no code implementations • 10 Oct 2023 • Ece Ozkan, Xavier Boix
We refer to this approach as multi-domain model and compare its performance to that of specialized models.
1 code implementation • 15 Sep 2023 • Amir Rahimi, Vanessa D'Amario, Moyuru Yamada, Kentaro Takemoto, Tomotake Sasaki, Xavier Boix
We demonstrate that this result is independent of the similarity between the training and testing data and applies to well-known families of neural network architectures for VQA (i. e. monolithic architectures and neural module networks).
1 code implementation • 15 Jun 2023 • Ian Mason, Anirban Sarkar, Tomotake Sasaki, Xavier Boix
In this work we develop a compositional image classification task where, given a few elemental corruptions, models are asked to generalize to compositions of these corruptions.
no code implementations • 17 Mar 2023 • Anirban Sarkar, Matthew Groth, Ian Mason, Tomotake Sasaki, Xavier Boix
Deep Neural Networks (DNNs) often fail in out-of-distribution scenarios.
1 code implementation • 27 Jan 2022 • Moyuru Yamada, Vanessa D'Amario, Kentaro Takemoto, Xavier Boix, Tomotake Sasaki
We reveal that Neural Module Networks (NMNs), i. e., question-specific compositions of modules that tackle a sub-task, achieve better or similar systematic generalization performance than the conventional Transformers, even though NMNs' modules are CNN-based.
1 code implementation • 25 Jan 2022 • Kimberly Villalobos, Vilim Štih, Amineh Ahmadinejad, Shobhita Sundaram, Jamell Dozier, Andrew Francl, Frederico Azevedo, Tomotake Sasaki, Xavier Boix
Only recurrent networks trained with small images learn solutions that generalize well to almost any curve.
1 code implementation • 17 Dec 2021 • Dimitris Bertsimas, Xavier Boix, Kimberly Villalobos Carballo, Dick den Hertog
We introduce a new approach to adversarial training by minimizing an upper bound of the adversarial loss that is based on a holistic expansion of the network instead of separate bounds for each layer.
1 code implementation • 8 Dec 2021 • Shobhita Sundaram, Darius Sinha, Matthew Groth, Tomotake Sasaki, Xavier Boix
Symmetry is omnipresent in nature and perceived by the visual system of many species, as it facilitates detecting ecologically important classes of objects in our environment.
1 code implementation • 30 Oct 2021 • Akira Sakai, Taro Sunagawa, Spandan Madan, Kanata Suzuki, Takashi Katoh, Hiromichi Kobashi, Hanspeter Pfister, Pawan Sinha, Xavier Boix, Tomotake Sasaki
While humans have a remarkable capability of recognizing objects in out-of-distribution (OoD) orientations and illuminations, Deep Neural Networks (DNNs) severely suffer in this case, even when large amounts of training examples are available.
no code implementations • 28 Sep 2021 • Avi Cooper, Xavier Boix, Daniel Harari, Spandan Madan, Hanspeter Pfister, Tomotake Sasaki, Pawan Sinha
The capability of Deep Neural Networks (DNNs) to recognize objects in orientations outside the distribution of the training data is not well understood.
no code implementations • 13 Jul 2021 • Vanessa D'Amario, Sanjana Srivastava, Tomotake Sasaki, Xavier Boix
Datasets often contain input dimensions that are unnecessary to predict the output label, e. g. background in object recognition, which lead to more trainable parameters.
1 code implementation • 30 Jun 2021 • Spandan Madan, Tomotake Sasaki, Hanspeter Pfister, Tzu-Mao Li, Xavier Boix
This result provides evidence supporting theories attributing adversarial examples to the proximity of data to ground-truth class boundaries, and calls into question other theories which do not account for this more stringent definition of adversarial attacks.
1 code implementation • NeurIPS 2021 • Vanessa D'Amario, Tomotake Sasaki, Xavier Boix
Neural Module Networks (NMNs) aim at Visual Question Answering (VQA) via composition of modules that tackle a sub-task.
no code implementations • 1 Jan 2021 • Spandan Madan, Timothy Henry, Jamell Arthur Dozier, Helen Ho, Nishchal Bhandari, Tomotake Sasaki, Fredo Durand, Hanspeter Pfister, Xavier Boix
We find that learning category and viewpoint in separate networks compared to a shared one leads to an increase in selectivity and invariance, as separate networks are not forced to preserve information about both category and viewpoint.
no code implementations • 1 Jan 2021 • Vanessa D'Amario, Sanjana Srivastava, Tomotake Sasaki, Xavier Boix
In this paper, we investigate the impact of unnecessary input dimensions on one of the central issues of machine learning: the number of training examples needed to achieve high generalization performance, which we refer to as the network's data efficiency.
