Pascal Vincent

Joint Embedding Self-Supervised Learning (JE-SSL) has seen rapid developments in recent years, due to its promise to effectively leverage la… (see more)rge unlabeled data. The development of JE-SSL methods was driven primarily by the search for ever increasing downstream classification accuracies, using huge computational resources, and typically built upon insights and intuitions inherited from a close parent JE-SSL method. This has led unwittingly to numerous pre-conceived ideas that carried over across methods e.g. that SimCLR requires very large mini batches to yield competitive accuracies; that strong and computationally slow data augmentations are required. In this work, we debunk several such ill-formed a priori ideas in the hope to unleash the full potential of JE-SSL free of unnecessary limitations. In fact, when carefully evaluating performances across different downstream tasks and properly optimizing hyper-parameters of the methods, we most often -- if not always -- see that these widespread misconceptions do not hold. For example we show that it is possible to train SimCLR to learn useful representations, while using a single image patch as negative example, and simple Gaussian noise as the only data augmentation for the positive pair. Along these lines, in the hope to democratize JE-SSL and to allow researchers to easily make more extensive evaluations of their methods, we introduce an optimized PyTorch library for SSL.

2023-03-03

ArXiv (preprint)

Disentanglement of Correlated Factors via Hausdorff Factorized Support

Karsten Roth

Mark Ibrahim

Zeynep Akata

Diane Bouchacourt

A grand goal in deep learning research is to learn representations capable of generalizing across distribution shifts. Disentanglement is on… (see more)e promising direction aimed at aligning a model's representation with the underlying factors generating the data (e.g. color or background). Existing disentanglement methods, however, rely on an often unrealistic assumption: that factors are statistically independent. In reality, factors (like object color and shape) are correlated. To address this limitation, we consider the use of a relaxed disentanglement criterion -- the Hausdorff Factorized Support (HFS) criterion -- that encourages only pairwise factorized \emph{support}, rather than a factorial distribution, by minimizing a Hausdorff distance. This allows for arbitrary distributions of the factors over their support, including correlations between them. We show that the use of HFS consistently facilitates disentanglement and recovery of ground-truth factors across a variety of correlation settings and benchmarks, even under severe training correlations and correlation shifts, with in parts over

2023-02-01

ICLR.cc/2023/Conference (poster)

ImageNet-X: Understanding Model Mistakes with Factor of Variation Annotations

Badr Youbi Idrissi

Diane Bouchacourt

Randall Balestriero

Ivan Evtimov

Caner Hazirbas

Nicolas Ballas

Michal Drozdzal

David Lopez-Paz

Mark Ibrahim

Deep learning vision systems are widely deployed across applications where reliability is critical. However, even today's best models can fa… (see more)il to recognize an object when its pose, lighting, or background varies. While existing benchmarks surface examples challenging for models, they do not explain why such mistakes arise. To address this need, we introduce ImageNet-X—a set of sixteen human annotations of factors such as pose, background, or lighting the entire ImageNet-1k validation set as well as a random subset of 12k training images. Equipped with ImageNet-X, we investigate 2,200 current recognition models and study the types of mistakes as a function of model’s (1) architecture, e.g. transformer vs. convolutional, (2) learning paradigm, e.g. supervised vs. self-supervised, and (3) training procedures, e.g., data augmentation. Regardless of these choices, we find models have consistent failure modes across ImageNet-X categories. We also find that while data augmentation can improve robustness to certain factors, they induce spill-over effects to other factors. For example, color-jitter augmentation improves robustness to color and brightness, but surprisingly hurts robustness to pose. Together, these insights suggest to advance the robustness of modern vision models, future research should focus on collecting additional data and understanding data augmentation schemes. Along with these insights, we release a toolkit based on ImageNet-X to spur further study into the mistakes image recognition systems make.

2023-02-01

ICLR.cc/2023/Conference (notable)

The Hidden Uniform Cluster Prior in Self-Supervised Learning

Mahmoud Assran

Randall Balestriero

Quentin Duval

Florian Bordes

Ishan Misra

Piotr Bojanowski

Michael Rabbat

Nicolas Ballas

A successful paradigm in representation learning is to perform self-supervised pretraining using tasks based on mini-batch statistics (e.g.,… (see more) SimCLR, VICReg, SwAV, MSN). We show that in the formulation of all these methods is an overlooked prior to learn features that enable uniform clustering of the data. While this prior has led to remarkably semantic representations when pretraining on class-balanced data, such as ImageNet, we demonstrate that it can hamper performance when pretraining on class-imbalanced data. By moving away from conventional uniformity priors and instead preferring power-law distributed feature clusters, we show that one can improve the quality of the learned representations on real-world class-imbalanced datasets. To demonstrate this, we develop an extension of the Masked Siamese Networks (MSN) method to support the use of arbitrary features priors.

