.png "back to home page")
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/15810597/Screen_Shot_2019_03_06_at_11.23.46_AM.png)
In collaboration with Google, we're releasing Activation Atlases: a new technique for visualizing what interactions between neurons can represent.
— OpenAI (@OpenAI) March 6, 2019
?Blog: https://t.co/nLfd0vpED4
?Paper: https://t.co/qh2gozddHE
?Code: https://t.co/VwLN5j3uZo
?️Demo: https://t.co/9d1PmPqt6H pic.twitter.com/7c6Q6GvNLq
Congratulations @openai...this is fascinating.
— Grady Booch (@Grady_Booch) March 6, 2019
Looking at this from the lens of systems engineering, it tells me that we are just at the threshold of developing entirely new classes of tools for instrumenting, testing, and debugging AI-saturated systemshttps://t.co/hMe6J39XOv
In collaboration with Google, we're releasing Activation Atlases: a new technique for visualizing what interactions between neurons can represent.
— OpenAI (@OpenAI) March 6, 2019
?Blog: https://t.co/nLfd0vpED4
?Paper: https://t.co/qh2gozddHE
?Code: https://t.co/VwLN5j3uZo
?️Demo: https://t.co/9d1PmPqt6H pic.twitter.com/7c6Q6GvNLq
Cool tools from us & Google on neural nets visualization: https://t.co/T0w2gkq1jU
— Wojciech Zaremba (@woj_zaremba) March 6, 2019
A new tool from Google and OpenAI lets us better see through the eyes of artificial intelligence. #aiethics #ArtificialIntelligence #AI https://t.co/eB7Asfge1Z
— Nordic AI Artificial Intelligence Institute (@nordicinst) March 6, 2019
A new tool from Google and OpenAI lets us better see through the eyes of artificial intelligence https://t.co/yoeVjLVUIy pic.twitter.com/kda6g4EI8T
— The Verge (@verge) March 6, 2019
A new tool from Google and OpenAI lets us better see through the eyes of artificial intelligence#AI https://t.co/jvnLvwAXWT
— Martin Ford (@MFordFuture) March 6, 2019
For the last few years, one of the staples of my research has been visualizing individual neurons in vision models. But that's only a partial picture -- neurons work together.
— Chris Olah (@ch402) March 6, 2019
Activation Atlases are a way to explore the space neurons jointly represent.https://t.co/TEa9WuYBMg pic.twitter.com/yb3zvpxmhq
Very excited about the visualization techniques explored in this article! The potential for understanding AI is awesome and the implication for understanding our data is huge.https://t.co/Jsw52xjXSP
— Ian Johnson (@enjalot) March 6, 2019
I learn so much working with @shancarter @zanstrong @colah and @ludwigschubert
???#AI: Exploring Neural Networks with Activation Atlases https://t.co/UbIQQjWyqi #DeepLearning #datascience #bigdata #artificialintelligence #machinelearning pic.twitter.com/4TZHjv6UkG
— sjoerd (@sjoerdapp) March 6, 2019
Exploring Neural Networks with Activation Atlases https://t.co/ZaK257BWfA by @GoogleAI & @OpenAI
— Alexandr Kalinin (@alxndrkalinin) March 6, 2019
- feature visualizations of averaged activations to visualize a global picture seen by neural net
- understanding high-level confusions
- highlighting class-specific activations
etc pic.twitter.com/YIEOUHOMD7
In collaboration with colleagues at @OpenAI, we're publishing "Exploring Neural Networks with Activation Atlases", a new @distillpub that describes a technique to help determine what image classification neural networks "see" when provided an image. https://t.co/hSDCELi3AP
— Google AI (@GoogleAI) March 6, 2019
The activation atlas paper is a collaboration between @GoogleAI and @OpenAI. For overview articles see https://t.co/D5NQUKuXTMhttps://t.co/egQEkUvgBV
— Shan Carter (@shancarter) March 6, 2019
AI에 있어서 활성화지도, 즉 액티베이션 아틀라스 https://t.co/w9VY8q8yUP
— editoy (@editoy) March 7, 2019
This is how #machine vision #algorithms make sense of the world: https://t.co/DQUGZLLANn#ArtificialIntelligence #Machinelearning #computervision #google @ch402 @jjvincent
— Advanced Intelligent Systems (@Advintellsyst) March 7, 2019
A new tool from Google and OpenAI lets us better see through the eyes of artificial intelligence https://t.co/53P4e5jmMK #ai #ArtificialInteligence #google #openai pic.twitter.com/KfW7XACQdy
— Nige Willson (@nigewillson) March 7, 2019
How does AI view the world? Not like a human, is one of the big takeaways from Google and OpenAI's latest research on this topic https://t.co/03YygagRQb.
