Limits...
Computational state space models for activity and intention recognition. A feasibility study.

Krüger F, Nyolt M, Yordanova K, Hein A, Kirste T - PLoS ONE (2014)

Bottom Line: Our results suggest that the combinatorial explosion caused by rich CSSM models does not inevitably lead to intractable inference or inferior performance.This means that the potential benefits of CSSM models (knowledge-based model construction, model reusability, reduced need for training data) are available without performance penalty.However, our results also show that research on CSSMs needs to consider sufficiently complex domains in order to understand the effects of design decisions such as choice of heuristics or inference procedure on performance.

View Article: PubMed Central - PubMed

Affiliation: Computer Science Institute, University of Rostock, Rostock, Germany.

ABSTRACT

Background: Computational state space models (CSSMs) enable the knowledge-based construction of Bayesian filters for recognizing intentions and reconstructing activities of human protagonists in application domains such as smart environments, assisted living, or security. Computational, i. e., algorithmic, representations allow the construction of increasingly complex human behaviour models. However, the symbolic models used in CSSMs potentially suffer from combinatorial explosion, rendering inference intractable outside of the limited experimental settings investigated in present research. The objective of this study was to obtain data on the feasibility of CSSM-based inference in domains of realistic complexity.

Methods: A typical instrumental activity of daily living was used as a trial scenario. As primary sensor modality, wearable inertial measurement units were employed. The results achievable by CSSM methods were evaluated by comparison with those obtained from established training-based methods (hidden Markov models, HMMs) using Wilcoxon signed rank tests. The influence of modeling factors on CSSM performance was analyzed via repeated measures analysis of variance.

Results: The symbolic domain model was found to have more than 10(8) states, exceeding the complexity of models considered in previous research by at least three orders of magnitude. Nevertheless, if factors and procedures governing the inference process were suitably chosen, CSSMs outperformed HMMs. Specifically, inference methods used in previous studies (particle filters) were found to perform substantially inferior in comparison to a marginal filtering procedure.

Conclusions: Our results suggest that the combinatorial explosion caused by rich CSSM models does not inevitably lead to intractable inference or inferior performance. This means that the potential benefits of CSSM models (knowledge-based model construction, model reusability, reduced need for training data) are available without performance penalty. However, our results also show that research on CSSMs needs to consider sufficiently complex domains in order to understand the effects of design decisions such as choice of heuristics or inference procedure on performance.

Show MeSH

Related in: MedlinePlus

CSSM DBN structure.Boxes represent tuples of random variables. An arc starting/ending at a box ( =  a tuple) represents a set of arcs connected to the tuple's components. Nodes with double outline signify observed random variables.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4220990&req=5

pone-0109381-g002: CSSM DBN structure.Boxes represent tuples of random variables. An arc starting/ending at a box ( =  a tuple) represents a set of arcs connected to the tuple's components. Nodes with double outline signify observed random variables.

Mentions: For the probabilistic model of Sec. 1.2 (see Appendix S2 for details), a DBN with the structure given in Fig. 2 was used. is the observation data for time step , i. e. the sensor data as discussed in Sec. 2.1.4. is the associated time stamp, required to be strictly increasing. defines the hidden state. For this study, , the current goal, could be assumed to be constant, namely that the user has prepared the meal, eaten, and cleaned afterwards. A new action is selected according to the action selection heuristic , which incorporates the distance from the current state to the goal (for more details on action selection heuristics, see Sec. 4.1.5 of Appendix S4). is the LTS state for time step : either the result of applying the new action to the previous state, or by carrying over the old state. For the purpose of this study, actions could be assumed to be deterministic and with instantaneous effect. In contrast to the model defined in Sec. 1.2, actions in our model may last longer than a single time step. A model was chosen where multiple observations may correspond to a single action. This model introduces a real-valued random variable representing the starting time of an action and a boolean random variable signaling termination status of the previous action .