2023-02-01

ICLR.cc/2023/Conference (poster)

Guillotine Regularization: Why removing layers is needed to improve generalization in Self-Supervised Learning

Florian Bordes

Randall Balestriero

Quentin Garrido

Adrien Bardes

One unexpected technique that emerged in recent years consists in training a Deep Network (DN) with a Self-Supervised Learning (SSL) method,… (see more) and using this network on downstream tasks but with its last few projector layers entirely removed. This trick of throwing away the projector is actually critical for SSL methods to display competitive performances on ImageNet for which more than 30 percentage points can be gained that way. This is a little vexing, as one would hope that the network layer at which invariance is explicitly enforced by the SSL criterion during training (the last projector layer) should be the one to use for best generalization performance downstream. But it seems not to be, and this study sheds some light on why. This trick, which we name Guillotine Regularization (GR), is in fact a generically applicable method that has been used to improve generalization performance in transfer learning scenarios. In this work, we identify the underlying reasons behind its success and show that the optimal layer to use might change significantly depending on the training setup, the data or the downstream task. Lastly, we give some insights on how to reduce the need for a projector in SSL by aligning the pretext SSL task and the downstream task.

2023-01-01

Trans. Mach. Learn. Res. (published)

The Emergence of Argument Structure in Artificial Languages

Tom Bosc

Abstract Computational approaches to the study of language emergence can help us understand how natural languages are shaped by cognitive an… (see more)d sociocultural factors. Previous work focused on tasks where agents refer to a single entity. In contrast, we study how agents predicate, that is, how they express that some relation holds between several entities. We introduce a setup where agents talk about a variable number of entities that can be partially observed by the listener. In the presence of a least-effort pressure, they tend to discuss only entities that are not observed by the listener. Thus we can obtain artificial phrases that denote a single entity, as well as artificial sentences that denote several entities. In natural languages, if we ignore the verb, phrases are usually concatenated, either in a specific order or by adding case markers to form sentences. Our setup allows us to quantify how much this holds in emergent languages using a metric we call concatenability. We also measure transitivity, which quantifies the importance of word order. We demonstrate the usefulness of this new setup and metrics for studying factors that influence argument structure. We compare agents having access to input representations structured into pre-segmented objects with properties, versus unstructured representations. Our results indicate that the awareness of object structure yields a more natural sentence organization.

2022-12-01

Transactions of the Association for Computational Linguistics (published)

High Fidelity Visualization of What Your Self-Supervised Representation Knows About

Florian Bordes

Randall Balestriero

Discovering what is learned by neural networks remains a challenge. In self-supervised learning, classification is the most common task used… (see more) to evaluate how good a representation is. However, relying only on such downstream task can limit our understanding of what information is retained in the representation of a given input. In this work, we showcase the use of a Representation Conditional Diffusion Model (RCDM) to visualize in data space the representations learned by self-supervised models. The use of RCDM is motivated by its ability to generate high-quality samples -- on par with state-of-the-art generative models -- while ensuring that the representations of those samples are faithful i.e. close to the one used for conditioning. By using RCDM to analyze self-supervised models, we are able to clearly show visually that i) SSL (backbone) representation are not invariant to the data augmentations they were trained with -- thus debunking an often restated but mistaken belief; ii) SSL post-projector embeddings appear indeed invariant to these data augmentation, along with many other data symmetries; iii) SSL representations appear more robust to small adversarial perturbation of their inputs than representations trained in a supervised manner; and iv) that SSL-trained representations exhibit an inherent structure that can be explored thanks to RCDM visualization and enables image manipulation.

