Measuring and analyzing behavior using AI and machine learning

Organizer: Prof. Lucas Noldus, Radboud University Nijmegen and Noldus Information Technology, the Netherlands

Description: Compared to the important role for Computer Vision applied in research methodologies successfully for many years, the role of Deep Learning networks and algorithms can be viewed as the new kid on the behavioral research block. A promising world ahead of us, just as in the early years of many technologies, where many problems and challenges will be solved by applying Deep Learning.

In practice, the use of Deep Learning has its challenges itself. Training neural networks requires very large amounts of properly annotated data, which are hard to come by. Trained networks create results that are difficult to verify and reproduce scientifically in new situations. The dataset used for training is often not representative enough, leading to bias in the algorithms. And yet, there is a huge group of scientist working daily on improving their algorithms and training their networks resulting in a vastly growing deep learning toolset to perform quality research on behavior.

This symposium will address the data challenges while at the same time presenting successes from a deep learning approach to measuring behavior.