Deep Learning with TensorFlow and PyTorch
Why colleges run this
Deep learning powers modern AI products; a student who has trained and measured a real neural network in both major frameworks holds a strongly differentiated skill.
Course outcomes
CO1
Explain how a neural network learns through backpropagation
K2CO2
Build and train a neural network in TensorFlow and PyTorch
K3CO3
Construct a convolutional network for image classification
K3CO4
Apply regularisation to improve model generalisation
K4CO5
Fine-tune a pre-trained model and report its measured accuracy
K4Modules tap a module for its theory & lab
Hours shown are the recommended 5-day format — module time scales to the duration you pick.
01Neural network foundations5 h
Theory
Neurons, activation functions, the forward pass, loss
Lab
Compute a forward pass by hand and in code
Output
Forward-pass notebook
02Training mechanics5 h
Theory
Backpropagation, gradient descent, optimisers
Lab
Train a small network from scratch
Output
Training notebook
03Frameworks5 h
Theory
TensorFlow and PyTorch, tensors, automatic differentiation
Lab
Rebuild the network in both frameworks
Output
Two-framework notebook
04Convolutional networks5 h
Theory
Convolution, pooling, image-classification architecture
Lab
Build an image classifier
Output
CNN notebook
05Sequence models5 h
Theory
Recurrent networks and an introduction to attention
Lab
Build a sequence model
Output
Sequence-model notebook
06Regularisation and tuning5 h
Theory
Dropout, batch normalisation, learning-rate schedules
Lab
Improve generalisation on a model
Output
Tuning report
07Transfer learning5 h
Theory
Fine-tuning pre-trained models
Lab
Fine-tune a pre-trained model
Output
Fine-tuning notebook
08Capstone5 h
Theory
An image or text model to a measured accuracy
Lab
Train and measure a capstone model
Output
Capstone model + accuracy report
Every participant receives
Certificate of completion Course material LMS access Interview question bank Mock interview & viva practice Optional nasscom NSQF assessment