no code implementations • 25 Apr 2024 • Mazda Moayeri, Michael Rabbat, Mark Ibrahim, Diane Bouchacourt
We propose a method to encode and account for diversity within a class using inferred attributes, still in the zero-shot setting without retraining.
no code implementations • 15 Nov 2023 • Cian Eastwood, Julius von Kügelgen, Linus Ericsson, Diane Bouchacourt, Pascal Vincent, Bernhard Schölkopf, Mark Ibrahim
Self-supervised representation learning often uses data augmentations to induce some invariance to "style" attributes of the data.
no code implementations • 28 Sep 2023 • Mohammad Pezeshki, Diane Bouchacourt, Mark Ibrahim, Nicolas Ballas, Pascal Vincent, David Lopez-Paz
Successful out-of-distribution generalization requires environment annotations.
1 code implementation • NeurIPS 2023 • Florian Bordes, Shashank Shekhar, Mark Ibrahim, Diane Bouchacourt, Pascal Vincent, Ari S. Morcos
Synthetic image datasets offer unmatched advantages for designing and evaluating deep neural networks: they make it possible to (i) render as many data samples as needed, (ii) precisely control each scene and yield granular ground truth labels (and captions), (iii) precisely control distribution shifts between training and testing to isolate variables of interest for sound experimentation.
no code implementations • 24 Jul 2023 • Megan Richards, Polina Kirichenko, Diane Bouchacourt, Mark Ibrahim
Second, we study model generalization across geographies by measuring the disparities in performance across regions, a more fine-grained measure of real world generalization.
1 code implementation • 11 Apr 2023 • Laura Gustafson, Megan Richards, Melissa Hall, Caner Hazirbas, Diane Bouchacourt, Mark Ibrahim
As an example, we show that mitigating a model's vulnerability to texture can improve performance on the lower income level.
no code implementations • 3 Nov 2022 • Badr Youbi Idrissi, Diane Bouchacourt, Randall Balestriero, Ivan Evtimov, Caner Hazirbas, Nicolas Ballas, Pascal Vincent, Michal Drozdzal, David Lopez-Paz, Mark Ibrahim
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.
no code implementations • 24 Oct 2022 • Mark Ibrahim, Quentin Garrido, Ari Morcos, Diane Bouchacourt
We study not only how robust recent state-of-the-art models are, but also the extent to which models can generalize variation in factors when they're present during training.
no code implementations • 24 Oct 2022 • Mark Ibrahim, Diane Bouchacourt, Ari Morcos
Our approach applies the formalism of Lie groups to capture continuous transformations to improve models' robustness to distributional shifts.
1 code implementation • 13 Oct 2022 • Karsten Roth, Mark Ibrahim, Zeynep Akata, Pascal Vincent, Diane Bouchacourt
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 $+60\%$ in relative improvement over existing disentanglement methods.
no code implementations • 20 Jul 2022 • David Lopez-Paz, Diane Bouchacourt, Levent Sagun, Nicolas Usunier
By highlighting connections to the literature in domain generalization, we propose to measure fairness as the ability of the system to generalize under multiple stress tests -- distributions of examples with social relevance.
1 code implementation • NeurIPS 2021 • Diane Bouchacourt, Mark Ibrahim, Ari S. Morcos
While prior work has focused on synthetic data, we attempt here to characterize the factors of variation in a real dataset, ImageNet, and study the invariance of both standard residual networks and the recently proposed vision transformer with respect to changes in these factors.
1 code implementation • NeurIPS 2021 • Diane Bouchacourt, Mark Ibrahim, Ari S. Morcos
While prior work has focused on synthetic data, we attempt here to characterize the factors of variation in a real dataset, ImageNet, and study the invariance of both standard residual networks and the recently proposed vision transformer with respect to changes in these factors.
1 code implementation • 10 Feb 2021 • Diane Bouchacourt, Mark Ibrahim, Stéphane Deny
A core challenge in Machine Learning is to learn to disentangle natural factors of variation in data (e. g. object shape vs. pose).
no code implementations • 1 Jan 2021 • Diane Bouchacourt, Mark Ibrahim, Stephane Deny
A core challenge in Machine Learning is to disentangle natural factors of variation in data (e. g. object shape vs pose).
