Deep Learning section of the Algorithms in Machine Learning class at ISAE-Supaero
Adapted from Emmanuel Rachelson's Machine Learning class
This class covers deep learning from a theoretical basis to example applications. We start with simple multi-layer perceptrons, backpropogation, and gradient descent, exploring at the fundamental aspects of deep learning in depth. We cover a wide range of deep learning topics, from Natural Language Processing to Generative Adversarial Networks; the full schedule is below. The goal is that students understand the capacities of deep learning, the current state of the field, and the challenges of using and developing deep learning algorithms. By the end of this class, we expect students that students will be able to understand recent literature in deep learning, implement novel neural network architectures, use and understand the PyTorch library in many ways, and apply deep learning to different domains.
| Schedule | ||
|---|---|---|
| 12/11 | Artificial Neural Networks | ANNs, backpropagation, Stochastic Gradient Descent |
| 12/11 | Deep Learning | layers, convolution, architectures, training |
| 18/11 | Deep Learning for Computer Vision, pt 1 | Convolutional Neural Networks, satellite imagery |
| 18/11 | Deep Learning for Computer Vision, pt 2 | |
| 25/11 | Image generation | VAEs, GANs, and Diffusion Models |
| 02/12 | RNNs | Recurrent Neural Networks, LSTM, GRU |
| 02/12 | Transformers | Transformers |
| 09/12 | Model Analysis | Autoencoders, t-SNE, SAEs |
| 09/12 | LLMs | Large Language Models |