Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Emotion Sense: From Design to Deployment

824 views

Published on

Research talk @ St. Andrew's University, October 2014

Published in: Mobile
  • Be the first to like this

Emotion Sense: From Design to Deployment

  1. 1. Emotion Sense: From Design to Deployment @neal_lathia – October 2014
  2. 2. Apply to new domains Analyse data Market & react Re-design & launch First Trial Collecting Sensor Data Background
  3. 3. Analyse data Market & react Re-design & launch First Trial Collecting Sensor Data Background Ongoing work. Lathia et. al. “Contextual Dissonance: Design Bias in Sensor Based Experience Sampling Methods.” Ubicomp 2013. Lathia et. al. “Smartphones for Large-Scale Behaviour Change Interventions.” IEEE Pervasive. Apply to new domains Lathia et. al. “Open-Source Smartphone Libraries for Computational Social Science.” MCSS 2013.
  4. 4. Analyse data Market & react Re-design & launch First Trial Collecting Sensor Data Background Lathia et. al. “Smartphones for Large-Scale Behaviour Change Interventions.” IEEE Pervasive. Apply to new domains
  5. 5. ● …[short term] to collect research data about moods and behaviours as people experience them ● …[long term] to explore whether machine learning approaches could infer people's subjective responses/complex behaviours ● …[vision] to understand the extent that behavioural support can be automated and personalised with sensor data
  6. 6. Apply to new domains Analyse data Market & react Re-design & launch First Trial Collecting Sensor Data Background Lathia et. al. “Open-Source Smartphone Libraries for Computational Social Science.” MCSS 2013.
  7. 7. “... a number of challenges remain in the development of sensor-based applications [...] there is mixed API and operating system (OS) support to access the low-level sensors...” “A Survey of Mobile Phone Sensing,” Lane et. al
  8. 8. Android ESSensorManager ● Everything as a “sensor” ● Simple API with two modes (get, subscribe); sensor data in two lines of code ● API exposes battery issues, configuration to programmer ● Student project-led evaluation ● https://github.com/xsenselabs
  9. 9. Apply to new domains Analyse data Market & react Re-design & launch First Trial Collecting Sensor Data Background Lathia et. al. “Contextual Dissonance: Design Bias in Sensor Based Experience Sampling Methods.” Ubicomp 2013. How should we design the relationship between interacting and collecting 'labelled' data?
  10. 10. We built a system that includes: sensor data collection, ESM interfaces, etc., and remote reconfiguration.
  11. 11. 22 users; 1-month; questions about mood & current context (location, sociability); background sensing from many sensors; triggers remotely reconfigured weekly.
  12. 12. Dissonance; a tension or clash resulting from the combination of two disharmonious elements
  13. 13. Dissonance; between using sensor states to trigger ESM surveys while using sensor data to quantify context and behaviour.
  14. 14. Apply to new domains Analyse data Market & react Re-design & launch First Trial Collecting Sensor Data Background
  15. 15. Ongoing work. Apply to new domains Analyse data Market & react Re-design & launch First Trial Collecting Sensor Data Background
  16. 16. We have: State/trait surveys Context labels Sensor data
  17. 17. signals of behaviour
  18. 18. Ongoing work. Apply to new domains Analyse data Market & react Re-design & launch First Trial Collecting Sensor Data Background
  19. 19. Generalise sensor-enhanced experience sampling tool. Currently in alpha testing.
  20. 20. Emotion Sense: From Design to Deployment @neal_lathia – October 2014

×