We can no longer imagine a world without interacting with machines. Staying informed on developments in human-computer interaction research is important; therefore, VicarVision is happy to be a member of the Man-Machine Interaction Platform (MMI). MMI is a Dutch platform with people from business, universities, and research institutes that aims to support collaboration and knowledge exchange in this field.
VicarVision hosted the most recent (online) member meeting on our own expertise: the use of artificial intelligence for emotion recognition. We addressed this topic from multiple viewpoints with presentations on the technology and applications, but also presentations related to ethics of A.I.
Emotion Recognition Technology and Applications
Our CEO Tim den Uyl introduced the company and the wide area of applications VicarVision is involved in: from consumer and behavior analysis tools to e-health applications. Amogh Gudi, our machine learning expert, introduced the technology behind deep learning and facial analysis. An interesting fact is the exponential growth of data that we are experiencing in the last 10 years (50x increase). He explained how A.I. has surpassed human level in the task of face recognition with the use of deep learning. He also showed how we use these methods in a smart way to perform facial analysis for emotion classification. Next, Tess den Uyl gave a live demonstration of a psychology experiment from start to finish (on the specific emotion schadenfreude). The audience could start the FaceReader Online experiment via the browser and saw a short video that was meant to elicit some laughter (see the results below).
These FaceReader Online graphs show the results of the MMI mini-experiment. The question was: is schadenfreude a universal joy, or do some people really not laugh at the misery of others? In this video with a surprising and painful ending those that liked schadenfreude (left graph) were anticipating and laughing more than those who didn’t like it (although perhaps a small laugh couldn’t be repressed in the end).
Ethics of A.I.
We also invited Merel Noorman (a former employee of VicarVision) and now an assistant professor at Tilburg University on A.I. ethics. She presented some striking examples of where problems have arisen in the past and the dangers to consider when doing facial analysis. She also gave examples of concerns to address to prevent misuse, such as explainability and automation bias. We thought it was very interesting because we also think it is an important topic. To address this, we have recently created a task force to formulate which aspects of responsible A.I. we find important as a company. The final presentation was by MMI member Lucas Noldus, who gave practical examples on how to implement ethics into a solid sales guideline and policy for FaceReader.
Overall, it was a very interesting event, where we learned something, and hopefully the audience as well. Of course, one of the drawbacks of online meetings is the lack of a social drink afterward. Luckily, developers have found ways to recreate a cafe online and we had a nice afterparty in SpatialChat.