Yoshua Bengio

TL;DR: In this paper , the authors proposed an improved version of Mixup, theoretically justified to deliver better generalization performance than the vanilla mixup, which implicitly regularizes infinitely many directional derivatives of all orders.

TL;DR: To imitate the target behavior more closely by incorporating the lookback phenomenon in GNNs, two methods are proposed: (a) target smoothing for the standard cross-entropy loss function, and (b) adding a Parent-as-Target (PAT) Lookback regularizer term.

TL;DR: In this paper, an attention mechanism dynamically selects which modules can be adapted to the current task, and the parameters of the selected modules are allowed to change quickly as the learner is confronted with variations in what it experiences.

TL;DR: This work establishes that a physical system can perform statistical learning without gradient computations, via an Agnostic Equilibrium Propagation procedure that combines energy minimization, homeostatic control, and nudging towards the correct response.

TL;DR: In this article, the authors propose a particular training framework in which the pieces of knowledge an agent needs, as well as its reward function are stationary and can be re-used across tasks, and find that meta-learning the modular aspects of the proposed system greatly help in achieving faster learning, in experiments with a reinforcement learning setup involving navigation in a partially observed grid world with image-level input.