Timezone: »

On orthogonality and learning RNNs with long term dependencies
Eugene Vorontsov · Chiheb Trabelsi · Christopher Pal · Samuel Kadoury

Tue Aug 08 01:30 AM -- 05:00 AM (PDT) @ Gallery #39

It is well known that it is challenging to train deep neural networks and recurrent neural networks for tasks that exhibit long term dependencies. The vanishing or exploding gradient problem is a well known issue associated with these challenges. One approach to addressing vanishing and exploding gradients is to use either soft or hard constraints on weight matrices so as to encourage or enforce orthogonality. Orthogonal matrices preserve gradient norm during backpropagation and may therefore be a desirable property. This paper explores issues with optimization convergence, speed and gradient stability when encouraging or enforcing orthogonality. To perform this analysis, we propose a weight matrix factorization and parameterization strategy through which we can bound matrix norms and therein control the degree of expansivity induced during backpropagation. We find that hard constraints on orthogonality can negatively affect the speed of convergence and model performance.

Author Information

Eugene Vorontsov (MILA)
Chiheb Trabelsi (Ecole Polytechnique de Montreal)
Christopher Pal (École Polytechnique de Montréal)
Samuel Kadoury (Ecole Polytechnique de Montreal)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors