しましまのICML2015まとめ

Neil Lawrence @lawrennd

That's a big screen. Jorge Nocedal's ideas filling the stage at the Deep Learning *Workshop* #icml2015 pic.twitter.com/zfDCHxoKac

拡大

Gael Varoquaux @GaelVaroquaux.bsky.social @GaelVaroquaux

Notes from reproducible machine learning discussion at #icml2015 MLOSS workshop gist.github.com/GaelVaroquaux/… #Reproducible #Research

™ @tmaehara

Ax = b の圧縮センシングでAが正定値しか対応してない話について「これ一般化できない？」と聞いたら「A'Ax = A'b とすれば常にこうできるから必要ないのでは」とコメントされた．

Salim Jouili @jouilis

ReNet an alternative to convolutional neural network #icml2015 #machinelearning pic.twitter.com/mwDFXJORTa

拡大

Andrew B. Collier @datawookie

Gaël Varoquaux "Most lines of code written by programmers in academia never reach an audience." #mloss15 #ICML2015

ML @ REDDIT @mxlearn

ICML2015 Drama bit.ly/1Hid9HQ

Sebastien Treguer @ST4Good

Understanding Neural Nets throught deep Visualisation yosinski.com/deepvis #ICML2015 #DeepLearning pic.twitter.com/rSoSNUEjvB

拡大

Sebastien Treguer @ST4Good

Object detection with deep CNN & Bayesian Optimization http:// bit.ly/fgs-obj #ICML2015 #DeepLearning pic.twitter.com/NugsEUvjit

拡大

Mathblogging.org All @MathbloggingAll

#ICML2015 codes: DeepVis, Deep Visualization Toolbox: Understanding Neural Networks Through Deep Visualization dlvr.it/BTj64Z

しましま @shima__shima

明日のプレゼンにジョークを二つ仕込んだけど受けるかなぁ？今まで受けたのは kamishima.net/archive/2009-p… のP6の4行目のネタだけなんだよな… これ，そんなに面白いつもりはなかったんだけどなぁ… なかなか国際会議でジョークは難しいよなぁ…

しましま @shima__shima

7月11日 (土) ICML2015 6日目 ワークショップ 2日目

しましま @shima__shima

Future Directions of Fairness-Aware Data Mining slideshare.net/shima__shima/f… My presentation slide at #icml2015 workshop fatml.org

しましま @shima__shima

公正性のワークショップ 20〜30 といったところか．これで，この分野の研究者の8割ぐらいカバーだな

Balázs Kégl @balazskegl

MIchele Sebag on algorithm recommendation as collaborative filtering. #AutoML #icml2015 pic.twitter.com/n2mrznPIG4

拡大

Matthias Rolf @MatthiasRolf1

Learning of Action Selection for Robot-Assisted Therapy. Senft, Baxter, @TonyBelpaeme at MLIS ws #ICML2015 @DREAM_FP7 pic.twitter.com/JOaYI3b4PQ

拡大

Matthias Rolf @MatthiasRolf1

"Teaching iCub to recognize objects using deep Convolutional Neural Networks" at MLIS ws #ICML2015 @IITalk pic.twitter.com/Bo823jPprZ

拡大

Alison B. Lowndes ✿ @AlisonBLowndes

@google #DeepMind's Gorilla mashing #deeplearning with reinforcement #disbelief parallelization @icml15 #icml2015 pic.twitter.com/xnEdII9Lz6

拡大

Man Group Quant Tech @ManQuantTech

David Silver talks about Gorila, a massively parallel system for deep reinforcement learning at #icml2015 pic.twitter.com/oi5moMtgEX

拡大

Salim Jouili @jouilis

#Gorila: distributed architecture for deep reinforcement learning from #googledeepmind #machinelearning #icml2015 pic.twitter.com/j7XlbjXwBj

拡大

Levity @Al_levity

#ICML2015 DLW: Demo of Mike Tyson on steroids. Google Deepmind. pic.twitter.com/PRUQTtgBQB

拡大

Man Group Quant Tech @ManQuantTech

The Google DeepMind massively parallel GoRiLA architecture at #icml2015 pic.twitter.com/A4X1BYpac2

拡大

Salim Jouili @jouilis

A neural conversational model #icml2015 #machinelearning #nlp pic.twitter.com/WixDQgjpnF

拡大

しましま @shima__shima

Future Directions of Fairness-Aware Data Mining slideshare.net/shima__shima/f… updated my slides #icml2015 workshop fatml.org

Ioannis Katakis @iokat

Big cities, big broblems, big data. @EPratsini at #mud15 #icml2015 pic.twitter.com/zHN6ONeZUR

拡大

しましま @shima__shima

プレゼンに仕込んだ二つのジョークのうち kamishima.net/archive/2015-w… P17の1行目 多様体ネタはすべったが，P22の論文リジェクトネタはちょっとウケた