Paper Reading #3: Lowering the barrier to applying machine learning

Comments:

Jacob Lillard:
http://chi-jacob.blogspot.com/2011/01/paper-reading-3-lowering-barrier-to.html

Angel Narvaez:
http://angel-at-chi.blogspot.com/2011/01/paper-reading-3-lowering-barrier-to.html

Reference Information:

Title: Lowering the barrier to applying machine learning
Author: Kayur Patel
Presentation Venue: CHI EA '10 Proceedings of the 28th of the international conference extended abstracts on Human factors in computing systems

Summary:

Machine learning algorithms have been applied to difficult problems by researchers, but their findings have shown that such implementations are themselves difficult. This is one aim that this research paper is attempting to address. The researcher has noticed that many of the previous findings on machine learning have not lead to more effective tools. The creation of tools is one of the main motivations for the research.

Prior research of this topic in the area of tools creation has focused on structured tools. One tool example that is provided is once called Crayons, which allows the users to create models for image segmentation tasks. The researchers claim that structured tools tend to lower the level of flexibility needed for good machine learning techniques. They plan to address this flexibility issue with their work.

Discussion:

I think that making machine learning more effective could lead to advances in AI and simplify the way in which humans interact with computers. There are many benefits we could derive from this, including better tools for machine interaction, smarter AI tools and constructs, etc.

More flexible tools could lead to improved interfaces. We would likely achieve more efficient algorithms that would allow us the ability to create software more quickly. This could lead to machines offering suggestions as to more efficient ways in which they also could interact back with us.