Tracking with ranked signals
Tianyang Li, Harsh Pareek, Pradeep Ravikumar, Dhruv Balwada, Kevin G. Speer · 2015
We present a novel graphical model approach for a problem not previously considered in the ma-chine learning literature: that of tracking with ranked signals. The problem consists of track-ing a single target given observations about the target that consist of ranked continuous signals, from unlabeled sources in a cluttered environ-ment. We introduce appropriate factors to handle the imposed ordering assumption, and also incor-porate various systematic errors that can arise in this problem, particularly clutter or noise signals as well as missing signals. We show that infer-ence in the obtained graphical model can be sim-plified by adding bipartite structures with appro-priate factors. We apply a hybrid approach con-sisting of belief propagation and particle filter-ing in this mixed graphical model for inference and validate the approach on simulated data. We were motivated to formalize and study this prob-lem by a key task in Oceanography, that of track-ing the motion of RAFOS ocean floats, using range measurements sent from a set of fixed bea-cons, but where the identities of the beacons cor-responding to the measurements are not known. However, unlike the usual tracking problem in artificial intelligence, there is an implicit rank-ing assumption among signal arrival times. Our experiments show that the proposed graphical model approach allows us to effectively leverage the problem constraints and improve tracking ac-curacy over baseline tracking methods yielding results similar to the ground truth hand-labeled data. 1