自動微分のサーベイ論文 “Automatic differentiation in machine learning: a survey” を 森田が PyTorch のコードを読みつつ紹介します。感想などはハッシュタグ #misreading か [email protected] にお寄せください。
https://misreading.chat/wp-content/uploads/2018/05/ep09.mp3
[1502.05767] Automatic differentiation in machine learning: a surveyAutomatic differentiation in PyTorch | OpenReviewChainer: a next-generation open source framework for deep learningImplementation of Control Flow in TensorFlowpytorch/pytorch: Tensors and Dynamic neural networks in Python with strong GPU accelerationRelease alpha-1 release · pytorch/pytorch aka. v0.1.1Yann LeCun on Twitter: “Dude, there were graph-based deep learning frameworks with AD back in 1991 (SN at AT&T) and its open source successor Lush, released in 2002.… https://t.co/sL2wPnaOJY”深層学習 (Goodfellow の教科書日本語訳)Follow-up and Correction
Netflix ResearchHow to Adopt Modern C++17 into Your C++ Code – YouTubeAmazon.com: A Tour of C++Carnegie Mellon Launches Undergraduate Degree in Artificial Intelligence | Carnegie Mellon School of Computer ScienceCurriculum | Carnegie Mellon School of Computer SciencePyTorch の論文は 2017 の Autodiff Workshop で発表されていたようです。10:25 あたりで計算途中の変数が Tensor になっていると言っていますが、Variable の間違いです。