Parameter Learning and Change Detection Using a Particle Filter With Accelerated Adaptation
This addresses the challenge of detecting and adapting to parameter changes in dynamic models for applications like finance or signal processing, but it appears incremental as it combines existing particle filter and genetic algorithm techniques.
The paper tackles the problem of online sequential filtering when data mismatches the model posterior due to parameter changes, such as regime shifts or stochastic volatility, by constructing a particle filter with genetic algorithm elements. The result is a filter that adapts rapidly to regime shifts and provides a heuristic to distinguish them from stochastic volatility, though no concrete numbers are provided.
This paper presents the construction of a particle filter, which incorporates elements inspired by genetic algorithms, in order to achieve accelerated adaptation of the estimated posterior distribution to changes in model parameters. Specifically, the filter is designed for the situation where the subsequent data in online sequential filtering does not match the model posterior filtered based on data up to a current point in time. The examples considered encompass parameter regime shifts and stochastic volatility. The filter adapts to regime shifts extremely rapidly and delivers a clear heuristic for distinguishing between regime shifts and stochastic volatility, even though the model dynamics assumed by the filter exhibit neither of those features.