Paper Reading #7: Real-time Interaction with Supervised Learning

Comments
Derek Landini
Jacob Lillard

References
Real-time Interaction with Supervised Learning
Rebecca Fiebrink
CHI 2010 Doctoral Consortium
http://wiki.cs.princeton.edu/index.php/ChucK/Wekinator/Instructions

http://wiki.cs.princeton.edu/index.php/ChucK/Wekinator/Instructions

Summary
In this paper, Rebecca Fiebrink discusses the possibility of creating interfaces for machine-learning interactions while focusing on the music composition and performance domain. In this research paper, Fiebrink sets to design a user interface that can guide musicians in the creation of datasets and building, evaluating, and refining trained models. Her work is inspired by the Weka interface that users to apply machine learning techniques to their own dataset. This being said, the author designed the Wekinator in order to achieve her goals of creating a general purpose user interface for handling real-time data.
Fiebrink recognizes that one of the easiest ways to improve a model's performance is to modify or add more training data as opposed to modify the learning algorithms or its parameters. However, the author recognizes that there are no user interfaces that facilitate such capability. Also, musicians may be concerned with timing issues which further complicates the design of such interactive interfaces.

Discussion
This paper was somewhat interesting since it attempts to bridge the gap between very specialized computer fields and other non-computer areas. If this paper successfully makes this gap smaller, then humanity may see computer science as a more concrete field.