GitHub - jivoi/awesome-ml-for-cybersecurity: Machine Learning for Cyber SecurityYou signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session. Dismiss alert
Optimization for Deep Learning Highlights in 2017
Optimization for Deep Learning Highlights in 2017 Different gradient descent optimization algorithms have been proposed in recent years but Adam is still most commonly used. This post discusses the most exciting highlights and most promising recent approaches that may shape the way we will optimize our models in the future. This post discusses the most exciting highlights and most promising direct
Automated Analysis of Cybercriminal Markets
We have developed an automated, top-down approach for analyzing underground forums. Using natural language processing and machine learning, our approach can automatically generate high-level information about underground forums with high accuracy. Read the paper About This is a collaborative study among researchers from the International Computer Science Institute, UC Berkeley, UC San Diego, and N
Detecting Malicious Requests with Keras & Tensorflow
Security is a concern for any public facing web application. Good development practices can assist with defending against attempts from users looking to expose data or bring down an app. However, sometimes not all attack vectors are handled and new exploits are bound to be discovered. This is where security software can assist with monitoring and preventing unforeseen attacks. So what if you could
åˆæˆå¤‰é‡ã¨ã‚¢ãƒ³ã‚µãƒ³ãƒ–ル：回帰森ã¨åŠ 法モデルã®è¦ç‚¹
