DBATU · Third Year B.Tech Artificial Intelligence & Allied · NEP 2020
Deep Learning
26AFAIPC601 · Semester VI · 3 credits · Continuous assessment 20 + mid-semester 20 + end-semester 60
Unit 1: Machine Learning Basics and Neural Networks
5 hoursLinear and logistic regression, k-nearest neighbours, probabilistic classifiers, decision trees, random forests and SVMs; the artificial neuron and perceptron, stochastic gradient descent and backpropagation; network architectures with one hidden layer and multiple outputs; network hyperparameters.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Biological and artificial neurons
- Perceptron learning rule
- Backpropagation
- Gradient descent variants
- Feed-forward network architectures
Unit 2: Deep Architectures and CNNs
5 hoursWhy deep networks; hyperparameters: encoding, layers, loss, learning rate, momentum, optimisation, regularisation, dropout and batch normalisation; vanishing gradients; from dense layers to convolutions and pooling; AlexNet, VGG, NiN, GoogLeNet, ResNet and DenseNet; segmentation and object detection.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Hyperparameters in deep learning
- Regularisation and dropout
- Vanishing and exploding gradients
- Convolution and feature maps
- CNN architectures: LeNet to ResNet
- Object detection with YOLO
Unit 3: Deep Sequence Models
4 hoursSequence modelling problems; recurrent networks and backpropagation through time; GRUs, LSTMs and deep RNNs; image captioning and video-to-text with LSTMs.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Recurrent neural networks
- LSTM
- GRU
- Bidirectional and deep RNNs
- Image captioning and video to text
Unit 4: Deep Unsupervised Learning and Transfer Learning
5 hoursLatent variable models and autoencoders; variational autoencoders and GANs for image generation; transfer learning, its process and data augmentation.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Autoencoders and VAEs
- Generative adversarial networks
- Transfer learning
- Data augmentation
Unit 5: Responsible AI, MLOps and Applications
5 hoursExplainable AI, fairness and bias, responsible and ethical AI; evaluation metrics and deployment strategies; MLOps, cloud deployment, monitoring and maintenance; AI agents and tool use; case studies in healthcare, finance, manufacturing, smart cities and autonomous systems.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Explainable and responsible AI
- Deploying deep learning models
- MLOps and model monitoring
- AI agents and tool use
Unit contents are summarised from the official DBATU syllabus . AiCanCode.org is not affiliated with Dr. Babasaheb Ambedkar Technological University (DBATU).