Inferring Relevance from Eye Movements Using Generic Neural Microcircuits
In: NIPS 2005 Workshop on Machine Learning for Implicit Feedback and User Modeling, 10 Dec 2005, Whistler, Canada.
In the ''Inferring Relevance from Eye Movements Challenge 2005'', contestants were asked to apply machine learning techniques for predicting sentence relevancies based on eye movements of the readers. The two-part competition consisted of a classification problem and a time series analysis problem. In our winning solution to the time series analysis problem, we applied generic neural microcircuit as a nonlinear operator on time series with discriminative classifier as a readout. The results suggest that the model used has some practical merits as a generic time series analysis tool.