2 code implementations • 15 Jul 2020 • Spandan Madan, Timothy Henry, Jamell Dozier, Helen Ho, Nishchal Bhandari, Tomotake Sasaki, Frédo Durand, Hanspeter Pfister, Xavier Boix
In this paper, we investigate when and how such OOD generalization may be possible by evaluating CNNs trained to classify both object category and 3D viewpoint on OOD combinations, and identifying the neural mechanisms that facilitate such OOD generalization.
no code implementations • 30 Jun 2020 • Hojin Jang, Syed Suleman Abbas Zaidi, Xavier Boix, Neeraj Prasad, Sharon Gilad-Gutnick, Shlomit Ben-Ami, Pawan Sinha
Our results with state-of-the-art DCNNs indicate that invariant neural representations do not always drive robustness to transformations, as networks show robustness for categories seen transformed during training even in the absence of invariant neural representations.
1 code implementation • 10 Dec 2019 • Stephen Casper, Xavier Boix, Vanessa D'Amario, Ling Guo, Martin Schrimpf, Kasper Vinken, Gabriel Kreiman
We identify two distinct types of "frivolous" units that proliferate when the network's width is increased: prunable units which can be dropped out of the network without significant change to the output and redundant units whose activities can be expressed as a linear combination of others.
no code implementations • 25 Sep 2019 • Kimberly M Villalobos, Vilim Stih, Amineh Ahmadinejad, Jamell Dozier, Andrew Francl, Frederico Azevedo, Tomotake Sasaki, Xavier Boix
At the heart of image segmentation lies the problem of determining whether a pixel is inside or outside a region, which we denote as the "insideness" problem.
no code implementations • ICLR 2019 • Sanjana Srivastava, Guy Ben-Yosef, Xavier Boix
Ullman et al. 2016 show that a slight modification of the location and size of the visible region of the minimal image produces a sharp drop in human recognition accuracy.
no code implementations • 29 Jun 2018 • Tomaso Poggio, Qianli Liao, Brando Miranda, Andrzej Banburski, Xavier Boix, Jack Hidary
Here we prove a similar result for nonlinear multilayer DNNs near zero minima of the empirical loss.
no code implementations • 30 Dec 2017 • Tomaso Poggio, Kenji Kawaguchi, Qianli Liao, Brando Miranda, Lorenzo Rosasco, Xavier Boix, Jack Hidary, Hrushikesh Mhaskar
In this note, we show that the dynamics associated to gradient descent minimization of nonlinear networks is topologically equivalent, near the asymptotically stable minima of the empirical error, to linear gradient system in a quadratic potential with a degenerate (for square loss) or almost degenerate (for logistic or crossentropy loss) Hessian.
no code implementations • 14 Nov 2016 • Ece Ozkan, Gemma Roig, Orcun Goksel, Xavier Boix
We show that the algorithm to extract diverse M -solutions from a Conditional Random Field (called divMbest [1]) takes exactly the form of a Herding procedure [2], i. e. a deterministic dynamical system that produces a sequence of hypotheses that respect a set of observed moment constraints.
no code implementations • CVPR 2016 • Anna Volokitin, Michael Gygli, Xavier Boix
Many computational models of visual attention use image features and machine learning techniques to predict eye fixation locations as saliency maps.
no code implementations • ICCV 2015 • Xun Huang, Chengyao Shen, Xavier Boix, Qi Zhao
Saliency in Context (SALICON) is an ongoing effort that aims at understanding and predicting visual attention.
no code implementations • 19 Nov 2015 • Yan Luo, Xavier Boix, Gemma Roig, Tomaso Poggio, Qi Zhao
To see this, first, we report results in ImageNet that lead to a revision of the hypothesis that adversarial perturbations are a consequence of CNNs acting as a linear classifier: CNNs act locally linearly to changes in the image regions with objects recognized by the CNN, and in other regions the CNN may act non-linearly.
no code implementations • NeurIPS 2014 • Xavier Boix, Gemma Roig, Salomon Diether, Luc V. Gool
Within this processing pipeline, the common trend is to learn the feature coding templates, often referred as codebook entries, filters, or over-complete basis.
no code implementations • 29 Aug 2014 • Xavier Boix, Gemma Roig, Luc van Gool
In a series of papers by Dai and colleagues [1, 2], a feature map (or kernel) was introduced for semi- and unsupervised learning.
1 code implementation • 16 Sep 2013 • Michael Van den Bergh, Xavier Boix, Gemma Roig, Luc van Gool
We define a robust and fast to evaluate energy function, based on enforcing color similarity between the bound- aries and the superpixel color histogram.
no code implementations • 19 Jul 2013 • Gemma Roig, Xavier Boix, Luc van Gool
SVMs suffer from various drawbacks in terms of selecting the right kernel, which depends on the image descriptors, as well as computational and memory efficiency.
no code implementations • CVPR 2013 • Xavier Boix, Michael Gygli, Gemma Roig, Luc van Gool
We demonstrate the capabilities of our formulation for both keypoint matching and image classification.