EmProx: Neural Network Performance Estimation For Neural Architecture Search

TL;DR

EmProx introduces a fast and accurate neural network performance estimation method for neural architecture search, utilizing embedding proximity and kNN to significantly reduce training time while maintaining high prediction accuracy.

Contribution

The paper presents EmProx, a novel performance estimation strategy that maps architectures to an embedding space and estimates performance with kNN, offering comparable accuracy to existing methods but with much faster training.

Findings

EmProx achieves similar accuracy to NAO's MLP predictor.

EmProx is nearly nine times faster to train than NAO.

Outperforms other strategies with 5 to 80 times faster estimation times.

Abstract

Common Neural Architecture Search methods generate large amounts of candidate architectures that need training in order to assess their performance and find an optimal architecture. To minimize the search time we use different performance estimation strategies. The effectiveness of such strategies varies in terms of accuracy and fit and query time. This study proposes a new method, EmProx Score (Embedding Proximity Score). Similar to Neural Architecture Optimization (NAO), this method maps candidate architectures to a continuous embedding space using an encoder-decoder framework. The performance of candidates is then estimated using weighted kNN based on the embedding vectors of architectures of which the performance is known. Performance estimations of this method are comparable to the MLP performance predictor used in NAO in terms of accuracy, while being nearly nine times faster to train compared to NAO. Benchmarking against other performance estimation strategies currently used shows similar to better accuracy, while being five up to eighty times faster.