Regularizing Towards Permutation Invariance in Recurrent Models
This addresses the challenge of applying temporal architectures to permutation-invariant problems, offering a novel regularization approach that is incremental compared to existing invariant-by-design methods.
The paper tackles the problem of making recurrent neural networks (RNNs) permutation invariant for tasks where output should not depend on input order, by introducing a stochastic regularization method that results in compact models and outperforms existing approaches on synthetic and real-world datasets.
In many machine learning problems the output should not depend on the order of the input. Such "permutation invariant" functions have been studied extensively recently. Here we argue that temporal architectures such as RNNs are highly relevant for such problems, despite the inherent dependence of RNNs on order. We show that RNNs can be regularized towards permutation invariance, and that this can result in compact models, as compared to non-recurrent architectures. We implement this idea via a novel form of stochastic regularization. Existing solutions mostly suggest restricting the learning problem to hypothesis classes which are permutation invariant by design. Our approach of enforcing permutation invariance via regularization gives rise to models which are \textit{semi permutation invariant} (e.g. invariant to some permutations and not to others). We show that our method outperforms other permutation invariant approaches on synthetic and real world datasets.