Page Not Found
Page not found. Your pixels are in another canvas.
Sitemap
A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.
Pages
Hello There !
About me
Page not in menu
This is a page not in th emain menu
Posts
Understanding Few-Shot Multi-Task Representation Learning Theory
Published:
Multi-Task Representation Learning (MTR) is a popular paradigm to learn shared representations from multiple related tasks. It has demonstrated its efficiency for solving different problems, ranging from machine translation for natural language processing to object detection in computer vision. On the other hand, Few-Shot Learning is a recent problem that seeks to mimic the human capability to quickly learn how to solve a target task with little supervision. For this topic, researchers have turned to meta-learning that learns to learn a new task by training a model on a lot of small tasks. As meta-learning still suffers from a lack of theoretical understanding for its success in few-shot tasks, an intuitively appealing approach would be to bridge the gap between it and multi-task learning to better understand the former using the results established for the latter. In this post, we dive into a recent ICLR 2021 paper by S. Du, W. Hu, S. Kakade, J. Lee and Q. Lei, that demonstrated novel learning bounds for multi-task learning in the few-shot setting and go beyond it by establishing the connections that allow to better understand the inner workings of meta-learning algorithms as well.
Optimal Transport as a Defense Against Adversarial Attacks
Published:
We present Sinkhorn Adversarial Training (SAT), a robust adversarial training method based on the latest theory of optimal transport. We also propose a new metric, the Area Under Accuracy Curve (AUAC), to quantify more precisely the robustness of a model to adversarial attacks over a wide range of perturbation sizes.
Vulnerability of Person Re-Identification Models to Metric Adversarial Attacks
Published:
We investigate different possible attacks on metric learning models depending on the number and type of guides available. Two particularly effective attacks stand out. To defend against these attacks, we adapt the adversarial training protocol for metric learning. Let us guide you !
portfolio
Portfolio item number 1
Short description of portfolio item number 1
Portfolio item number 2
Short description of portfolio item number 2
publications
Vulnerability of Person Re-Identification Models to Metric Adversarial Attacks.
Published in CVPRW, 2020
Recommended citation: "Vulnerability of Person Re-Identification Models to Metric Adversarial Attacks." Quentin Bouniot, Romaric Audigier, Angélique Loesch (2020). IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops.
Optimal Transport as a Defense Against Adversarial Attacks
Published in ICPR, 2021
Recommended citation: "Optimal Transport as a Defense Against Adversarial Attacks." Quentin Bouniot, Romaric Audigier, Angélique Loesch (2021). IEEE International Conference on Pattern Recognition (ICPR) 2020.
Understanding Few-Shot Multi-Task Representation Learning Theory
Published in ICLR Blog Track, 2022
Recommended citation: "Understanding Few-Shot Multi-Task Representation Learning Theory" Quentin Bouniot & Ievgen Redko, ICLR Blog Track, 2022
Improving Few-Shot Learning through Multi-task Representation Learning Theory
Published in ECCV, 2022
Recommended citation: "Improving Few-Shot Learning through Multi-task Representation Learning Theory" Quentin Bouniot, Ievgen Redko, Romaric Audigier, Angélique Loesch, Amaury Habrard. Computer Vision–ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part XX. Cham: Springer Nature Switzerland, 2022.
Towards Few-Annotation Learning for Object Detection: Are Transformer-Based Models More Efficient?
Published in WACV, 2023
Recommended citation: "Towards Few-Annotation Learning for Object Detection: Are Transformer-Based Models More Efficient?" Quentin Bouniot, Angélique Loesch, Romaric Audigier, Amaury Habrard; Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2023, pp. 75-84
Proposal-Contrastive Pretraining for Object Detection from Fewer Data
Published in ICLR, 2023
Recommended citation: "Proposal-Contrastive Pretraining for Object Detection from Fewer Data" Quentin Bouniot,Romaric Audigier, Angelique Loesch, Amaury Habrard; The Eleventh International Conference on Learning Representations, 2023
Towards Few-Annotation Learning in Computer Vision: Application to Image Classification and Object Detection tasks
Published in PhD Thesis, 2023
Recommended citation: "Towards Few-Annotation Learning in Computer Vision: Application to Image Classification and Object Detection tasks" Quentin Bouniot (2023). PhD Thesis.
talks
Vulnerability of Person Re-Identification Models to Metric Adversarial Attacks
Published:
We investigate different possible attacks on metric learning models depending on the number and type of guides available. Two particularly effective attacks stand out. To defend against these attacks, we adapt the adversarial training protocol for metric learning. Let us guide you !
Vulnerability of Person Re-Identification Models to Metric Adversarial Attacks
Published:
We investigate different possible attacks on metric learning models depending on the number and type of guides available. Two particularly effective attacks stand out. To defend against these attacks, we adapt the adversarial training protocol for metric learning. Let us guide you !
Putting Theory to Work: From Learning Bounds to Meta-Learning Algorithms
Published:
In this work, we want to apply the latest insights from meta-learning theory in practice to ensure a good meta-learning.