機械å¦ç¿’ã«ãŠã‘る「木ã€ã‚„「森ã€ã®ãƒ¢ãƒ‡ãƒ«ã®æ¦‚è¦ [招待講演] åˆæˆå¤‰é‡ã¨ã‚¢ãƒ³ã‚µãƒ³ãƒ–ル：回帰森ã¨åŠ 法モデルã®è¦ç‚¹ ○瀧å·ä¸€å¦ï¼ˆåŒ—大） ä¿¡å·å‡¦ç†ç ”究会（SIP）https://goo.gl/PxAbbK 2017å¹´ 6月19æ—¥(月) - 2017å¹´ 6月20æ—¥(ç«) 回路ã¨ã‚·ã‚¹ãƒ†ãƒ ç ”ç©¶ä¼šï¼ˆCAS）/ VLSIè¨è¨ˆæŠ€è¡“ç ”ç©¶ä¼šï¼ˆVLD）/ システム数ç†ã¨å¿œç”¨ç ”究会（MSS） 備考) 機械å¦ç¿’ç ”ç©¶è€…ãƒ»ç€§å·ä¸€å¦ã•ã‚“ [北大人図鑑 No.5] https://youtu.be/XNz3D26wy0oRead less
Swarm
SWARM serverless hosting incentivised peer-to-peer storage and content distribution Swarm is a distributed storage platform and content distribution service, a native base layer service of the ethereum web3 stack. The primary objective of Swarm is to provide a decentralized and redundant store for dapp code and data as well as block chain and state data. Swarm is also set out to provide various ba
メルカリã®ä»Šå¹´1å¹´é–“ã®æ©Ÿæ¢°å¦ç¿’ã®å–り組ã¿ã¨ã“ã‚Œã‹ã‚‰
ã“ã‚Œã¯Mercari Advent Calendar 2017 ã®2日目ã®è¨˜äº‹ã§ã™ã€‚ 昨日㯠@stanaka ã®åˆ†æ•£ãƒ•ã‚¡ã‚¤ãƒ«ã‚·ã‚¹ãƒ†ãƒ ã¯ãƒ–ãƒãƒƒã‚¯ãƒã‚§ãƒ¼ãƒ³ã®å¤¢ã‚’見る㋠ã§ã—ãŸã€‚ 今回ã¯@Hmj_kd ãŒã€ãƒ¡ãƒ«ã‚«ãƒªã®æ©Ÿæ¢°å¦ç¿’ã®å–り組ã¿ã‚„機械å¦ç¿’エンジニアã®ä»Šå¹´è¡Œã£ã¦ããŸæ´»å‹•ã®ã„ãã¤ã‹ã‚’ã”紹介ã—ãŸã„ã¨æ€ã„ã¾ã™ã€‚ 以下ã€ä¸€éƒ¨ã«ã¦æ©Ÿæ¢°å¦ç¿’ã‚’MLã¨ç•¥ã—ã¾ã™ã€‚ ã“ã®1å¹´é–“ã§æ©Ÿæ¢°å¦ç¿’ã§å–り組んã ã“㨠ç§ãŒå…¥ç¤¾ã—ãŸã®ã¯2017/01ã§ã€ç¤¾å†…ã§ã¯äºŒäººç›®ã®æ©Ÿæ¢°å¦ç¿’エンジニアã§ã—ãŸã€‚ ãã®å½“時㯠speakerdeck.com ã«ã‚るよã†ã«ã€ãƒ—ãƒãƒ€ã‚¯ãƒˆã«MLã«é–¢ã™ã‚‹ã‚‚ã®ã¯ã‚ã‚Šã¾ã›ã‚“ã§ã—ãŸã€‚ ç¾åœ¨ã¯ã€ãƒãƒ¼ãƒ 全体ã§ãƒžãƒãƒ¼ã‚¸ãƒ£ã‚‚å«ã‚ã¦ç´„10人ã»ã©ã®çµ„ç¹”ã«ãªã‚Šã€æ´»ç™ºã«å®Ÿé¨“や開発ãŒè¡Œã‚ã‚Œã¦ã„ã¾ã™ã€‚ ã„ãã¤ã‹ã‚’列挙ã—ã¾ã™ã¨ - 商å“出å“時ã®ä¾¡æ ¼æŽ¨å®šã¨ã‚µã‚¸ã‚§ã‚¹ãƒˆ - 一部カテゴリã«ã¤ã„ã¦å•†å“ã‚¿ã‚°ã®æŽ¨å®š - 商å“出å“時ã®ã‚«ãƒ†ã‚´ãƒª
