no code implementations • 14 Feb 2022 • Pierre Marza, Corentin Kervadec, Grigory Antipov, Moez Baccouche, Christian Wolf
We also study the impact of two methods to incorporate the information about objects necessary for answering a question, in the reasoning module directly, and earlier in the object selection stage.
no code implementations • NeurIPS 2021 • Corentin Kervadec, Christian Wolf, Grigory Antipov, Moez Baccouche, Madiha Nadri
Methods for Visual Question Anwering (VQA) are notorious for leveraging dataset biases rather than performing reasoning, hindering generalization.
no code implementations • CVPR 2021 • Corentin Kervadec, Theo Jaunet, Grigory Antipov, Moez Baccouche, Romain Vuillemot, Christian Wolf
Since its inception, Visual Question Answering (VQA) is notoriously known as a task, where models are prone to exploit biases in datasets to find shortcuts instead of performing high-level reasoning.
1 code implementation • 2 Apr 2021 • Theo Jaunet, Corentin Kervadec, Romain Vuillemot, Grigory Antipov, Moez Baccouche, Christian Wolf
First, as a result of a collaboration of three fields, machine learning, vision and language reasoning, and data analytics, the work lead to a better understanding of bias exploitation of neural models for VQA, which eventually resulted in an impact on its design and training through the proposition of a method for the transfer of reasoning patterns from an oracle model.
no code implementations • 10 Jun 2020 • Corentin Kervadec, Grigory Antipov, Moez Baccouche, Christian Wolf
Since its appearance, Visual Question Answering (VQA, i. e. answering a question posed over an image), has always been treated as a classification problem over a set of predefined answers.
1 code implementation • CVPR 2021 • Corentin Kervadec, Grigory Antipov, Moez Baccouche, Christian Wolf
Models for Visual Question Answering (VQA) are notorious for their tendency to rely on dataset biases, as the large and unbalanced diversity of questions and concepts involved and tends to prevent models from learning to reason, leading them to perform educated guesses instead.
no code implementations • 6 Dec 2019 • Corentin Kervadec, Grigory Antipov, Moez Baccouche, Christian Wolf
The large adoption of the self-attention (i. e. transformer model) and BERT-like training principles has recently resulted in a number of high performing models on a large panoply of vision-and-language problems (such as Visual Question Answering (VQA), image retrieval, etc.).
no code implementations • 15 Mar 2019 • Juan-Manuel Pérez-Rúa, Valentin Vielzeuf, Stéphane Pateux, Moez Baccouche, Frédéric Jurie
We tackle the problem of finding good architectures for multimodal classification problems.
no code implementations • 1 Aug 2018 • Juan-Manuel Perez-Rua, Moez Baccouche, Stephane Pateux
We demonstrate with experiments on the CIFAR-10 dataset that our method, denominated Efficient progressive neural architecture search (EPNAS), leads to increased search efficiency, while retaining competitiveness of found architectures.
2 code implementations • 7 Feb 2017 • Grigory Antipov, Moez Baccouche, Jean-Luc Dugelay
It has been recently shown that Generative Adversarial Networks (GANs) can produce synthetic images of exceptional visual fidelity.