Jill-Jênn Vie
Researcher at
Inria
Publications
Projects
Teaching
CV
Music
Slides
Deep learning at ENS Ulm
Introduction to Deep Learning
Lab:
Fine-tune VGG for cats vs. dogs
Automatic differentiation
Homework due Nov 6:
PyTorch from scratch
Guide soon (ask questions on Discord)
CNNs and ResNets
Lab:
CNNs and ResNets
(correction on Discord)
Homework due Dec 4:
class activation map, adversarial learning
(new link)
Regularization and Optimization
Lab:
Momentum and adaptive step sizes
Transformers I
Previous years
Embeddings, unsupervised learning
Lab:
word2vec as matrix factorization
Sequences: seq2seq, attention
Lab:
GPT
Homework: Code your own
Neural Machine Translator
in a
Kaggle competition
Deep reinforcement learning
Lab: Breakout
Graph Neural Networks
Tabular Deep Learning
&
why transformers work
Ethics, Biases,
Environmental Impact
Deep Generative Models
Multilayer Perceptrons: Expressiveness, overfitting, regularization
Stability and adversarial learning
Implicit layers
Fairness: Learning fair representations (ICML 2013) and FARE: provably fair representation learning (ICML 2023)
Competitive programming at École polytechnique
See
all slides
.
Intro to ICPC
Intro to pathfinding
Strings
Matching & flows
Segment trees
Machine learning
Jupyter notebooks
on GitHub
Supervised learning: regression, classification
Regression, classification
Write your own gradient descent
Write your own
backpropagation
Classification of handwritten digits
Unsupervised learning
Dimensionality reduction
Recommender systems
k
-means
Skip-gram & negative sampling: word2vec
t-SNE
Reinforcement learning
Policy improvement
Value iteration
Préparation à l’agrégation d’informatique
Exemples d’algorithmes d’apprentissage supervisé et non supervisé : KNN,
K
-moyennes
A* : heuristiques pour les plus courts chemins
Hors programme :
Modèles graphiques probabilistes
TP Viterbi :
Doudou le hamster surveillé