Name
Abstract | ||
|---|---|---|
Maintaining the a-posteriori distribution of categorical states given a sequence of noisy and ambiguous observations, e.g. sensor data, can lead to situations where one observation can correspond to a large number of different states. We call these states symmetrical as they cannot be distinguished given the observation. Considering each of them during the inference is computationally infeasible, even for small scenarios. However, the number of situations (called hypotheses) can be reduced by abstracting from particular ones and representing all symmetrical in a single abstract state. We propose a novel Bayesian Filtering algorithm that performs this abstraction. The algorithm that we call Lifted Marginal Filtering (LiMa) is inspired by Lifted Inference and combines techniques known from Computational State Space Models and Multiset Rewriting Systems to perform efficient sequential inference on a parametric multiset state description. We demonstrate that our approach is working by comparing LiMa with conventional filtering. |
Year | DOI | Venue |
|---|---|---|
2017 | 10.1007/978-3-319-67190-1_17 | Lecture Notes in Artificial Intelligence |
DocType | Volume | ISSN |
Conference | 10505 | 0302-9743 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
5 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Max Schröder | 1 | 4 | 5.86 |
| Stefan Lüdtke | 2 | 3 | 5.15 |
| Sebastian Bader | 3 | 51 | 14.66 |
| Frank Krüger | 4 | 53 | 10.43 |
| Thomas Kirste | 5 | 93 | 18.37 |