Course code DEEP LEARNING AND ITS APPLICATIONS L T P J C
CSE6037 2 0 2 0 3
Pre-requisite Nil Syllabus version
v. xx.xx
Course Objectives:
1. To understand the theoretical foundations, algorithms and methodologies of Neural Network
2. To design and develop an application using specific deep learning models
3. To provide the practical knowledge in handling and analysing real world applications.
Expected Course Outcomes:
Upon completion of the course, the students will be able to
1. Recognize the characteristics of deep learning models that are useful to solve real-world problems.
2. Understand different methodologies to create application using deep nets.
3. Identify and apply appropriate deep learning algorithms for analyzing the data for variety of
problems.
4. Implement different deep learning algorithms
5. Design the test procedures to assess the efficacy of the developed model.
6. Combine several models in to gain better results
Module:1 MACHINE LEARNING BASICS 3 hours
Learning algorithms, Maximum likelihood estimation, Building machine learning algorithm, Neural
Networks Multilayer Perceptron, Back-propagation algorithm and its variants Stochastic gradient
decent, Curse of Dimensionality
Module:2 DEEP LEARNING ARCHITECTURES 9 hours
Machine Learning and Deep Learning, Representation Learning, Width and Depth of Neural
Networks, Activation Functions: RELU, LRELU, ERELU, Unsupervised Training of Neural
Networks, Restricted Boltzmann Machines, Auto Encoders, Deep Learning Applications
Module:3 CONVOLUTIONAL NEURAL NETWORKS 3 hours
Architectural Overview,Motivation, Layers, Filters, Parameter sharing, Regularization, Popular CNN
Architectures: ResNet, AlexNet - Applications
Module:4 TRANSFER LEARNING 5 hours
Transfer learning Techniques, Variants of CNN: DenseNet, PixelNet.
Module:5 SEQUENCE MODELLING RECURRENT AND 3 hours
RECURSIVE NETS
Recurrent Neural Networks, Bidirectional RNNs, Encoder-decoder sequence to sequence
architectures - BPTT for training RNN, Long Short Term Memory Networks.
Module:6 AUTO ENCODERS 3 hours
Under complete Auto encoder, Regularized Auto encoder, stochastic Encoders and Decoders,
Contractive Encoders.
Module:7 DEEP GENERATIVE MODELS 2 hours
Deep Belief networks, Boltzmann Machines,Deep Boltzmann Machine, Generative Adversial
Networks.
Module:8 RECENT TRENDS 2 hours
Total Lecture hours: 30 hours
Text Book(s) and Journals
1.
2. Josh Patterson, Adam Gibson "Deep Learning: A Practitioner's Approach", O'Reilly Media,
2017
3. Umberto Michelucci Deep Learning. A Case-based Approach to Understanding
Deep Neural Apress, 2018.
Reference Books
1. Kevin P. Murphy "Machine Learning: A Probabilistic Perspective", The MIT Press, 2012.
2. EthemAlpaydin,"Introduction to Machine MIT Press, Prentice Hall of India, Third
3. Edition 2014.
4. Giancarlo Zaccone, Md. RezaulKarim, Ahmed Menshawy "Deep Learning with TensorFlow:
5. Explore neural networks with Python", Packt Publisher, 2017.
Antonio Gulli, Sujit Pal "Deep Learning with Keras", Packt Publishers, 2017.
Francois Chollet "Deep Learning with Python", Manning Publications, 2017.
Mode of Evaluation: CAT / Assignment / Quiz / FAT / LAB / Seminar
List of Indicative Experiments
1. Train a Deep learning model to classify a given image using pre trained model
2. Object detection using Convolution Neural Network
3. Recommendation system from sales data using Deep Learning
4. Improve the Deep learning model by tuning hyper parameters
5. Perform Sentiment Analysis in network graph using RNN
6. Image generation using GAN
Total Laboratory Hours 30 hours
Mode of assessment: CAT / Assignment / Quiz / FAT S
Recommended by Board of Studies 11-06-2019
Approved by Academic Council No. 56 Date 24-09-2019
