Yoshua Bengio

TL;DR: In this article, a continuation method is proposed to learn an easier (possibly convex) objective function and let it evolve during training until it eventually becomes the original, difficult to optimize objective function.

TL;DR: This special issue highlights a number of algorithmic approaches that are fundamental to data analysis, both in formulating and solving problems that relate to Big Data.

TL;DR: This paper considers several tasks - Mujoco based control tasks and Atari games - and proposes an auxiliary cost function to ensure consistency between what the agent observes and what it imagines, and helps to train powerful policies and better dynamics models.

TL;DR: This article propose an attentive reader and writer (STRAW) model for character-level translation, which computes alignments between the source and target sequences not only for a single time-step but for the next k time-steps as well by constructing a matrix of proposed future alignments and a commitment vector that governs whether to follow or recompute the plan.

TL;DR: In this paper, the authors propose a methodology to quickly predict expected tactical descriptions using machine learning and operations research, which can be used to predict expected descriptions of tactical objectives and objectives.