Abstract:Efficient neural architecture search (ENAS) achieves novel efficiency for learning architecture with high-performance via parameter sharing and reinforcement learning (RL). In the phase of architecture search, ENAS employs deep scalable architecture as search space whose training process consumes most of the search cost. Moreover, time-consuming model training is proportional to the depth of deep scalable architecture. Through experiments using ENAS on CIFAR-10, we find that layer reduction of scalable architecture is an effective way to accelerate the search process of ENAS but suffers from a prohibitive performance drop in the phase of architecture estimation. In this article, we propose a broad neural architecture search (BNAS) where we elaborately design broad scalable architecture dubbed broad convolutional neural network (BCNN) to solve the above issue. On the one hand, the proposed broad scalable architecture has fast training speed due to its shallow topology. Moreover, we also adopt RL and parameter sharing used in ENAS as the optimization strategy of BNAS. Hence, the proposed approach can achieve higher search efficiency. On the other hand, the broad scalable architecture extracts multi-scale features and enhancement representations, and feeds them into global average pooling (GAP) layer to yield more reasonable and comprehensive representations. Therefore, the performance of broad scalable architecture can be promised. In particular, we also develop two variants for BNAS that modify the topology of BCNN. In order to verify the effectiveness of BNAS, several experiments are performed and experimental results show that 1) BNAS delivers 0.19 days which is 2.37× less expensive than ENAS who ranks the best in RL-based NAS approaches; 2) compared with small-size (0.5 million parameters) and medium-size (1.1 million parameters) models, the architecture learned by BNAS obtains state-of-the-art performance (3.58% and 3.24% test error) on CIFAR-10; and 3) the learned architecture achieves 25.3% top-1 error on ImageNet just using 3.9 million parameters.

A Study of the Learning Progress in Neural Architecture Search Techniques

Understanding Neural Architecture Search Techniques

Understanding Architectures Learnt by Cell-based Neural Architecture Search

Efficient Neural Architecture Search via Parameter Sharing

EPNAS: Efficient Progressive Neural Architecture Search

BNAS: Efficient Neural Architecture Search Using Broad Scalable Architecture

Improving the sample-efficiency of neural architecture search with reinforcement learning

Neural Architecture Search in Embedding Space

UCB-ENAS based on Reinforcement Learning

Evolving Search Space for Neural Architecture Search

Neural Architecture Search with Progressive Evaluation and Sub-Population Preservation

Advances in neural architecture search

A Hierarchical Evolution of Neural Architecture Search Method Based on State Transition Algorithm

A Survey on Neural Architecture Search

BNAS:An Efficient Neural Architecture Search Approach Using Broad Scalable Architecture

Guided evolutionary Neural Architecture Search with efficient performance estimation

Evaluating the Search Phase of Neural Architecture Search

Continual and Multi-Task Architecture Search

Neural Architecture Search using Progressive Evolution

EENA: Efficient Evolution of Neural Architecture.

AS-NAS: Adaptive Scalable Neural Architecture Search with Reinforced Evolutionary Algorithm for Deep Learning.