2. OUTLINE
3
1,Active Learning
2,論文紹介
・Learning Loss for Active Learning
(2019,CVPR)
3,Bayesian Neural Network
4,論文紹介
・BatchBALD: Efficient and Diverse Batch Acquisition
for Deep Bayesian Active Learning (2019, NeurIPS)
・Bayesian Generative Active Deep Learning (2019, ICML)
まとめ
3. OUTLINE
4
1,Active Learning
2,論文紹介
・Learning Loss for Active Learning
(2019,CVPR)
3,Bayesian Neural Network
4,論文紹介
・BatchBALD: Efficient and Diverse Batch Acquisition
for Deep Bayesian Active Learning (2019, NeurIPS)
・Bayesian Generative Active Deep Learning (2019, ICML)
まとめ
14. OUTLINE
15
1,Active Learning
2,論文紹介
・Learning Loss for Active Learning
(2019,CVPR)
3,Bayesian Neural Network
4,論文紹介
・BatchBALD: Efficient and Diverse Batch Acquisition
for Deep Bayesian Active Learning (2019, NeurIPS)
・Bayesian Generative Active Deep Learning (2019, ICML)
まとめ
15. 発表論文
16
・Learning Loss for Active Learning
(Yoo et al.,2019,CVPR)
・BatchBALD: Efficient and Diverse Batch Acquisition
for Deep Bayesian Active Learning
(Kirsch et al., 2019, NeurIPS)
・Bayesian Generative Active Deep Learning
(Tran et al., 2019, ICML)
26. OUTLINE
27
1,Active Learning
2,論文紹介
・Learning Loss for Active Learning
(2019,CVPR)
3,Bayesian Neural Network
4,論文紹介
・BatchBALD: Efficient and Diverse Batch Acquisition
for Deep Bayesian Active Learning (2019, NeurIPS)
・Bayesian Generative Active Deep Learning (2019, ICML)
まとめ
38. OUTLINE
39
1,Active Learning
2,論文紹介
・Learning Loss for Active Learning
(2019,CVPR)
3,Bayesian Neural Network
4,論文紹介
・BatchBALD: Efficient and Diverse Batch Acquisition
for Deep Bayesian Active Learning (2019, NeurIPS)
・Bayesian Generative Active Deep Learning (2019, ICML)
まとめ
39. 発表論文
BatchBALD:
Efficient and Diverse Batch Acquisition
for Deep Bayesian Active Learning
(2019, NeurIPS)
Andreas Kirsch, Joost van Amersfoort, Yarin Gal
50. 発表論文
51
・Learning Loss for Active Learning
(Yoo et al.,2019,CVPR)
・BatchBALD: Efficient and Diverse Batch Acquisition
for Deep Bayesian Active Learning
(Kirsch et al., 2019, NeurIPS)
・Bayesian Generative Active Deep Learning
(Tran et al., 2019, ICML)
51. Bayesian Generative Active Deep Learning
(2019, ICML)
Toan Tran, Thanh-Toan Do, Ian Reid, Gustavo Carneiro
61. 発表論文(再掲)
62
・Learning Loss for Active Learning (Yoo et al.,2019,CVPR)
・Batch BALD: Efficient and Diverse Batch Acquisition
for Deep Bayesian Active Learning (Kirsch et al., 2019, NeurIPS)
・Bayesian Generative Active Deep Learning (Tran et al., 2019, ICML)
タスクに非依存な形で低コストに計算できる損失関数を定義し,
クエリを抽出する手法を提案した論文
バッチ内のデータの相関を考慮していないBALDの課題を解消する
Batch BALDを提案した論文
Bayesian ALと,Data Augmentationを組み合わせることで,
モデルの学習に効率的な画像のAugmentationする手法を提案した論文
63. 参考
64
[1] Yarin Gal, Zoubin Ghahramani,
Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning, 2016
( https://arxiv.org/abs/ 1506.02142)
[2]須山敦志 著,ベイズ深層学習,講談社,2020
[3]須山敦志 著,杉山将 監修,ベイズ推論による機械学習入門,講談社,2018
[4] Yarin Gal, Riashat Islam, Zoubin Ghahramani,
Deep Bayesian Active Learning with Image Data, 2017(ICML)
(https://dl.acm.org/doi/pdf/10.5555/3305381.3305504)
[5] Raymond W. Yeung , A New Outlook on Shannon’s Information Measures(IEEE) ,1991
(https://pdfs.semanticscholar.org/a37e/ab85f532cdc027260777815d78f164eb93aa.pdf)
[6] O. Sener and S. Savarese. Active learning for convolutional neural networks: A core-set approach.
In International Conference on Learning Representations, 2018.
(https://arxiv.org/pdf/1708.00489.pdf)
[7] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun, Identity Mappings in Deep Residual Networks,
2016 (https://arxiv.org/pdf/1603.05027.pdf)
[8] Toan Tran, A Bayesian Data Augmentation Approach for Learning Deep Models,2017
(https://arxiv.org/pdf/1710.10564.pdf)