Former teaching material

This is a part of the teaching material I created when I was associate professor (maître de conférences, HDR) at INSA de Lyon and the LIRIS laboratory from 2005 to the end of 2021.

Cours+TD+TP Deep Learning and Differentiable Programming (IF - 5ème année)

Part 1 - Introduction 1.1: Introduction: machine learning, a couple of applications [36 slides] 1.2: A short history of deep learning [8 slides] 1.3: An extremely short crash course on fitting and generalization [16 slides] Part 2 - Neural Networks and PyTorch 2.1: Frameworks and Tensors {+PyTorch} [24 slides] 2.2: (Generalized) Linear Models {+PyTorch} [32 slides] 2.3: Multi layer models + Universal approximation theorem {+PyTorch} [32 slides] 2.4: Train/Validation/Test split; Tensorboard {+PyTorch} [14 slides] 2.5: Gradient Backpropagation & Autograd {+PyTorch} [17 slides] 2.6: Stochastic Gradient Descent (SGD) and Variants (ADAM, RMSProp) {+PyTorch} [15 slides] 2.7: Shift invariance (equivariance) and Convolutions {+PyTorch} [38 slides] (see also blog post on shift equivariance) Part 3 - Scaling up: computer vision, transfer learning, visualization 3.1: Computer Vision [41 slides] 3.2: Visualization of learned knowledge [20 slides] 3.3: Transfer learning [19 slides] 3.4: Semi-supervised and self-supervised learning [8 slides] 3.5: GPUs - software {+CUDA, +PyTorch} [14 slides] Part 4 - Structure: sequences, graphs, attention 4.1: Recurrent Neural Networks and variants [35 slides] 4.2: Attention mechanisms in computer vision [41 slides] 4.3: Graphs and relational reasoning [37 slides] 4.4: Self-attention and transformers [19 slides] Part 5 - Advanced applications 5.1: Reinforcement Learning [25 slides] 5.2: Should we model or learn? [27 slides] 5.3: Learning robot navigation [61 slides] 5.4: Machine Learning as an Experimental Science [9 slides] 5.5: Conclusions [4 slides] Part 6 - A gentle introduction into learning theory 5.1: A gentle introduction into learning theory [45 slides] 5.2: A concrete application of the usage of learning theory in DL [35 slides] PyTorch sources: The full PyTorch sources of the slides are here. Pratical Excercise session: (will be held by Edward Beeching, web site will be published shortly.) Neural models in python for object detection and future forecasting. [Project page]

TD Advanced algorithms and data structures for Artificial Intelligence (IF - 3ème année)

With Christine Solnon: Shortest path problems, topological orderings of graphs, minimum spanning trees, the travelling saleman problem etc.

Cours+TD+TP Computer Architecture / Architectures d'ordinateurs (IF - 3ème année)

Session 1: Instruction Set Architecture, assembler (1) [PDF] Session 2: Instruction Set Architecture, assembler (2) [PDF] Session 3: Building a RISC processor from scratch. Instruction level parallelism [PDF] Session 4: Super-scalar processors, multi-threading, multi-core; ARM vs. x86 [PDF] Session 5: Memory hierarchy : cache, virtual memory [PDF] Session 6: Massively parallel processors (GPUs) [PDF] Full content on Moodle (in French)

Cours Robotique et vision par ordinateur OT (Option transversale 5A) "Robotique")

Vision for mobile robotics Visual SLAM with Kinect Object and face detection : deep learning Human-robot interaction : pose estimation Human-robot interaction : gesture recognition Human-robot interaction : face-to-face interaction The slides [PDF]

Cours+TD+TP Lanières SCAN et L - Groupes 72,54 (PC - 2ème année)

Full content on Moodle (in French): Winter Semester (S3), Summer Semester (S4), Database part (S3) Algorithmics and the java programming language: Classes and object oriented programming User interface design: event based programming Dynamic data structures: lists, trees Mini projet (Screenshots of previous miniprojects) Useful links: A cute java applet which visualizes how various data structures work.