# Analyzing Team Actions with Cascading HMM

### Cite as:

B. White, N. Blaylock, and L. Bölöni. Analyzing Team Actions with Cascading HMM. In The 22nd International FLAIRS Conference, , pp. 129–135, May 2009.

### Abstract:

While team action \em recognition has a relatively extended literature, less attention has been given to the detailed realtime analysis of the internal structure of the team actions. This includes recognizing the current state of the action, predicting the next state, recognizing deviations from the standard action model, and handling ambiguous cases. The underlying probabilistic reasoning model has a major impact on the type of data it can extract, its accuracy, and the computational cost of the reasoning process. In this paper we are using Cascading Hidden Markov Models (CHMM) to analyze Bounding Overwatch, an important team action in military tactics. The team action is represented in the CHMM as a plan tree. Starting from real-world recorded data, we identify the subteams through clustering and extract team oriented discrete features. In an experimental study, we investigate whether the better scalability and the more structured information yielded by CHMM comes with an unacceptable cost in accuracy. We find the a properly parametrized CHMM estimating the current goal chain of the Bounding Overwatch plan tree comes very close to a flat HMM estimating only the overall Bounding Overwatch state (a subset of the goal chain) at a respective overall state accuracy of 95\% vs 98\%, making the CHMM a good candidate for deployed systems.

### BibTeX:

@inproceedings{White-2009-FLAIRS,
author = "B. White and N. Blaylock and L. B{\"o}l{\"o}ni",
title = "Analyzing Team Actions with Cascading {HMM}",
booktitle = "The 22nd International FLAIRS Conference, ",
location = "Sanibel Island, Florida",
year = "2009",
month = "May",
pages = "129-135",
abstract = {
While team action {\em recognition} has a relatively extended
literature, less attention has been given to the detailed realtime
analysis of the internal structure of the team actions. This includes
recognizing the current state of the action, predicting the next
state, recognizing deviations from the standard action model, and
handling ambiguous cases. The underlying probabilistic reasoning model
has a major impact on the type of data it can extract, its accuracy,
and the computational cost of the reasoning process. In this paper we
are using Cascading Hidden Markov Models (CHMM) to analyze Bounding
Overwatch, an important team action in military tactics.
The team action is represented in the CHMM as a plan tree. Starting
from real-world recorded data, we identify the subteams through
clustering and extract team oriented discrete features. In an
experimental study, we investigate whether the better scalability and
the more structured information yielded by CHMM comes with an
unacceptable cost in accuracy. We find the a properly parametrized
CHMM estimating the current goal chain of the Bounding Overwatch plan
tree comes very close to a flat HMM estimating only the overall
Bounding Overwatch state (a subset of the goal chain) at a respective
overall state accuracy of 95\% vs 98\%, making the CHMM a good
candidate for deployed systems.
},
}


Generated by bib2html.pl (written by Patrick Riley, Lotzi Boloni ) on Fri Dec 29, 2017 18:25:30