Yoshua Bengio

TL;DR: This work takes a principled approach towards building robust ensembles and develops an algorithm for learning an ensemble with maximum margin, which not only outperforms existing ensembling techniques, but also large models trained in an end-to-end fashion.

TL;DR: Experiments are performed on 10 English vowels showing a recognition rate higher than 95% for new speakers, suggesting that MLNs suitably fed by the data computed by an ear model have good generalization capabilities over new speakers and new sounds.

TL;DR: An algorithm for the global optimization of this globally optimized hybrid system for phone recognition achieved a recognition accuracy of 86% on an 8 class recognition problem (7 plosives and one class corresponding to all other phonemes).

TL;DR: This paper designs a novel architecture that predicts the spatially continuous solution of a PDE given a spatial position query by augmenting coordinate-based architectures with Graph Neural Networks (GNN), and enables zero-shot generalization to new non-uniform meshes and long-term predictions up to 250 frames ahead that are physically consistent.

TL;DR: This work introduces Continuous-Time Meta-Learning (COMLN), a meta-learning algorithm where adaptation follows the dynamics of a gradient vector field, and devise an efficient algorithm based on forward mode differentiation, whose memory requirements do not scale with the length of the learning trajectory, thus allowing longer adaptation in constant memory.