| Course label : | Machine learning 3: Deep learning |
|---|---|
| Teaching departement : | EEA / Electrotechnics - Electronics - Control Systems |
| Teaching manager : | Mister PIERRE-ANTOINE THOUVENIN / Mister PIERRE CHAINAIS |
| Education language : | |
| Potential ects : | 0 |
| Results grid : | |
| Code and label (hp) : | MR_DS_S2_ML3 - Machine learning 3: Deep learn |
Education team
Teachers : Mister PIERRE-ANTOINE THOUVENIN / Mister PIERRE CHAINAIS
External contributors (business, research, secondary education): various temporary teachers
Summary
● reminder (ML1) + some complements: introduction to formal neural networks, the perceptron, training a perceptron, multilayer perceptron, full presentation of the backpropagation of the gradient of the error (incl. tricks to make it work in practice) ● elements of the formal analysis of neural networks (incl. MLP and their approximation ability) ● limitations of the classical MLP + backprop approach (vanishing or exploding gradient, etc.) ● the renewal of neural networks: deep learning and convolutional networks (conv / pool layers) ● deep net as a representation learner (auto-encoders, restricted Boltzmann machines) ● efficient deep net training (batch normalization, dropout, regularization etc.) ● recurrent neural networks and long-short time memories ● generative adversarial networks.
Educational goals
After successfully taking this course, a student should: ● know and understand the main concepts related to neural networks ● know the main types of neural networks (feedforward, convolutional, recurrent) and neurons ● know the main algorithms to train a neural network in practice ● understand the design of a deep neural network ● know how to use a neural network in practice to solve a particular supervised learning problem ● understand the limits of neural networks ● know theoretical properties of deep nets
Sustainable development goals
Knowledge control procedures
Continuous Assessment
Comments: Labs, grading scale: (min) 0 – 20 (max)
Exam, grading scale: (min) 0 – 20 (max)
Online resources
Hastie & Tibshirani, The Elements of Statistical Learning, Springer 2009. Goodfellow, Bengio, Courville, The Deep Learning book, MIT Press, 2016.
Pedagogy
24 hours, 6 lectures 6 exercises / labs Language of instruction is specified in the course offering information in the course and programme directory. English is the default language.
Sequencing / learning methods
| Number of hours - Lectures : | 12 |
|---|---|
| Number of hours - Tutorial : | 12 |
| Number of hours - Practical work : | 0 |
| Number of hours - Seminar : | 0 |
| Number of hours - Half-group seminar : | 0 |
| Number of student hours in TEA (Autonomous learning) : | 0 |
| Number of student hours in TNE (Non-supervised activities) : | 0 |
| Number of hours in CB (Fixed exams) : | 0 |
| Number of student hours in PER (Personal work) : | 0 |
| Number of hours - Projects : | 0 |
Prerequisites
ML1. ML2. Python & tools for research. Bases of optimization.