— James Vincent (@jjvincent) March 7, 2019
A new tool from Google and OpenAI lets us better see through the eyes of artificial intelligence - The Verge
— Iain Brown, PhD (@IainLJBrown) March 6, 2019
Read more here: https://t.co/kU9c4TdZ76#ArtificialIntelligence #AI #DataScience #MachineLearning #BigData #DeepLearning #NLP #Robots #IoT
A new tool from Google and OpenAI lets us better see through the eyes of artificial intelligence https://t.co/v3cuzqPhnu
— Darwin's Religion (@DarwinsReligion) March 7, 2019
This is how machine vision #algorithms make sense of the world. #AI #ML @kashthefuturist @HaroldSinnott @YuHelenYu @terence_mills @kuriharan @MHcommunicate @evankirstel @ipfconline1 @kenlittle https://t.co/KTytgEHZOK
— Terence Leung (@TerenceLeungSF) March 7, 2019
Trying to understand how #AI sees the world is not easy.#computerscience #Technology #Future #Tech #Robots #DeepLearning @mikequindazzi @ronald_vanloon@ipfconline1 @deeplearn007@chidambara09 @machinelearn_dhttps://t.co/qApAzOYeiV
— Taiwan Tech Arena (@TaiwanTechArena) March 7, 2019
Feature visualizations, letting us see through the eyes of the neural network. The hidden layers of neural networks are quite fun to inspect. Great interactive article by Distill. https://t.co/f4pWO5kdxs pic.twitter.com/PgpU7QkfEH
— Reza Zadeh (@Reza_Zadeh) March 7, 2019
Exploring Neural Networks with Activation Atlases https://t.co/aE3b7MXpaz #deeplearning
— Eclipse DL4J (@deeplearning4j) March 6, 2019
Google AI、Activation Atrasを公開。CNNがどのような画像特徴量に反応するか、全体をカタログ化。 #このCNN可視化がすごい2019 https://t.co/nztgZfpYcE
— Kazunori Sato (@kazunori_279) March 6, 2019
A new way to visualize how multiple neurons, in an image classifier, interact to identify features (eyes, edges, etc.). @OpenAI + @GoogleAI present an "Activation Atlas." It is also a good way to find adversarial features that can trick the classifier. https://t.co/Q79VgxyWXx https://t.co/oKUjBSQ9AT
— Cezanne Camacho (@cezannecam) March 6, 2019
OpenAI and Google released Activation Atlases, a way to inspect neural nets https://t.co/7SMDQsGC4m
— Mark O. Riedl (@mark_riedl) March 7, 2019
They held this under press embargo. Is the special media attention warranted? It’s neat, but (even they admit) relatively incremental.
#google #ai 各レイヤー各ニューロンが、画像のどの要素（ex. 犬の鼻）に反応しやすいかを可視化。
— Atsushi Tabata (@atsushieeeee) March 6, 2019
最近こういう研究は多いけど、他の類似した先行研究との差は何なのかな？
ちゃんと論文読まないとわからんなーhttps://t.co/KVOIcTvA5p