EasyDeepLearn

How do Successive Halving / Hyperband / ASHA speed up hyperparameter search?

hard

Answer

  • Instead of running every config to completion, allocate a small budget to many configs, kill the worst half (or bottom N/eta), double the budget for survivors, repeat.
  • Successive Halving: fixed budget schedule.
  • Hyperband: run multiple 'brackets' with different tradeoffs between exploration (many configs, small budget) and exploitation (few configs, full budget).
  • ASHA (Asynchronous Successive Halving): parallel version, ideal for distributed clusters.
  • Combined with a smart sampler (random + TPE), Hyperband/ASHA are the fastest generic hyperparameter search algorithms.
Check yourself — multiple choice
  • Run every configuration to full training
  • Multi-fidelity: give few resources to many configs, kill the worst, promote survivors, repeat
  • Same as random search
  • Only used for regression

SH/Hyperband/ASHA: multi-fidelity — allocate budget iteratively to the best-performing configs.

#hyperparameter-tuning

Practise Supervised Learning

215 interview questions in this topic.

Related questions