分散ファイルシステムã¯ãƒ–ãƒãƒƒã‚¯ãƒã‚§ãƒ¼ãƒ³ã®å¤¢ã‚’見る㋠| メルカリエンジニアリング
今年ã‹ã‚‰ãƒ¡ãƒ«ã‚«ãƒªã§ã‚‚Mercari Advent Calendar 2017ã¨ç§°ã—ã¦Advent Calendarを始ã‚ã‚‹ã“ã¨ã¨ãªã‚Šã¾ã—ãŸã€‚ åˆæ—¥ã¯ id:stanaka / @stanaka ãŒãƒãƒ³ãƒ‰ãƒ³ã‚ˆã‚ŠãŠå±Šã‘ã—ã¾ã™ã€‚ 分散ファイルシステムã¨ã„ã†è¨€è‘‰ã‚’èžãã¨ã€ãƒˆãƒ©ã‚¦ãƒžã‚’刺激ã•ã‚Œã€ã†ã£ã¨é を抱ãˆã‚‹äººã‚‚多ã„ã‹ã¨æ€ã„ã¾ã™ã€‚ç§ã‚‚ãã®ä¸€äººã§ã€ä»¥å‰ã«PBクラスã¾ã§ã¯ã„ã‹ãšã¨ã‚‚æ•°TBã®HDDを数百å°ä¸¦ã¹ãŸã‚·ã‚¹ãƒ†ãƒ ã®ãŠå®ˆã‚Šã‚’ã—ãŸã“ã¨ãŒã‚ã‚Šã€æ—¥ã€…壊れ続ã‘ã‚‹HDDã«è² è·ã«æ‚²é³´ã‚’上ã’るメタデータDBãªã©ãƒã‚¿ã«ã¯å›°ã‚‰ãªã„状æ³ã§ã—ãŸã€‚ãã†ã„ã†æ™‚ã«AWS S3を触るã¨ã€ã€Œã‚ã‚ã€ã“ã‚Œã¯å¤©å›½ã ..ã€ã¨ã‚‚ã¯ã‚„éŽåŽ»ã«ã¯æˆ»ã‚Œãªã„æ€ã„ã‚’ã—ãŸã‚‚ã®ã§ã™ã€‚ 最近ã§ã¯åˆ†æ•£ãƒ•ã‚¡ã‚¤ãƒ«ã‚·ã‚¹ãƒ†ãƒ ã‚’é‹ç”¨ã—ã¦ã„ã‚‹ã¨ã“ã‚ã‚‚ã‚ã£ãり減ã£ã¦ã„ã¦*1ã‚‚ã†éŽåŽ»ã®åˆ†é‡Žã‹ãªã€ã¨æ€ã£ã¦ã„ãŸã®ã§ã™ãŒã€ã“ã“æ•°å¹´ã§ã€Œãƒ–ãƒãƒƒã‚¯ãƒã‚§ãƒ¼ãƒ³ x 分散ファイルシステムã€ã¨ã„ã†
Preferred Networksã€æ·±å±¤å¦ç¿’ã®å¦ç¿’速度ã«ãŠã„ã¦ä¸–ç•Œæœ€é€Ÿã‚’å®Ÿç¾ - æ ªå¼ä¼šç¤¾Preferred Networks
大è¦æ¨¡ãªä¸¦åˆ—コンピュータを活用ã—ã€åˆ†æ•£å¦ç¿’パッケージChainerMNã§ImageNetã®å¦ç¿’ã‚’15分ã§å®Œäº† æ ªå¼ä¼šç¤¾Preferred Networks（本社：æ±äº¬éƒ½åƒä»£ç”°åŒºã€ä»£è¡¨å–ç· å½¹ç¤¾é•·ï¼šè¥¿å·å¾¹ã€ãƒ—リファードãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã‚¹ã€ä»¥ä¸‹ã€PFN）ã¯ã€å¤§è¦æ¨¡ãªä¸¦åˆ—コンピュータ「MN-1※1ã€ã‚’活用ã—ã€æ·±å±¤å¦ç¿’（ディープラーニング）ã®å¦ç¿’速度ã«ãŠã„ã¦ä¸–界最速を実ç¾ã—ã¾ã—ãŸã€‚ 深層å¦ç¿’モデルã®ç²¾åº¦ã‚’å‘上ã•ã›ã‚‹ãŸã‚ã€å¦ç¿’データã®ã‚µã‚¤ã‚ºã‚„モデルã®ãƒ‘ラメータ数ãŒå¢—åŠ ã—ã€ãã‚Œã«ã¨ã‚‚ãªã£ã¦è¨ˆç®—時間も増大ã—ã¦ã„ã¾ã™ã€‚1回ã®å¦ç¿’ã«æ•°é€±é–“ã‹ã‹ã‚‹ã“ã¨ã‚‚稀ã§ã¯ã‚ã‚Šã¾ã›ã‚“。複数ã®ã‚³ãƒ³ãƒ”ュータを連æºã•ã›ã¦å¦ç¿’を高速化ã™ã‚‹ã“ã¨ã¯ã€æ–°ãŸãªã‚¢ã‚¤ãƒ‡ã‚£ã‚¢ã®è©¦è¡ŒéŒ¯èª¤ã‚„検証ã«è¦ã™ã‚‹æ™‚間を圧縮ã—ã€ç´ æ—©ãç ”ç©¶æˆæžœã‚’ã‚ã’ã¦ã„ããŸã‚ã«éžå¸¸ã«é‡è¦ã§ã™ã€‚ 一方ã§ã€è¤‡æ•°ã®ã‚³ãƒ³ãƒ”ュータを使ã£ãŸä¸¦åˆ—分散å¦ç¿’ã«ãŠã„ã¦ã¯ã€é€šå¸¸ã€GPU数を増やã™ã»ã©ãƒãƒƒãƒã‚µ