Optimal Transport as a Defense Against Adversarial Attacks
Published:
We present Sinkhorn Adversarial Training (SAT), a robust adversarial training method based on the latest theory of optimal transport. We also propose a new metric, the Area Under Accuracy Curve (AUAC), to quantify more precisely the robustness of a model to adversarial attacks over a wide range of perturbation sizes.
Vers une meilleure compréhension des méthodes de méta-apprentissage à travers la théorie de l’apprentissage de représentations multi-tâches
Published:
Dans nos travaux nous avons cherché à faire le lien entre le meta-learning et l’apprentissage de représentation multi-tâche, qui possède une importante littérature théorique ainsi que des bornes d’apprentissage solides. Et en analysant les bornes les plus récentes d’apprentissage de représentation multi-tâches et leurs hypothèses, nous avons mis en évidence des critères qui permettent un méta-apprentissage plus efficace.
Improving Few-Shot Learning through Multi-task Representation Learning Theory
Published:
We consider the framework of multi-task representation (MTR) learning where the goal is to use source tasks to learn a representation that reduces the sample complexity of solving a target task. We start by reviewing recent advances in MTR theory and show that they can provide novel insights for popular meta-learning algorithms when analyzed within this framework. In particular, we highlight a fundamental difference between gradient-based and metric-based algorithms and put forward a theoretical analysis to explain it. Finally, we use the derived insights to improve the generalization capacity of meta-learning methods via a new spectral-based regularization term and confirm its efficiency through experimental studies on classic few-shot classification benchmarks. To the best of our knowledge, this is the first contribution that puts the most recent learning bounds of MTR theory into practice for the task of few-shot classification
On Few-Annotation Learning and Non-Linearity in Deep Neural Networks
Published:
Learning something new in real life does not necessarily mean going through a lot of examples in order to capture the essence of it. Humans are able to build upon prior experience and have the ability to adapt, allowing to combine previous observations with only little evidence for fast learning. This is particularly the case for recognition tasks, for which we are often capable of differentiating between two distinct objects after having seen only a few examples of them. In this talk, I will develop three different contributions for Machine Learning with limited labels, and more specifically for Computer Vision tasks, addressing theoretical, algorithmic and experimental aspects. In a first contribution, we are interested in bridging the gap between theory and practice for popular Meta-Learning algorithms used in Few-Shot Classification. We make connections to Multi-Task Representation Learning, which benefits from solid theoretical foundations, to verify the best conditions for a more efficient meta-learning. Then, to leverage unlabeled data when training object detectors based on the Transformer architecture, we propose an unsupervised pretraining approach that improves contrastive learning for object detectors through the introduction of the localization information. Finally, we present the first theoretically sound tool to track non-linearity propagation in deep neural networks, with a specific focus on computer vision applications. Our proposed affinity score allows us to gain insights into the inner workings of a wide range of different architectures and learning paradigms. We present extensive experimental results that highlight the practical utility of the proposed affinity score and its potential for long-reaching applications.
Understanding Deep Neural Networks Through the Lens of their Non-Linearity
Published:
The remarkable success of deep neural networks (DNN) is often attributed to their high expressive power and their ability to approximate functions of arbitrary complexity. Indeed, DNNs are highly non-linear models, and activation functions introduced into them are largely responsible for this. While many works studied the expressive power of DNNs through the lens of their approximation capabilities, quantifying the non-linearity of DNNs or of individual activation functions remains an open problem. In this work, we propose the first theoretically sound solution to track non-linearity propagation in deep neural networks with a specific focus on computer vision applications. Our proposed affinity score allows us to gain insights into the inner workings of a wide range of different architectures and learning paradigms. We provide extensive experimental results that highlight the practical utility of the proposed affinity score and its potential for long-reaching applications.
teaching
Algorithms and complexity
Teaching Assistant, CentraleSupélec, Université Paris-Saclay, 1900
First year Computer Science course for the main engineering track at CentraleSupélec
Recent Advances in Responsible AI
Course Lecturer, Télécom Paris, Institut Polytechnique de Paris, 1900
This course is part of the Master 2 Data Science program from IP Paris.
tips
Bash
Published:
Useful commands for bash.
Conda
Published:
Manage Python environments with conda.
Pytorch
Published:
Basic notions of Pytorch and useful functions to manipulate tensors.
Computation time
Published:
A trick to improve computation time when working with lists.
f-string
Published:
Clean and efficient string formatting in Python >3.6
Pathlib
Published:
Easy path handling in Python >3.4
Tqdm
Published:
Beautiful progress bars for loops in Python
Einsum
Published:
Einstein Summation in Numpy of Pytorch
Kubernetes
Published:
Basic notions for kubernetes
Git
Published:
Basic notions of Git. Branching, Merging and Stashing.
Regex
Published:
Basic Notions of Regex
Vim
Published:
Basic notions of Vim and useful shortcuts.
PDF files
Published:
Manipulate PDF document and images.