1 code implementation • 29 Sep 2020 • Christina Heinze-Deml, Diane Bouchacourt
Contrarily to humans who have the ability to recombine familiar expressions to create novel ones, modern neural networks struggle to do so.
1 code implementation • ICLR 2020 • Jonathan Gordon, David Lopez-Paz, Marco Baroni, Diane Bouchacourt
Humans understand novel sentences by composing meanings and roles of core language components.
1 code implementation • ACL 2020 • Rahma Chaabouni, Eugene Kharitonov, Diane Bouchacourt, Emmanuel Dupoux, Marco Baroni
Third, while compositionality is not necessary for generalization, it provides an advantage in terms of language transmission: The more compositional a language is, the more easily it will be picked up by new learners, even when the latter differ in architecture from the original agents.
4 code implementations • NeurIPS 2020 • Laura Ruis, Jacob Andreas, Marco Baroni, Diane Bouchacourt, Brenden M. Lake
In this paper, we introduce a new benchmark, gSCAN, for evaluating compositional generalization in situated language understanding.
no code implementations • 5 Nov 2019 • Roberto Dessì, Diane Bouchacourt, Davide Crepaldi, Marco Baroni
Research in multi-agent cooperation has shown that artificial agents are able to learn to play a simple referential game while developing a shared lexicon.
no code implementations • 25 Sep 2019 • Diane Bouchacourt, Ludovic Denoyer
Therefore, we propose a new self-interpretable model that performs output prediction and simultaneously provides an explanation in terms of the presence of particular concepts in the input.
no code implementations • 14 Aug 2019 • Mathijs Mul, Diane Bouchacourt, Elia Bruni
A typical setup to achieve this is with a scripted teacher which guides a virtual agent using language instructions.
no code implementations • IJCNLP 2019 • Eugene Kharitonov, Rahma Chaabouni, Diane Bouchacourt, Marco Baroni
There is renewed interest in simulating language emergence among deep neural agents that communicate to jointly solve a task, spurred by the practical aim to develop language-enabled interactive AIs, as well as by theoretical questions about the evolution of human language.
1 code implementation • ICML 2020 • Eugene Kharitonov, Rahma Chaabouni, Diane Bouchacourt, Marco Baroni
There is growing interest in studying the languages that emerge when neural agents are jointly trained to solve tasks requiring communication through a discrete channel.
1 code implementation • ACL 2019 • Diane Bouchacourt, Marco Baroni
Recent research studies communication emergence in communities of deep network agents assigned a joint task, hoping to gain insights on human language evolution.
no code implementations • 28 May 2019 • Diane Bouchacourt, Ludovic Denoyer
Therefore, we propose a new self-interpretable model that performs output prediction and simultaneously provides an explanation in terms of the presence of particular concepts in the input.
no code implementations • EMNLP 2018 • Diane Bouchacourt, Marco Baroni
There is growing interest in the language developed by agents interacting in emergent-communication settings.
2 code implementations • 24 May 2017 • Diane Bouchacourt, Ryota Tomioka, Sebastian Nowozin
We would like to learn a representation of the data which decomposes an observation into factors of variation which we can independently control.
no code implementations • NeurIPS 2016 • Diane Bouchacourt, Pawan K. Mudigonda, Sebastian Nowozin
We present a new type of probabilistic model which we call DISsimilarity COefficient Networks (DISCO Nets).
no code implementations • 8 Jun 2016 • Diane Bouchacourt, M. Pawan Kumar, Sebastian Nowozin
We present a new type of probabilistic model which we call DISsimilarity COefficient Networks (DISCO Nets).
no code implementations • ICCV 2015 • Diane Bouchacourt, Sebastian Nowozin, M. Pawan Kumar
To this end, we propose a novel prediction criterion that includes as special cases all previous prediction criteria that have been used in the literature.