Environment setup, jupyter, python, tensor basics with numpy and PyTorch
All tutorial materials will be available on this page.
Supervised learning framework, binary and multiclass logistic regression, pytorch and autograd basics
MLP model, activations, backprop, loss functions and optimization in PyTorch
Convolutional and pooling layers, architectures, spatial classification, residual nets.
RNN implementation, sentiment analysis, Temporal Convolution Networks.
Transfer learning definition and contexts, fine-tuning pre-trained models, unsupervised domain adaptation via an adversarial approach.
Attention mechanisms, sequence-to-sequence models, machine translation.
The RL setting, openAI Gym, Deep q-learning for Atari games.
Filters on graphs, graph convolution layers, semi-supervised node classification
The CUDA programming model, numba, implementing CUDA kernels in python, thread synchronization, shared memory