Deep Learning Summer School 2015

Lecture archive  ·  Montréal, August 3–12, 2015

This page archives the lecture slides from the 2015 CRM/CIFAR Deep Learning Summer School in Montréal, which I co-organized with Yoshua Bengio and Yann LeCun. The slides were originally hosted on the organizer's directory at the Université de Montréal (iro.umontreal.ca/~memisevr/dlss2015/); those links are no longer live. This page mirrors the archive to keep the materials reachable. Video recordings of most lectures are available via VideoLectures.net.

Day 1 — Monday 3 August

• Pascal Vincent Intro to machine learning slides
• Yoshua Bengio Theoretical motivations for representation learning & deep learning slides
• Léon Bottou Intro to multi-layer nets slides

Day 2 — Tuesday 4 August

• Hugo Larochelle Neural nets and backprop slides
• Léon Bottou Numerical optimization & SGD; structured problems & reasoning slides
• Hugo Larochelle Directed graphical models and NADE slides

Day 3 — Wednesday 5 August

• Aaron Courville Intro to undirected graphical models slides
• Honglak Lee Stacks of RBMs slides
• Pascal Vincent Denoising and contractive autoencoders, manifold view slides

Day 4 — Thursday 6 August

• Roland Memisevic Visual features slides
• Honglak Lee Convolutional networks slides
• Graham Taylor Learning similarity slides

Day 5 — Friday 7 August

• Chris Manning NLP 101 slides
• Graham Taylor Modeling human motion, pose estimation and tracking slides
• Chris Manning NLP / deep learning slides

Day 6 — Saturday 8 August

• Ruslan Salakhutdinov Deep Boltzmann Machines slides
• Adam Coates Speech recognition with deep learning slides
• Ruslan Salakhutdinov Multi-modal models slides

Day 7 — Sunday 9 August

• Ian Goodfellow Structure of optimization problems slides
• Adam Coates Systems issues & distributed training slides
• Ian Goodfellow Adversarial examples slides

Day 8 — Monday 10 August

• Phil Blunsom From language modeling to machine translation slides
• Richard Socher Recurrent neural networks slides
• Phil Blunsom Memory, reading, and comprehension slides

Day 9 — Tuesday 11 August

• Richard Socher Dynamic memory networks for NLP slides
• Mark Schmidt Smooth, finite, and convex optimization slides
• Roland Memisevic Visual features II slides

Day 10 — Wednesday 12 August

• Mark Schmidt Non-smooth, non-finite, and non-convex optimization slides
• Aaron Courville VAEs and deep generative models for vision slides
• Yoshua Bengio Generative models from autoencoders slides

Programming-tutorial materials (Theano / Fuel exercises) are on the MILA GitHub repository.

Note to speakers: if you'd prefer your slides not be mirrored here, please email me and I'll remove them.