GP-NAS-ensemble: a model for NAS Performance Prediction
This work addresses the time-consuming evaluation bottleneck in Neural Architecture Search, offering a domain-specific incremental improvement.
The paper tackles the problem of predicting neural architecture performance without full training by proposing GP-NAS-ensemble, a framework that improves upon GP-NAS with ensemble learning, achieving second place in the CVPR2022 lightweight NAS challenge.
It is of great significance to estimate the performance of a given model architecture without training in the application of Neural Architecture Search (NAS) as it may take a lot of time to evaluate the performance of an architecture. In this paper, a novel NAS framework called GP-NAS-ensemble is proposed to predict the performance of a neural network architecture with a small training dataset. We make several improvements on the GP-NAS model to make it share the advantage of ensemble learning methods. Our method ranks second in the CVPR2022 second lightweight NAS challenge performance prediction track.