• Corpus ID: 14203192

Probabilistic State-Dependent Grammars for Plan Recognition

  title={Probabilistic State-Dependent Grammars for Plan Recognition},
  author={David V. Pynadath and Michael P. Wellman},
Techniques for plan recognition under uncertainty require a stochastic model of the plangeneration process. We introduce probabilistic state-dependent grammars (PSDGs) to represent an agent's plan-generation process. The PSDG language model extends probabilistic contextfree grammars (PCFGs) by allowing production probabilities to depend on an explicit model of the planning agent's internal and external state. Given a PSDG description of the plan-generation process, we can then use inference… 

Figures and Tables from this paper

A probabilistic plan recognition algorithm based on plan tree grammars

Controlling the Hypothesis Space in Probabilistic Plan Recognition

A heuristic weighted model counting algorithm is presented that limits the number of generated plan execution models in order to recognize goals quickly by computing their lower and upper bound likelihoods.

Abstract Hidden Markov Models for Online Probabilistic Plan Recognition

The structure of the stochastic model rep- resenting the execution of the general AMP is analyzed and an efficient hybrid Rao-Blackwellised sampling method for policy recognition that scales well with the number of levels in the plan hierarchy is provided.

SLIM: Semi-Lazy Inference Mechanism for Plan Recognition

A new and efficient algorithm for online plan recognition called SLIM (Semi-Lazy Inference Mechanism), which combines both a bottom-up and top-down parsing processes, which allow it to commit only to the minimum necessary actions in real-time, but still provide complete hypotheses post factum.

Extending Bayesian Logic Programs for Plan Recognition and Machine Reading

This work develops an approach to abductive plan recognition using BLPs and extends BLPs to use logical abduction to construct Bayesian networks and calls the resulting model Bayesian Abductive Logic Programs (BALPs), demonstrating that BALPs outperform the existing state-of-art methods like Markov Logic Networks (MLNs).

Recognizing Plans with Loops Represented in a Lexicalized Grammar

It is shown how the loop-handling methods from formal grammars can be extended to the more general plan recognition problem and provide a method for encoding loops in an existing plan recognition system that scales linearly in the number of loop iterations.

Plan Recognition in Continuous Domains

It is shown that through mirroring, a formalization of recognition problems which admits continuous environments, as well as discrete domains, the authors can apply continuous-world motion planners in plan recognition through generalizing plan-recognition by planning.

Considering State in Plan Recognition with Lexicalized Grammars

  • C. Geib
  • Computer Science
  • 2012
This paper documents extending the ELEXIR (Engine for LEXicalized Intent Recognition) system with a world model and allows a number of additions to the algorithm, most significantly conditioning probabilities for recognized plans on the state of the world during execution.

Automatic Generation of Plan Libraries for Plan Recognition Performance Evaluation

A mechanism to automatically generate arbitrarily complex plan libraries is developed, such plan library generation can be directed through a number of parameters to allow for systematic experimentation.



Probabilistic grammars for plan recognition

New inference algorithms that generate a Bayesian network representation of the PCFG distribution and specialized algorithms that exploit the particular independence properties of the PSDG language to maintain a more efficient summary of evidence in the form of a belief state are presented.

A New Model of Plan Recognition

A new abductive, probabilistic theory of plan recognition that accounts for phenomena omitted from most previous plan recognition theories: notably the cumulative effect of a sequence of observations of partially-ordered, interleaved plans and the effect of context on plan adoption.

A Bayesian Model of Plan Recognition

Accounting for Context in Plan Recognition, with Application to Traffic Monitoring

A general Bayesian framework encompassing this view is presented, and it is shown how context can be exploited in plan recognition on a problem in traffic monitoring.

Generalized Queries on Probabilistic Context-Free Grammars

An algorithm for constructing Bayesian networks fromPCFGs is presented, and it is shown how queries or patterns of queries on the network correspond to interesting queries on PCFGs.

Towards History-based Grammars: Using Richer Models for Probabilistic Parsing

HBG incorporates lexical, syntactic, semantic, and structural information from the parse tree into the disambiguation process in a novel way and significantly outperforms P-CFG, increasing the parsing accuracy rate from 60% to 75%, a 37% reduction in error.

A Message Passing Algorithm for Plan Recognition

A general framework for plan recognition whose formulation is motivated by a general purpose algorithm for effective abduction, and the class of plans for which the method is applicable is wider than those previously proposed, as both recursive and optional plan components can be represented.

Effective Bayesian Inference for Stochastic Programs

This paper proposes a stochastic version of a general purpose functional programming language that contains random choices, conditional statements, structured values, defined functions, and recursion, and provides an exact algorithm for computing conditional probabilities of the form Pr(P(x) | Q(x)) where x is chosen randomly from this distribution.

Simulation-based inference for plan monitoring

This work investigates and extend simulation algorithms for approximate inference on Bayesian networks and proposes a new algorithm, called Rewind/Replay, for generating a set of simulations weighted by their likelihood given past observations.

Generalized Plan Recognition

A new theory of plan recognition that is significantly more powerful than previous approaches and employs circumscription to transform a first-order theory of action into an action taxonomy, which can be used to logically deduce the complex action(s) an agent is performing.