Combining Discriminative Features to Infer Complex Trajectories
June 14, 2006
We propose a new model for the probabilistic estimation of continuous state variables from a sequence of observations, such as tracking the position of an object in video. This mapping is modeled as a product of dynamics features and observation features. Individual features are flexible in that they can switch on or off at each time-step depending on their inferred relevance (or on additional side information), and discriminative in that they need not model the full generative likelihood of the data.
When trained conditionally, this permits the inclusion of a broad range of rich features (for example, features relying on observations from multiple time-steps), and allows the relevance of features to be learned from labeled sequences.(Joint work with Simon Osindero and Rich Zemel, at ICML 2006)