Preferred Networksã€æ°‘é–“ä¼æ¥ã®è¨ˆç®—環境ã¨ã—㦠国内最大級ã®ãƒ—ライベート・スーパーコンピュータを9月ã‹ã‚‰ç¨¼åƒ - æ ªå¼ä¼šç¤¾Preferred Networks
NTT Comグループã®ãƒžãƒ«ãƒãƒŽãƒ¼ãƒ‰åž‹GPUプラットフォームを採用 æ ªå¼ä¼šç¤¾Preferred Networks(本社：æ±äº¬éƒ½åƒä»£ç”°åŒºã€ä»£è¡¨å–ç· å½¹ç¤¾é•· æœ€é«˜çµŒå–¶è²¬ä»»è€…ï¼šè¥¿å· å¾¹ã€ä»¥ä¸‹ï¼šPFN)ã¯ã€è‡ªå‹•é‹è»¢æŠ€è¡“ã‚„ãŒã‚“診æ–ã‚’ã¯ã˜ã‚ã¨ã—ãŸæ·±å±¤å¦ç¿’(ディープラーニング)ãªã©ã®ç ”究開発用プライベート・スーパーコンピュータをã€2017å¹´9月より稼åƒã—ã¾ã—ãŸã€‚ã“ã®ã‚¹ãƒ¼ãƒ‘ーコンピュータã¯ã€NTTã‚³ãƒŸãƒ¥ãƒ‹ã‚±ãƒ¼ã‚·ãƒ§ãƒ³ã‚ºæ ªå¼ä¼šç¤¾(本社：æ±äº¬éƒ½åƒä»£ç”°åŒºã€ä»£è¡¨å–ç· å½¹ç¤¾é•·ï¼šåº„å¸ å“²ä¹Ÿã€ä»¥ä¸‹ï¼šNTT Com)ã¨æ ªå¼ä¼šç¤¾NTTPCコミュニケーションズ(本社：æ±äº¬éƒ½æ¸¯åŒºã€ä»£è¡¨å–ç· å½¹ç¤¾é•·ï¼šç”°ä¸ åŸºå¤«ã€ä»¥ä¸‹ï¼šNTTPC)ã®é«˜é€Ÿæ¼”算処ç†(GPU)プラットフォームを採用ã—ã€è¨ˆç®—ノードã«ã¯NVIDIA(R)製 Tesla(R) P100 GPUã‚’1,024基æ載ã—ã¦ã„ã¾ã™ã€‚æ°‘é–“ä¼æ¥ã®ãƒ—ライベートãªè¨ˆç®—環境ã¨ã—ã¦ã¯ã€å›½å†…最大級ã¨ãªã‚Šã¾
Nvidia’s New Policy Limits GeForce Data Center Usage | Hacker News
Developers should be mad about this. Here's where to start contributing to open alternatives:ROCm: https://rocm.github.io/ OpenCL: https://www.khronos.org/opencl/ TensorFlow: https://github.com/tensorflow/tensorflow Nouveau: https://nouveau.freedesktop.org/wiki/ TensorFlow depends exclusively (st time of writing) on CUDA for GPU acceleration. Leaning on them to support OpenCL fully would unlock th
Andrew Ng先生ã®æ©Ÿæ¢°å¦ç¿’講義を解説ã™ã‚‹å‹‰å¼·ä¼šè³‡æ–™ã‚’公開 - エンタープライズギークス (Enterprise Geeks)
