Grunnleggende konsepter
訓練フリーNASのロバスト性とパフォーマンス向上を目指すRoBoTアルゴリズムが提案された。
Statistikk
訓練フリーNAS: 重要性は高く、計算コスト削減可能(Mellor et al., 2021)
Precision @ T値: 上位ランキング予測性能を評価(Duan et al., 2021)
BOHB: 大規模なハイパーパラメータ最適化(Falkner et al., 2018)
Sitater
"Neural architecture search (NAS) has become a key component of AutoML and a standard tool to automate the design of deep neural networks."
"To address these challenges, we propose the robustifying and boosting training-free NAS (RoBoT) algorithm."
"Our findings suggest that RoBoT is an appropriate choice for ensuring the robustness of training-free metrics while having the potential to boost performance."