Reliable and Fast Recurrent Neural Network Architecture Optimization
This addresses the need for efficient architecture optimization in machine learning, though it appears incremental as it builds on existing evolutionary and evaluation methods.
The paper tackled the problem of optimizing recurrent neural network architectures by introducing RESN, a method that combines an evolutionary algorithm with training-free evaluation, achieving state-of-the-art error performance and reducing computational time by half.
This article introduces Random Error Sampling-based Neuroevolution (RESN), a novel automatic method to optimize recurrent neural network architectures. RESN combines an evolutionary algorithm with a training-free evaluation approach. The results show that RESN achieves state-of-the-art error performance while reducing by half the computational time.