; hypothesis space and inductive bias; evaluation and cross-validation; estimating hypothesis accuracy, confidence intervals, hypothesis testing and comparing learning algorithms.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Machine learning workflow
- Well-posed learning problems and designing a learning system
- Evaluating hypotheses
- Cross-validation and hyperparameter tuning
Unit 2: Bayesian Learning and Learning Theory
Bayes' theorem and concept learning; maximum-likelihood and least-squares hypotheses; minimum description length; the Bayes optimal classifier and Gibbs algorithm; PAC learning, sample complexity and the VC dimension; linear and logistic regression; instance-based learning; feature reduction; collaborative filtering; bias–variance, overfitting and regularisation.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Bayesian learning: MAP, MDL, Bayes optimal and Gibbs
- PAC learning and VC dimension
- Linear and polynomial regression
- Logistic regression
- K-nearest neighbours and naïve Bayes
- Bias–variance trade-off and responsible AI
Unit 3: Neural Networks and Optimisation
The McCulloch–Pitts neuron and threshold logic; perceptrons and their learning algorithm; multilayer perceptrons and sigmoid neurons; feedforward networks and backpropagation; gradient descent, momentum, Nesterov, stochastic GD, AdaGrad, RMSProp and Adam.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Perceptrons and multilayer perceptrons
- Backpropagation in feedforward networks
- Gradient descent optimisers: momentum to Adam
Unit 4: Autoencoders and Regularisation
PCA and SVD; autoencoders and their relation to PCA; denoising, sparse and contractive autoencoders; L2 regularisation, early stopping, data augmentation, parameter sharing, noise injection, ensembles and dropout; layer-wise pre-training, activation functions, weight initialisation and batch normalisation.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- PCA and dimensionality reduction
- Autoencoders: denoising, sparse and contractive
- Regularisation, early stopping and dropout
Unit 5: Convolutional and Recurrent Networks
LeNet, AlexNet, ZF-Net, VGGNet, GoogLeNet and ResNet; visualising CNNs, guided backpropagation, Deep Dream and fooling CNNs; recurrent networks, backpropagation through time, vanishing and exploding gradients, GRUs and LSTMs.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- CNN architectures: LeNet to ResNet
- Visualising convolutional networks
- Recurrent networks and backpropagation through time
- LSTM and GRU
Unit contents are summarised from the official DBATU syllabus . AiCanCode.org is not affiliated with Dr. Babasaheb Ambedkar Technological University (DBATU).