Course Title: 
Deep Learning
Course Code: 
Year Taught: 
Postgraduate (PG)
School of Engineering

'Deep Learning' is a Soft Core course offered for the M. Tech. in Computer Science and Engineering program at School of Engineering, Amrita Vishwa Vidyapeetham.

Neural Networks basics - Binary Classification, Logistic Regression, Gradient Descent, Derivatives, Computation graph, Vectorization, Vectorizing logistic regression – Shallow neural networks: Activation functions, non-linear activation functions, Backpropagation, Data classification with a hidden layer – Deep Neural Networks: Deep L-layer neural network, Forward and Backward propagation, Deep representations, Parameters vs Hyperparameters, Building a Deep Neural Network (Application) - Supervised Learning with Neural Networks – Practical aspects of Deep Learning: Train/Dev / Test sets, Bias/variance, Overfitting and regularization, Linear models and optimization, Vanishing/exploding gradients, Gradient checking – Logistic Regression, Convolution Neural Networks, RNN and Backpropagation – Convolutions and Pooling – Optimization algorithms: Mini-batch gradient descent, exponentially weighted averages, RMSprop, Learning rate decay, problem of local optima, Batch norm – Parameter tuning process.

Neural Network Architectures – Recurrent Neural Networks, Adversarial NN, Spectral CNN, Self-Organizing Maps, Restricted Boltzmann Machines, Long Short-Term Memory Networks (LSTM) and Deep Reinforcement Learning – TensorFlow, Keras or MatConvNet for implementation.


  1. Deep Learning, Ian Goodfellow, Yoshua Bengio and Aeron Courville, MIT Press,First Edition, 2016.
  2. Deep Learning, A practitioner’s approach, Adam Gibson and Josh Patterson, O’Reilly, First Edition, 2017.
  3. Hands-On Learning with Scikit-Learn and Tensorflow, Aurelien Geron, O’Reilly, First Edition, 2017.
  4. Deep Learning with Python, Francois Chollet, Manning Publications Co, First Edition, 2018.
  5. Python Machine Learning by Example, Yuxi (Hayden) Liu, First Edition, 2017.
  6. A Practical Guide to Training Restricted Boltzmann Machines, Geoffrey Hinton, 2010,
  Course Outcome Bloom’s Taxonomy Level
CO 1 Apply deep neural networks from building to training models L3
CO 2 Understand and use dropout regularization, Batch normalization and gradient checking in deep neural nets/td> L2
CO 3 Apply mini-batch, gradient descent, Momentum, RMSprop and Adam optimization algorithms with convergence L3
CO 4 Understand train/dev/test datasets and test bias/variance L4
CO 5 Analyse neural networks using tools - Tensorflow/Keras/MatConvNet. L4
CO 6 Analyse detection and recognition tasks using convolution/adversarial neural networks. L4