Computational state space models for activity and intention recognition. A feasibility study.

Krüger F, Nyolt M, Yordanova K, Hein A, Kirste T - PLoS ONE (2014)

CSSM DBN structure.Boxes represent tuples of random variables. An arc starting/ending at a box ( =  a tuple) represents a set of arcs connected to the tuple's components. Nodes with double outline signify observed random variables.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4220990&req=5

pone-0109381-g002: CSSM DBN structure.Boxes represent tuples of random variables. An arc starting/ending at a box ( =  a tuple) represents a set of arcs connected to the tuple's components. Nodes with double outline signify observed random variables.
Mentions: For the probabilistic model of Sec. 1.2 (see Appendix S2 for details), a DBN with the structure given in Fig. 2 was used. is the observation data for time step , i. e. the sensor data as discussed in Sec. 2.1.4. is the associated time stamp, required to be strictly increasing. defines the hidden state. For this study, , the current goal, could be assumed to be constant, namely that the user has prepared the meal, eaten, and cleaned afterwards. A new action is selected according to the action selection heuristic , which incorporates the distance from the current state to the goal (for more details on action selection heuristics, see Sec. 4.1.5 of Appendix S4). is the LTS state for time step : either the result of applying the new action to the previous state, or by carrying over the old state. For the purpose of this study, actions could be assumed to be deterministic and with instantaneous effect. In contrast to the model defined in Sec. 1.2, actions in our model may last longer than a single time step. A model was chosen where multiple observations may correspond to a single action. This model introduces a real-valued random variable representing the starting time of an action and a boolean random variable signaling termination status of the previous action .

Bottom Line: Our results suggest that the combinatorial explosion caused by rich CSSM models does not inevitably lead to intractable inference or inferior performance.This means that the potential benefits of CSSM models (knowledge-based model construction, model reusability, reduced need for training data) are available without performance penalty.However, our results also show that research on CSSMs needs to consider sufficiently complex domains in order to understand the effects of design decisions such as choice of heuristics or inference procedure on performance.

View Article: PubMed Central - PubMed

Affiliation: Computer Science Institute, University of Rostock, Rostock, Germany.

ABSTRACT

Background: Computational state space models (CSSMs) enable the knowledge-based construction of Bayesian filters for recognizing intentions and reconstructing activities of human protagonists in application domains such as smart environments, assisted living, or security. Computational, i. e., algorithmic, representations allow the construction of increasingly complex human behaviour models. However, the symbolic models used in CSSMs potentially suffer from combinatorial explosion, rendering inference intractable outside of the limited experimental settings investigated in present research. The objective of this study was to obtain data on the feasibility of CSSM-based inference in domains of realistic complexity.

Methods: A typical instrumental activity of daily living was used as a trial scenario. As primary sensor modality, wearable inertial measurement units were employed. The results achievable by CSSM methods were evaluated by comparison with those obtained from established training-based methods (hidden Markov models, HMMs) using Wilcoxon signed rank tests. The influence of modeling factors on CSSM performance was analyzed via repeated measures analysis of variance.

Results: The symbolic domain model was found to have more than 10(8) states, exceeding the complexity of models considered in previous research by at least three orders of magnitude. Nevertheless, if factors and procedures governing the inference process were suitably chosen, CSSMs outperformed HMMs. Specifically, inference methods used in previous studies (particle filters) were found to perform substantially inferior in comparison to a marginal filtering procedure.

Conclusions: Our results suggest that the combinatorial explosion caused by rich CSSM models does not inevitably lead to intractable inference or inferior performance. This means that the potential benefits of CSSM models (knowledge-based model construction, model reusability, reduced need for training data) are available without performance penalty. However, our results also show that research on CSSMs needs to consider sufficiently complex domains in order to understand the effects of design decisions such as choice of heuristics or inference procedure on performance.

Show MeSH
Related in: MedlinePlus