Our teams from Warwick DCS have won both the 1st and 2nd place at the Zero Cost NAS Competition held in conjunction with the AutoML'22 conference.

Over the recent year, Neural Architecture Search (NAS) has attracted a lot of attention. While being able to automate the discovery of better performing neural architectures than hand-crafted ones, it comes at a great price, requiring thousands of GPU hours to perform the search. The Zero Cost NAS competition challenges the participants to design efficient proxies for NAS, using negligible computational resources to evaluate neural architectures.

In collaboration with the AutoCAML team at Samsung AI Cambridge (led by Dr. Hongkai Wen), our research students, Lichuan Xiang and Youyang Sha, proposed new zero-cost NAS metrics that exploit the compressibility of neural networks. Our metrics are extremely efficient to run (reducing search cost from weeks/days to minutes), and achieves impressive results across multiple search spaces and datasets. In the competition, our teams won both the 1st and 2nd places (using different scoring functions), and the performance gap with the 3rd winning team is almost 2x. Checkout our poster here.

Congratulations to Harita Dellaporta for receiving the Best Paper Award at the premier conference in Artificial Intelligence and Statistics (AISTATS) 2022 for her paper on Robust Bayesian Inference for Simulator-based Models via the MMD Posterior Bootstrap.

We are delighted to report that Dr Long Tran-Thanh has received a AIJ Prominent Paper Award for his first-authored paper, Efficient crowdsourcing of unknown experts using bounded multi-armed bandits, published in 2014 at Artificial Intelligence (AIJ), a premier journal in the field of artificial intelligence. The AIJ Prominent Paper Award recognises outstanding papers published in the journal in the last seven years that are exceptional in their significance and impact.

The paper developed the first comprehensive framework for the rigorous and principled mathematical analysis of task allocation algorithms in crowdsourcing systems. In addition, the paper proposed bounded bandits, a new sequential decision making model to solve task allocation problems with resource constraints. The work has had a significant impact on subsequent work carried out in both industry and academia. The award will be presented at IJCAI 2021, a top tier international conference in artificial intelligence.