2022-08-15

TMLR (accepted)

Masked Siamese Networks for Label-Efficient Learning

Mahmoud Assran

Mathilde Caron

Ishan Misra

Piotr Bojanowski

Florian Bordes

Armand Joulin

Michael Rabbat

Nicolas Ballas

We propose Masked Siamese Networks (MSN), a self-supervised learning framework for learning image representations. Our approach matches the … (see more)representation of an image view containing randomly masked patches to the representation of the original unmasked image. This self-supervised pre-training strategy is particularly scalable when applied to Vision Transformers since only the unmasked patches are processed by the network. As a result, MSNs improve the scalability of joint-embedding architectures, while producing representations of a high semantic level that perform competitively on low-shot image classification. For instance, on ImageNet-1K, with only 5,000 annotated images, our base MSN model achieves 72.4% top-1 accuracy, and with 1% of ImageNet-1K labels, we achieve 75.7% top-1 accuracy, setting a new state-of-the-art for self-supervised learning on this benchmark. Our code is publicly available.

2022-04-14

ArXiv (preprint)

Masked Siamese Networks for Label-Efficient Learning

Mahmoud Assran

Mathilde Caron

Ishan Misra

Piotr Bojanowski

Florian Bordes

Armand Joulin

Michael Rabbat

Nicolas Ballas

2022-04-14

ArXiv (preprint)

Enhanced Biomedical Knowledge Discovery From Unstructured Text Using Contextual Embeddings

Iz Beltagy

Kyle Lo

Arman Cohan. 2019

Scib-500

Yoshua Bengio

R´ejean Ducharme

Rishi Bommasani

Kelly Davis

Claire Cardie

Billy Chiu

Sampo Pyysalo

Ivan Vuli´c

Extracting knowledge from large, unstruc-001 tured text corpora presents a challenge. Re-002 cently, authors have utilized unsupervised, 003… (see more) static word embeddings to uncover "latent 004 knowledge" contained within domain-speciﬁc 005 scientiﬁc corpora. Here semantic-similarity 006 measures between representations of concepts, 007 objects or entities were used to predict re-008 lationships, which were later veriﬁed using 009 physical methods. Static language models 010 have recently been surpassed at most down-011 stream tasks by massively pre-trained, contex-012 tual language models like BERT. Some have 013 postulated that contextualized embeddings po-014 tentially yield word representations superior 015 to static ones for knowledge-discovery pur-016 poses. In an effort to address this ques-017 tion, two biomedically-trained BERT models 018 (BioBERT, SciBERT) were used to encode 019 n = 500, 1000 or 5000 sentences containing 020 words of interest extracted from a biomedical 021 corpus (Coronavirus Open Research Dataset). 022 The n representations for the words of inter-023 est were subsequently extracted and then ag-024 gregated to yield static-equivalent word rep-025 resentations. These words belonged to the 026 vocabularies of intrinsic benchmarking tools 027 for the biomedical domain (Bio-SimVerb and 028 Bio-SimLex), which assess quality of word 029 representations using semantic-similarity and 030 relatedness measures. Using intrinsic bench-031 marking tasks, feasibility of using contextual-032 ized word representations for knowledge dis-033 covery tasks can be assessed: Word represen-034 tations that better encode described reality are 035 expected to perform better (i.e. closer to do-036 main experts). As postulated, BERT embed-037 dings outperform static counterparts

Accounting for Variance in Machine Learning Benchmarks

Xavier Bouthillier

Pierre Delaunay

Mirko Bronzi

Assya Trofimov

Brennan Nichyporuk

Justin Szeto

Naz Sepah

Edward Raff

Kanika Madan

Vikram Voleti

Samira Ebrahimi Kahou

Vincent Michalski

Dmitriy Serdyuk

Tal Arbel

Chris Pal

Gael Varoquaux

Strong empirical evidence that one machine-learning algorithm A outperforms another one B ideally calls for multiple trials optimizing the l… (see more)earning pipeline over sources of variation such as data sampling, data augmentation, parameter initialization, and hyperparameters choices. This is prohibitively expensive, and corners are cut to reach conclusions. We model the whole benchmarking process, revealing that variance due to data sampling, parameter initialization and hyperparameter choice impact markedly the results. We analyze the predominant comparison methods used today in the light of this variance. We show a counter-intuitive result that adding more sources of variation to an imperfect estimator approaches better the ideal estimator at a 51 times reduction in compute cost. Building on these results, we study the error rate of detecting improvements, on five different deep-learning tasks/architectures. This study leads us to propose recommendations for performance comparisons.

2021-01-01

MLSys (published)

Cooperative Semi-Supervised Transfer Learning of Machine Reading Comprehension

Oliver Bender

Franz Josef Och

Yoshua Bengio

R´ejean Ducharme