Discriminative Viewer Identification using Generative Models of Eye Gaze
This addresses viewer identification for applications like security or personalization, but it is incremental as it builds on existing generative models with a discriminative approach.
The paper tackled the problem of identifying viewers from their eye gaze on arbitrary images by deriving Fisher kernels from generative stochastic models of eye movements, and found that using an SVM with Fisher kernel improved classification performance over the underlying generative models.
We study the problem of identifying viewers of arbitrary images based on their eye gaze. Psychological research has derived generative stochastic models of eye movements. In order to exploit this background knowledge within a discriminatively trained classification model, we derive Fisher kernels from different generative models of eye gaze. Experimentally, we find that the performance of the classifier strongly depends on the underlying generative model. Using an SVM with Fisher kernel improves the classification performance over the underlying generative model.