A Classification Framework for Partially Observed Dynamical Systems
This work addresses classification challenges in fields like biology and physics where data is incomplete, but it is incremental as it builds on existing model-based approaches.
The authors tackled the problem of classifying partially observed dynamical systems by learning in the model space using posterior distributions to account for uncertainty, and showed that classifier performance remains robust even with simpler inferential models if they capture essential task characteristics, as demonstrated on a biological pathway and stochastic double-well system.
We present a general framework for classifying partially observed dynamical systems based on the idea of learning in the model space. In contrast to the existing approaches using model point estimates to represent individual data items, we employ posterior distributions over models, thus taking into account in a principled manner the uncertainty due to both the generative (observational and/or dynamic noise) and observation (sampling in time) processes. We evaluate the framework on two testbeds - a biological pathway model and a stochastic double-well system. Crucially, we show that the classifier performance is not impaired when the model class used for inferring posterior distributions is much more simple than the observation-generating model class, provided the reduced complexity inferential model class captures the essential characteristics needed for the given classification task.