Ambulatory Sensing Systems

                       May 16th 2012, 10:00-16:00

   You may have seen the recent blog entry from Stephen Wolfram [1] in 
 which he analyzes data from his e-mails to obtained interesting facts 
 about his sleep trends, working habits, and even whereabouts during 
 the last decades. Or you may have heard about Deb Roy's work at MIT, 
 recording and processing all waking moments of his son's life in audio
 and video to study language development [2]. Collecting and mining 
 such observations over long periods of time is interesting for many 
 reasons, one of which is that it provides other scientific disciplines 
 such as psychology and psychiatry with valuable tools to assert better 
 hypotheses and diagnoses. The design of such systems is interesting on 
 its own since it combines a challenging monitoring environment with an
 inherently highly complex data structure.

   We will discuss in the course the use of embedded sensing techniques 
 to monitor human behavior outside laboratory environments, i.e., when 
 they lead their normal life. Current topics such as sleep monitoring,
 activity recognition, and proactive healthcare monitoring will be in-
 troduced, and as part of the course, sensors will be made available to 
 any students interested in monitoring themselves [3].

 This course will involve both a theoretical session and a practical 
 (literally) hands-on session.

 In the practical sessions, you will get the chance to learn or refresh
 your skills in:
 - prototyping miniature sensor devices:
    - advanced micro-electronics soldering techniques
    - embedded programming of routines that process sensor readings
    - using advanced features of MEMS inertial sensors
 - building and evaluating sensor-based recognition systems
    - recording real-world data and getting ground truth annotation
    - avoiding bias from working with human participants
    - setting up proper cross-validation experiments for classifiers