Attendees of the AMPLab Summer 2011 retreat used the OpinionSpace system (developed by Prof. Ken Goldberg’s group) to brainstorm the key technical challenges to be addressed by AMPLab.

Altogether, the effort involved 78 participants and 760 ratings of the generated ideas.   Ideas were ranked by the ratings they received by other participants. The top rated ideas for the AMPLab’s research agenda ranged from opinions on data integration, user feedback in large-scale machine learning, and questions of privacy and inference surrounding the ever-increasing number of ubiquitous computing devices.

The three top-ranked submissions were:

1. User Corrections and ML: How do we incorporate user corrections into large, distributed, high dimensional statistical models?
2. Data Integration:  How do you take data from many different sources,all of different types, sometimes built for different purposes, and combine them in a way that can allow for meaningful queries?
3. Mobile Applications. There are already 5 *billion* phones in use in the world, and they are in many ways revolutionizing society in evenmore dramatically than the PC did.  Though many of the issues with mobile data are similar to those from other forms of “big data”, there are unique challenges too.

A full report on the results from the Summer Retreat OpinionSpace exercise, including the top-ranked ideas, is posted here.