AWP

[riow1983]

Wu, Xia, and Wang (2020), Adversarial Weight Perturbation Helps Robust Generalization https://arxiv.org/abs/2004.05884

Authors' implementation: https://github.com/csdongxian/AWP

[riow1983]

[What is threat model?]

An adversarial threat model defines a set of perturbations that may be made to an image in order to produce an adversarial example. Common threat models include L2 and L∞ threat models, which constrain adversarial examples to be close to the original image in L2 or L∞ distances.

https://arxiv.org/pdf/2006.12655.pdf

[riow1983]

ATまとめ

https://www.slideshare.net/ssuserc45ddf/adversarial-trainingpptx

[riow1983]

Fast AWP (Kaggle notebook) https://www.kaggle.com/code/junkoda/fast-awp

Video (Chinese) https://www.bilibili.com/s/video/BV1pt4y1Y7uU

[riow1983]

PACベイズの基礎についてまとめる https://qiita.com/student-i/items/341f1fd720d4b6b26d7b

[riow1983]

Kaggle notebook: https://www.kaggle.com/code/wht1996/feedback-nn-train/notebook

[riow1983]

社内発表完了.

[riow1983]

---

[SGDとシグナム関数] The Geometry of Sign Gradient Descent https://arxiv.org/pdf/2002.08056.pdf

---

[確率とバウンドについて] ヘフディングの不等式(Hoeffding's inequality)と諸々の確率の評価の不等式 https://ludu-vorton.hatenablog.com/entry/2019/06/06/073000

CS229 Supplemental Lecture notes Hoeffding’s inequality http://cs229.stanford.edu/extra-notes/hoeffding.pdf

---

[VC次元とDLの汎化誤差見積もりについて] 過学習に理論的に迫る https://qiita.com/kuroitu/items/e695907d9a2d28cc0cf9

(Not) Bounding the True Error https://papers.nips.cc/paper/2001/file/98c7242894844ecd6ec94af67ac8247d-Paper.pdf

Exploring Generalization in Deep Learning https://papers.nips.cc/paper/2017/file/10ce03a1ed01077e3e289f3e53c72813-Paper.pdf

Computing Nonvacuous Generalization Bounds for Deep (Stochastic) Neural Networks with Many More Parameters than Training Data https://arxiv.org/pdf/1703.11008.pdf

---

[PAC Bayes] An Introduction to PAC-Bayes https://www.youtube.com/watch?v=t5GBuBD0ibc

Some PAC-Bayesian Theorems (McAllester's theorem) https://link.springer.com/content/pdf/10.1023/A:1007618624809.pdf