KNAS Green Neural Architecture Search

Posted on July 4, 2022   1 minute read ∼ Filed in  : 

The paper tries to evaluate architecture without training based on the Gram Matrix of gradients of a mini-batch.

provement

Overall, the paper tries to find how the loss is related to the gradient of a mini-batch.

image-20220707144931469

Finally, the paper se the mean of GM - MGM - to evaluate an architecture.

image-20220707145116956

In practice, since the weight is too long, so the paper use sampling to only use a party of weights.

image-20220707145215773

Algorithm

image-20220707145234438

Evaluation

image-20220707145500009

The result shows that MGM has a good coefficient with real performance.

image-20220707145548608

This result shows that the

  1. KNAS is faster than search-based and gradient-based evaluation algorithms, and also has a good performance than them.
  2. KNAS is slower than training-free based algorithm but has better Acc on ImageNet than those 2.




END OF POST




Tags Cloud


Categories Cloud




It's the niceties that make the difference fate gives us the hand, and we play the cards.