Schedule#
The following is a tentative schedule
Time |
June 2 (Monday) |
June 3 (Tuesday) |
June 4 (Wednesday) |
June 5 (Thursday) |
June 6 (Friday) |
June 9 (Monday) |
June 10 (Tuesday) |
June 11 (Wednesday) |
June 12 (Thursday) |
June 13 (Friday) |
|---|---|---|---|---|---|---|---|---|---|---|
|
Registration |
|||||||||
|
1. Preflight: 1. matrix operations; 2. Linear Regression (OLS) (80’) |
Introduction to DL: 1. Preflight - calculus, chain rule; 2. Gradient Descent |
Toy Regression Problem: 1. Implement GD with autodiff; 2. Implement SGD with autodiff; 3. Implement Mini-batch SGD with autodiff |
Convolutional Neural Networks |
MNIST Classification: 1. Convolutional Neural Networks; 2. GCN - Superpixels dataset; 3. Compare/contrast approaches |
Introduction to Transformers; 1. General idea (“dot products”); 2. Self attention; 3. Multi-head Self attention |
Implementing a basic GPT text model |
Introduction to Bayesian Methods: 1. Bayes Theorem; 2. Prior vs posterior; 3. Mathematical examples |
Bayesian Deep Learning: 1. Introduction; 2. Posteriors over weights; 3. Simple implementation |
Bayesian Optimization |
|
Coffee Break (30’) |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
|
Linear Classifiers (Perceptron, Adaline) Historical approach (80’) |
Optimizing Networks through GD: 1. GD, SGD, Mini-batch SGD; 2. Optimizers; 3. Regularization; 4. Schedulers |
1. Pytorch Fundamentals; 2. Classification with Neural Networks (binary); 3. Multiclass approaches; 4. Implement binary classifier in pytorch |
Graph Neural Networks |
Introduction to Gen AI: 1. VAE; 2. GAN; 3. Advanced Gen AI; 4. Diffusion Models |
GPT Text Models: 1. How does it work?; 2. Pretraining vs post training; 3. Key words they may hear (KV caching, Full Query, Grouped Query, Single Query) |
Introduction to Vision Transformers; 1. An image is worth 16x16 words; 2. Treating patches as tokens |
1. Bayesian Regression; 2. Gaussian Process |
Advanced Bayesian DL |
Bayesian Optimization |
|
Lunch (80’) |
Lunch |
Lunch |
Lunch |
Lunch |
Lunch |
Lunch |
Lunch |
Lunch |
Lunch |
|
General introduction : Fusion and Magnetic Confinement (Saskia Mordijck) (60’) |
Tokamak Operations: Pegasus (Steffi Diem) |
Tokamak Measurements (Eva Kostadinova) |
Tokamak MHD stability: Disruption (Cristina Rea, Alex Saperstein) |
Tokamak MHD stability: ELMs (Saskia Mordijck) |
Collaborative research projects (Nick Murphy) |
Tutorial MAST/MAST-U (Nathan Cummings) |
Digilab: Equilibrium reconstruction uncertainty (Cyd Cowley) |
Digilab: Equilibrium reconstruction uncertainty (Cyd Cowley) |
Group presentations |
|
Coffee Break (30’) |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
Coffee Break |
|
Getting setup with software/hardware, data explanation |
Hands-on (Explanation of Dataset) |
Hands-on |
Hands-on |
Hands-on |
Hands-on |
Hands-on |
Hands-on |
Hands-on / Hands-on |
Group presentations |
|
Welcome Reception |
Al4Fusion Social |
AI4Fusion Social |
Closing remarks |