Easily Fooling Deep Neural Networks
Data Skeptic16 Jan 2015

Easily Fooling Deep Neural Networks

My guest this week is Anh Nguyen, a PhD student at the University of Wyoming working in the Evolving AI lab. The episode discusses the paper Deep Neural Networks are Easily Fooled [pdf] by Anh Nguyen, Jason Yosinski, and Jeff Clune. It describes a process for creating images that a trained deep neural network will mis-classify. If you have a deep neural network that has been trained to recognize certain types of objects in images, these "fooling" images can be constructed in a way which the network will mis-classify them. To a human observer, these fooling images often have no resemblance whatsoever to the assigned label. Previous work had shown that some images which appear to be unrecognizable white noise images to us can fool a deep neural network. This paper extends the result showing abstract images of shapes and colors, many of which have form (just not the one the network thinks) can also trick the network.

Det här avsnittet är hämtat från ett öppet RSS-flöde och publiceras inte av Podme. Det kan innehålla reklam.

Avsnitt(612)

TastePrint

TastePrint

What if we could 3D print food with customized flavors and textures? Yamato Miyatake joins us to explore how digital food fabrication, human-computer interaction, and AI could transform the future of ...

29 Okt 26min

Gestures Around the Ear

Gestures Around the Ear

The next evolution of wearable technology may not require new screens or controllers, but new ways to interact with the devices we already wear. Shimon explores how gesture-sensing earbuds could trans...

18 Okt 30min

Implicit Interactions

Implicit Interactions

How do we design robots and autonomous vehicles that understand the unwritten rules of human behavior? Kyle speaks with Cornell Tech professor Wendy Ju about implicit interaction, "Wizard of Oz" proto...

5 Okt 44min

The Lived Informatics Model

The Lived Informatics Model

The data we collect about ourselves can tell us a lot—but only if the technology collecting it actually fits into our lives. Daniel Epstein explores personal informatics, from fitness trackers and foo...

25 Sep 34min

Recommender Systems Today and Tomorrow

Recommender Systems Today and Tomorrow

In the final episode of our Recommender Systems season, we explore the growing questions of trust, manipulation, privacy, fairness, sustainability, and user control. From fake reviews and shilling att...

9 Sep 22min

Recommender Systems Optimization Goals

Recommender Systems Optimization Goals

In part two of the Data Skeptic Recommender Systems season finale, Kyle asks a deceptively difficult question: what should recommender systems actually optimize for? Drawing on conversations from acro...

1 Sep 31min

Recommender Systems Origin Story

Recommender Systems Origin Story

Where did recommender systems come from, and how do we know when they're actually working? In part one of Data Skeptic's three-part Recommender Systems finale, Kyle traces the field from collaborative...

18 Aug 25min

Social Choice for Fair Recommendations

Social Choice for Fair Recommendations

Recommender systems influence nearly every aspect of our digital lives—but what does it mean for those systems to be fair? Robin Burke joins Data Skeptic to discuss the history of recommender systems,...

27 Juli 42min

Populärt inom Vetenskap

p3-dystopia
dumma-manniskor
allt-du-velat-veta
rss-ronden
angestpodden
kapitalet-en-podd-om-ekonomi
sexet
hacka-livet
rss-ufobortom-rimligt-tvivel
medicinvetarna
rss-vetenskapsradion-2
rss-vetenskapsradion
rss-kriminologerna
det-morka-psyket
rss-experimentet
rss-spraket
rss-ai-med-jonas-benjamin
paranormalt-med-caroline-giertz
bildningspodden
ufo-sverige