Democratizing AI for End Users: Designing Machine Learning Models Using Interactive Applications

Tuesday, 8 March 2022 | 9:00am - 9:45am (JST)
Zoom Meeting
  • Wataru Kawabe Graduate School of Information Science and Technology, The University of Tokyo
  • Yusuke Sugano Associate Professor, Institute of Industrial Science, The University of Tokyo
Event Description

Recent advances in machine learning (ML), especially deep neural networks, have greatly expanded the opportunities for various real-world applications in our daily lives. However, while ML has great potential, it is still difficult for novice users to design a model for their own purpose since building an ML model requires programming skills and mathematical knowledge. Interactive machine learning (IML) aims at providing a method for them to interact with ML algorithms and prototype their own models. There have been many research attempts to create interactive image/sound/text recognition systems, and this time I show you some examples I found worth checking. I also introduce my own work attempting to provide a generic image recognition model with text as output, and how novice users responded to the prototype system and solved various image recognition tasks with it.

About the Speaker

Wataru Kawabe is a master’s student in the Graduate School of Information Science and Technology at the University of Tokyo, specializing in computer science. His research interests include applied machine learning and human-computer interaction, and his current research focuses on understanding end-users through interaction with image/sound/text recognition applications.