シリコンãƒãƒ¬ãƒ¼ã®æœ‰å大å¦ã§ã‚るスタンフォード大å¦ã§ã€Andrew Ng先生ã®æ•™ãˆã‚‹æ©Ÿæ¢°å¦ç¿’ã®è¬›ç¾©ãŒäººæ°—を集ã‚ã¦ã„る。ã“ã®è¬›ç¾©ã¯å½¢ã‚’変ãˆã€courseraã¨ã„ã†ç„¡æ–™ã®Web上オンラインコースã¨ã—ã¦ã‚‚å—講ã§ãã€ã“ã“日本ã§ã‚‚機械å¦ç¿’ã®å‹‰å¼·ãŒã—ãŸã„人é”ã®é–“ã§ã‚‚人気ã®è¬›ç¾©ã¨ãªã£ã¦ã„る。 [Machine Learning by Stanford University] https://www.coursera.org/learn/machine-learning/ 人気ã®ç§˜å¯† Andrew Ng先生ã®è¬›ç¾©ã¯å†…容ãŒéžå¸¸ã«å®Ÿç”¨çš„ã§ã‚ã‚‹ã“ã¨ãŒäººæ°—ã®ç†ç”±ã®1ã¤ã¨ã—ã¦æŒ™ã’られるã ã‚ã†ã€‚Andrew Ng先生ã¯ã‚·ãƒªã‚³ãƒ³ãƒãƒ¬ãƒ¼ã®ä¼æ¥ã‚’相手ã«æ©Ÿæ¢°å¦ç¿’ã®æŒ‡å°Žã‚’ã—ã¦ã„る。ãŠãらããã®çµŒé¨“ã‹ã‚‰ã€ç†è«–ã«èµ°ã‚Šã™ãŽãšã€æ©Ÿæ¢°å¦ç¿’ã‚’é©ç”¨ã™ã‚‹ãƒ—ãƒã‚¸ã‚§ã‚¯ãƒˆã«ãŠã„ã¦æœ¬å½“ã«æœ‰æ„義ãªå†…容ã«å¼·ãフォーカスã—ã¦ã„ã‚‹ã“ã¨ãŒæ„Ÿã˜ã‚‰ã‚Œã‚‹ã®ã 。Andr
ã¯ã˜ã‚ã¦ã®Adversarial Example
今回ã¯adversarial exampleã«ã¤ã„ã¦è§£èª¬ã—ã¦ã„ãã¾ã™ã€‚Adversarial exampleã¨ã„ã†ã®ã¯ã€ä¸‹å›³ã®ã‚ˆã†ã«æ‘‚動を与ãˆã‚‹ã“ã¨ã«ã‚ˆã‚Šãƒ¢ãƒ‡ãƒ«ã«é–“é•ã£ãŸç”ãˆã‚’出力ã•ã›ã¦ã—ã¾ã†ã‚‚ã®ã®ã“ã¨ã§ã™ã€‚ ã“ã®ä¾‹ã§ã¯ã€ã‚‚ã¨ã‚‚ã¨ãƒ¢ãƒ‡ãƒ«ãŒãƒ‘ンダã¨æ£ã—ã分類ã™ã‚‹ã“ã¨ãŒã§ãã¦ã„ãŸç”»åƒã«æ‘‚動を与ãˆã‚‹ã“ã¨ã§ã€ãƒ†ãƒŠã‚¬ã‚¶ãƒ«ã¨èª¤åˆ†é¡žã•ã›ã¦ã„ã¾ã™ã€‚ã—ã‹ã—ã€äººé–“ã«ã¯å…ƒã®ç”»åƒã¨ã®é•ã„ã¯ã»ã¨ã‚“ã©åˆ†ã‹ã‚‰ãšã€ãƒ‘ンダã®ã¾ã¾ã«è¦‹ãˆã¾ã™ã€‚ Adversarial exampleã¯æ©Ÿæ¢°å¦ç¿’モデルを実用化ã—ã¦ã„ã上ã§å¤§ããªå•é¡Œã¨ãªã‚Šã¾ã™ã€‚例ãˆã°ã€äº¤é€šæ¨™è˜ã‚’adversarial exampleã«ã—ã¦ã—ã¾ãˆã°ã€è‡ªå‹•é‹è»¢è»Šã‚’ã ã¾ã›ã¦ã—ã¾ã†å¯èƒ½æ€§ãŒã‚ã‚Šã¾ã™ã€‚ 注目を集ã‚ã¦ãã¦ã„ã‚‹ç ”ç©¶åˆ†é‡Žã§ã™ãŒã€ã¾ã ã¡ã‚ƒã‚“ã¨èª¿ã¹ãŸã“ã¨ãŒãªã„ã¨ã„ã†äººã‚‚多ã„ã‹ã¨æ€ã„ã¾ã™ã€‚今回もãªã‚‹ã¹ãä¸å¯§ã«è§£èª¬ã—ã¦ã„ããŸã„ã¨æ€ã„ã¾ã™ã€‚ 目次 基礎 攻撃 防御 論文紹介
リリースã€éšœå®³æƒ…å ±ãªã©ã®ã‚µãƒ¼ãƒ“スã®ãŠçŸ¥ã‚‰ã›
最新ã®äººæ°—エントリーã®é…ä¿¡
j次ã®ãƒ–ックマーク
kå‰ã®ãƒ–ックマーク
lã‚ã¨ã§èªã‚€
eコメント一覧を開ã
oページを開ã
Kaggle Blog – Medium
{{#tags}}- {{label}}
{{/tags}}