Neural Architecture Search on NAS-Bench-101 1.0 (test)
0.9432Test AccuracyOptimum
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Optimum2024.03 | 0.9432 | — | — | |
| DiNAS2024.03 | 0.9427 | — | 150 | |
| WeakNAS2024.03 | 0.9418 | — | 200 | |
| AG-Net2024.03 | 0.9418 | — | 192 | |
| arch2vec-RL#Queries=400, Encoding=Unsupervised, Search Method=REINFORCE2020.06 | 0.941 | — | — | |
| Arch2vecSearch Algorithm=RL2024.03 | 0.941 | — | 400 | |
| BANANAS2024.03 | 0.9409 | — | 192 | |
| BANANAS#Queries=500, Encoding=Supervised, Search Method=Bayesian Optimization2020.06 | 0.9408 | — | — | |
| arch2vec-BO#Queries=400, Encoding=Unsupervised, Search Method=Bayesian Optimization2020.06 | 0.9405 | — | — | |
| Arch2vecSearch Algorithm=BO2024.03 | 0.9405 | — | 400 | |
| Local Search2024.03 | 0.9397 | — | 192 | |
| Bayesian Optimisation2024.03 | 0.9396 | — | 192 | |
| Regularised Evolution2024.03 | 0.9389 | — | 192 | |
| BO (supervised baseline)#Queries=400, Encoding=Supervised, Search Method=Bayesian Optimization2020.06 | 0.9379 | — | — | |
| NAO#Queries=1000, Encoding=Supervised, Search Method=Gradient Decent2020.06 | 0.9374 | — | — | |
| RL (supervised baseline)#Queries=400, Encoding=Supervised, Search Method=REINFORCE2020.06 | 0.9374 | — | — | |
| BO#Queries=1000, Encoding=Discrete, Search Method=Bayesian Optimization2020.06 | 0.9372 | — | — | |
| RE#Queries=1000, Encoding=Discrete, Search Method=Evolution2020.06 | 0.9372 | — | — | |
| Random Search2024.03 | 0.9361 | — | 192 | |
| RL#Queries=1000, Encoding=Discrete, Search Method=REINFORCE2020.06 | 0.9358 | — | — | |
| Random Search#Queries=1000, Encoding=Discrete, Search Method=Random2020.06 | 0.9354 | — | — | |
| NAO2024.03 | 0.9349 | — | 192 | |
| AlphaXMechanism=Monte Carlo tree search, Evaluations=150, Trials=2002019.10 | — | 0.0623 | — | |
| BANANASMechanism=BO (path encoding), Evaluations=150, Trials=2002019.10 | — | 0.0592 | — | |
| BO w. GPMechanism=BO (Gaussian process), Evaluations=150, Trials=2002019.10 | — | 0.0627 | — | |
| BOHAMIANNMechanism=BO (Bayesian NN), Evaluations=150, Trials=2002019.10 | — | 0.0601 | — | |
| BOHBMechanism=BO (successive halving), Evaluations=150, Trials=2002019.10 | — | 0.0636 | — | |
| BONASMechanism=BO (GCN), Evaluations=150, Trials=2002019.10 | — | 0.0595 | — | |
| DNGOMechanism=BO (neural networks), Evaluations=150, Trials=2002019.10 | — | 0.0609 | — | |
| GCN Pred.Mechanism=GCN, Evaluations=150, Trials=2002019.10 | — | 0.0633 | — | |
| Local searchMechanism=Local search, Evaluations=150, Trials=2002019.10 | — | 0.0593 | — | |
| NASBOTMechanism=BO (Gaussian process), Evaluations=150, Trials=2002019.10 | — | 0.0625 | — | |
| Random searchMechanism=Random search, Evaluations=150, Trials=2002019.10 | — | 0.0634 | — | |
| Reg. EvolutionMechanism=Evolution, Evaluations=150, Trials=2002019.10 | — | 0.0611 | — | |
| REINFORCEMechanism=Reinforcement learning, Evaluations=150, Trials=2002019.10 | — | 0.0644 | — | |
| TPEMechanism=BO (Parzen windows), Evaluations=150, Trials=2002019.10 | — | 0.0642 | — |