Facilitated by Foo Café at Goto 10, Dr. Abbey Wadron discusses the surprising flaws in machine learning and how we as humans can fool the artificial intelligence.

Impressive breakthroughs have been made in the field of artificial intelligence over the past few years. Human level computer vision in particular has made many applications possible from surveillance to self-driving cars. But how intelligent is AI really? In this talk we'll explore some of the surprising ways in which machine learning algorithms can be fooled, from turtle-rifles to stop sign attacks for cars.

We'll take a quick overview of deep learning with convolutional neural networks (CNNs) and then we'll introduce adversarial examples and show how to use them to trick CNNs.

Speaker is Dr. Abbey Wadron. Abbey has a PhD in experimental particle physics from the University of Oxford and has done research at many of the World's major laboratories including CERN, Fermilab and J-PARC.

Abbey is also organiser of the AI Retreats initiative, https://www.computerscienceretreat.com/. Check it out!

Agenda

18:00 Arrive

18:15 Abbey Wadron's talk begins

19:00 Break with food and drink

19:30 Q & A

Welcome!