Classification Accuracy from the Perspective of the User - Real-Time Interaction with Physiological Computing

Explore a conference talk examining the relationship between mathematical classification accuracy and user-perceived accuracy in physiological computing systems. Delve into a study where participants watched movie trailers, indicated their interest levels, and received real-time feedback from a Support Vector Machine (SVM) algorithm classifying their psychophysiological data. Discover how psychological biases, expectations, and trust development influence perceived accuracy, and learn about the significance of the F1 score as a predictor. Gain insights into implicit tagging, adaptive curation, and the challenges of measuring interest in real-time interactive systems.

Intro
implicit tagging
adaptive curation
what to measure?
measurement of interest
study 1 results
perceived system accuracy
the system
two phases
design
perceived error rate
mathematical accuracy & perceived accuracy
summary