NAS (Neural Architecture Search) is the automated construction of Deep Learning architectures targeted towards given tasks. This competition aims to test NAS algorithms on new and unknown datasets to evaluate whether NAS is able to
adapt to problems outside the datasets used for standard benchmark experimentation.
Competition Timeline
- Phase 1 (Development): Participants use the provided starter kit to develop and test their method locally.
- Phase 2 (Validation): Submitted code is run as a “smoke test”. This phase runs from July 1st to July 31st (making it a great good fit for e.g. summer Master’s projects).
- Phase 3 (Final Evaluation): Validated methods are evaluated on secret datasets to determine the final rankings.
To take part or for more information, please visit the competition website: http://www.nascompetition.com/
Starter Kit:
https://github.com/Towers-D/NAS-Comp-Starter-Kit
Join the Discord:
https://discord.gg/EUUfhdjp
Contact:
nas-competition-contact@newcstle.ac.uk