I will start by talking about the convolution operation, the motivation of choosing it over others, and the basics of convolutional neural networks (convolutional layer, pooling layers).

I will then briefly go over the architecture of a famous convolution network. I will end by talking about some applications and sharing a Colab notebook of at least one convolutional network for semantic segmentation (using PyTorch).

If time permits and people are interested, I will be happy to touch on some pertinent advanced topics like transpose convolutions etc.

Useful Links

• Google Colab Notebook
• Stanford Introduction to Convolutional Neural Networks for Visual Recognition
• Stanford CS224N: NLP with Deep Learning
• Deep Learning by Ian Goodfellow, Yoshua Bengio and Aaron Courville