Select a box in the video to open that research area. Details and all videos on the research page →
Overview video (36 s). The clips are taken from the videos on the research page: the one pixel attack, the SyncMap run and the robot footage are real; the others are illustrations.
Exploring the robustness of deep neural networks.
Adaptating like rats to dynamic mazes.
Evolving neuron models, topology and weights.
Tackling autonomous driving and robotics.
In the news
AI image recognition fooled by single pixel changeMIT Technology ReviewHow do you turn a dog into a car? Change a single pixel.
AIのブラックボックス解決へ計算過程を可視化EurekAlert! · Kyushu UniversityUnlocking the 'black box': scientists reveal AI's hidden thoughts
Can you get from 'dog' to 'car' with one pixel? Japanese AI boffins can
A turtle is a rifleBoing BoingResearchers can fool machine-learning vision systems with a single, well-placed pixel
One pixel attack for deceiving deep neural networksResearch Opportunities
Our lab has in-house cutting edge GPU and CPU servers, it is located in one of the biggest cities in Japan with great infra and an exhuberating nature. For more details about our lab environment and opportunities to join us check the link below.
Join Us