ICLR 2020 將於2020年4 月 26 日於埃塞俄比亞首都亞的斯亞貝巴舉行。
在最終提交的 2594 篇論文中,有 687 篇被接收,接收率爲 26.5%;和之前一樣,所有被接收的論文都將進行 poster 展示;23% 的接收論文將進行口頭講述報告,其中108 篇論文爲 spotlight,演講時間 4 分鐘,48 篇論文將進行 10 分鐘以上的 talk。
對ICLR 2020接收論文中與元學習(Meta-learning)及小樣本學習(few-shot learning)相關方向的論文進行總結。若有遺漏或其他錯誤還請指正。
元學習(Meta-learning)
- Learning To Balance: Bayesian Meta-learning For Imbalanced And Out-of-distribution Tasks (Talk)
- Meta-learning With Warped Gradient Descent (Talk)
- Meta-learning Without Memorization (Spotlight)
- Meta-learning Acquisition Functions For Transfer Learning In Bayesian Optimization (Spotlight)
- Automated Relational Meta-learning (Poster)
- Empirical Bayes Transductive Meta-learning With Synthetic Gradients (Poster)
- Meta-learning Curiosity Algorithms (Poster)
- Towards Fast Adaptation Of Neural Architectures With Meta Learning (Poster)
- Differentially Private Meta-learning (Poster)
- Meta-learning Deep Energy-based Memory Models (Poster)
- Varibad: A Very Good Method For Bayes-adaptive Deep Rl Via Meta-learning (Poster)
- ES-MAML: Simple Hessian-Free Meta Learning (Poster)
- Watch, Try, Learn: Meta-learning From Demonstrations And Rewards (Poster)
Meta Reinforcement Learning
- Meta-Q-Learning (Talk)
- Improving Generalization In Meta Reinforcement Learning Using Neural Objectives (Spotlight)
- Meta Reinforcement Learning With Autonomous Inference Of Subtask Dependencies (Poster)
小樣本學習(Few-shot Learning)
- A Theoretical Analysis Of The Number Of Shots In Few-shot Learning (Poster)
- A Baseline For Few-shot Image Classification (Poster)
- Metapix: Few-shot Video Retargeting (Poster)
- Cross-domain Few-shot Classification Via Learned Feature-wise Transformation (Spotlight)
- Few-shot Learning On Graphs Via Super-classes Based On Graph Spectral Measures (Poster)
- Few-shot Text Classification With Distributional Signatures (Poster)
- Meta-dataset: A Dataset Of Datasets For Learning To Learn From Few Examples (Poster)