Home /
Research
Showing 319 - 324 / 904
An end-to-end data-driven optimisation framework for constrained trajectories
FlorentDewez, BenjaminGuedj, ArthurTalpaert....
Published date-11/24/2020
Many real-world problems require to optimise trajectories under constraints. Classical approaches are based on optimal control methods but require an exact knowledge of the underlying dynamics, which could be challenging …
Augmented Lagrangian Adversarial Attacks
JérômeRony, EricGranger, MarcoPedersoli....
Published date-11/24/2020
AdversarialAttack
Adversarial attack algorithms are dominated by penalty methods, which are slow in practice, or more efficient distance-customized methods, which are heavily tailored to the properties of the considered distance. We …
Dissecting Image Crops
BasileVanHoorick, CarlVondrick....
Published date-11/24/2020
DataAugmentation, ImageForensics, RepresentationLearning, Self-SupervisedLearning
The elementary operation of cropping underpins nearly every computer vision system, ranging from data augmentation and translation invariance to computational photography and representation learning. This paper investigates the subtle traces …
Play Fair: Frame Attributions in Video Models
WillPrice, DimaDamen....
Published date-11/24/2020
ActionRecognition, RelationalReasoning
In this paper, we introduce an attribution method for explaining action recognition models. Such models fuse information from multiple frames within a video, through score aggregation or relational reasoning. We …
Energy-Based Models for Continual Learning
ShuangLi, YilunDu, GidoM.vandeVen....
Published date-11/24/2020
ContinualLearning
We motivate Energy-Based Models (EBMs) as a promising model class for continual learning problems. Instead of tackling continual learning via the use of external memory, growing models, or regularization, EBMs …
Message Passing Networks for Molecules with Tetrahedral Chirality
LagnajitPattanaik, OctavianE.Ganea, IanColey....
Published date-11/24/2020
DrugDiscovery
Molecules with identical graph connectivity can exhibit different physical and biological properties if they exhibit stereochemistry-a spatial structural characteristic. However, modern neural architectures designed for learning structure-property relationships from molecular …