Description:
The multiplication of computing devices with which a person in the developed world interacts has grown rapidly. From laptops and mobile phones to automobiles and urban infrastructure, bits are being computed all around us. Each of these platforms brings with it a suite of sensors which are measuring the world of the user. Sometimes the sensors are intentional like GPS and speedometers, but other times they emerge from the repurposing of other hardware, such as Wi-Fi and camera phone location systems.
Each of these sensors, however, looks at the world in a very limited way, for example by generating latitude and longitude pairs or outputting miles per hour. This is not the way that most users reason about or communicate their situation. In order to use the portfolio of sensors around them, computers need to do a better job of rapidly transforming sensor readings into concepts which users can utilize for communication and decision making.
In this talk, I will describe an approach to solving a generalized version of the "Position to Place" problem, which I call the "Sensors to Semantics" problem. Using lightweight artificial intelligence techniques coupled with mobile user interface systems, we collect translations of sensor readings to semantic concepts. By learning across these translations we enable intelligent context-aware support for application developers.
Speaker(s):
Donald J. Patterson, Assistant Professor, Donald Bren School of Information and Computer Sciences, University of California at Irvine
|
