UIST 2010 New York, New York.
Author Bios: Kayur Patel is working on a PhD in Computer Science at Washington University.
Naomi Bancroft works for Google. She recently graduated from Washington University.
Steven M. Drucker is a Principal Researcher at Microsoft Research.
James Fogarty holds the position of Assistant Professor of Computer Science and Engineering at Washington University.
Andrew J. Ko is also an Assistant Professor at Washington University in the Information School.
James Landay is a Professor at Washington University who previously worked for Intel.
Summary
- Hypothesis: By using Gestalt, a development environment, developers will find it easier to apply machine learning.
- Methods: Eight participants were recruited to test Gestalt by using a baseline similar to MATLAB along with Gestalt to find and debug errors.
- Results: The participants preferred Gestalt and were able to more effectively locate errors with Gestalt.
- Contents: The paper begins with a brief description of the machine learning process. Then it goes on to describe what Gestalt is. Through implementation and analysis, programmers can manage classified pipelines and visualize computed data, respectively. The combination of Gestalt's features allows it to serve as a general-purpose tool for developers.
Discussion
I liked the concept of a development environment that aids in the incorporation of machine learning. Although I don't really have any experience in that field, I think such a development environment would still be an effective tool for the general programming population. Developers often have to deal with demands on a tight timetable, and something like Gestalt can help them meet those demands in a timely manner. The work in this paper can pave the way for such environments.
No comments:
Post a Comment