機械学習のためのソフトウェア工学に関する課題やチャレンジを議論する SE4ML meet-up https://mlxse.connpass.com/event/66905/ に申し込んでみた。 #MLxSE
# 募集が「機械学習の人」「ソフトウェア工学の人」の枠になってて、自分は「機械学習の人」と「ソフトウェア工学の人」のどちらでもないような気がして、どちらに申し込むが悩んだけれど (^^;
この辺りの議論は、NIPS2014併設の Software Engineering for Machine Learning (SW4ML) https://sites.google.com/site/software4ml/ 以降、
* NIPS2015併設の Workshop on Machine Learning Systems (LearningSys) http://learningsys.org
* ICML2016併設の ML Systems Workshop https://sites.google.com/site/mlsys2016/
* NIPS2016併設の Reliable Machine Learning in the Wild https://sites.google.com/site/wildml2016/
* ICML2017併設の Reliable Machine Learning in the Wild - ICML 2017 Workshop https://sites.google.com/site/wildml2017icml/
* SOSP2017併設で10月に開催予定の Workshop on AI Systems (AISys) at Symposium on Operating Systems Principles (SOSP) http://learningsys.org/sosp17/
と、継続的に議論されてきていて、盛り上がってきている話だと思う。
自分も機械学習モデルのテストや検証については興味があり、これまでここでも例えば以下のような研究を紹介してきたので、この辺りの話とか議論できれば良いな。
L. Pulina and A. Tacchella, "An Abstraction-Refinement approach to verification of artificial neural networks," in Computer Aided Verification (CAV 2010), https://www.researchgate.net/publication/221403485
紹介: https://plus.google.com/+MasahiroSakai/posts/WJAWeft778y
L. Pulina and A. Tacchella, "Challenging SMT solvers to verify neural networks," AI Commun., vol. 25, no. 2, pp. 117-135, Apr. 2012 http://dx.doi.org/10.3233/AIC-2012-0525 https://pdfs.semanticscholar.org/c94e/f735b9c587c9d0fbc7cf890a664aa6951982.pdf
紹介: https://plus.google.com/+MasahiroSakai/posts/QoLiMdiUQGe
K. Scheibler, L. Winterer, R. Wimmer, and B. Becker. “Towards verification of artificial neural networks”. In 18th Workshop on Methoden und Beschreibungssprachen zur Modellierung und Verifikation von Schaltungen und Systemen” (MBMV), 2015. http://www.avacs.org/fileadmin/Publikationen/Open/scheibler.mbmv2015.pdf
紹介: https://plus.google.com/+MasahiroSakai/posts/JWinbw9NQQW
G. Katz, C. Barrett, D. Dill, K. Julian, and M. Kochenderfer, "Reluplex: An efficient SMT solver for verifying deep neural networks," CAV 2017, http://arxiv.org/abs/1702.01135
紹介: https://plus.google.com/+MasahiroSakai/posts/913ZviJ4iVR
R. Ehlers, "Formal verification of Piece-Wise linear Feed-Forward neural networks," http://arxiv.org/abs/1705.01320
紹介: https://plus.google.com/+MasahiroSakai/posts/iif74B92Vu7
K. Pei, Y. Cao, J. Yang, and S. Jana, "DeepXplore: Automated whitebox testing of deep learning systems," Jun. 2017. http://arxiv.org/abs/1705.06640
紹介: https://plus.google.com/+MasahiroSakai/posts/63uHxCWNkn2
SE4ML meet-up
