Joe Cabrera
Stuart Jones
Reference Info:
Gestalt: Integrated Support for Implementation and Analysis in Machine Learning
Kayur Patel, Naomi Bancroft, Steven M. Drucker, James Fogarty, Andrew J. Ko, James A. Landay
UIST '10 Proceedings of the 23nd annual ACM symposium on User interface software and technology
Summary:
Gestalt provides users a graphical way to display and perfect the field of machine learning. Gestalt allows users to implement the learning, analyze data and provide easy transitions from implementation to data. Widgets were also added that gave different abilities to the users such as parsing data or train and test models. There is also a table with shared analysis so that developers may see what attributes were used to classify things in the learning. Test data was presented was used to test against the learning code.
Discussion:
I enjoyed this paper because I feel I understood it more having taken AI and a good foundation of machine learning. This tools seems like a great way to test machine learning, especially when one needs instant results of the code displayed. I feel like this if a big step forward in machine learning and hopefully will help out the field alot.
The idea of graphically mapping machine learning is a fairly innovative one, this paper sounded very interesting. It's nice to know that people are continuing to innovate in the field.
ReplyDeleteI agree with you that this system is indeed a breakthrough in developer tools. Machine learning is becoming increasingly important to implement in software. I would interested in using this software.
ReplyDelete