## Event Recognition in Sensor Networks by Means of Grammatical Inference

### Cached

### Download Links

Citations: | 8 - 5 self |

### BibTeX

@MISC{Geyik_eventrecognition,

author = {Sahin Cem Geyik and Boleslaw K. Szymanski},

title = {Event Recognition in Sensor Networks by Means of Grammatical Inference},

year = {}

}

### OpenURL

### Abstract

Abstract—Modern military and civilian surveillance applications should provide end users with the high level representation of events observed by sensors rather than with the raw data measurements. Hence, there is a need for a system that can infer higher level meaning from collected sensor data. We demonstrate that probabilistic context free grammars (PCFGs) can be used as a basis for such a system. To recognize events from raw sensor network measurements, we use a PCFG inference method based on Stolcke(1994) and Chen(1996). We present a fast algorithm for deriving a concise probabilistic context free grammar from the given observational data. The algorithm uses an evaluation metric based on Bayesian formula for maximizing grammar a posteriori probability given the training data. We also present a real-world scenario of monitoring a parking lot and the simulation based on this scenario. We described the use of PCFGs to recognize events in the results of such a simulation. We finally demonstrate the deployment details of such an event recognition system. I.

### Citations

688 | An Efficient Context-Free Parsing Algorithm
- Earley
- 1970
(Show Context)
Citation Context ...ent recognition steps. In [8], the authors are using a visual system to recognize human gestures and to detect interactions in a parking lot environment. Authors utilize an Earley-Stolcke parser [2], =-=[9]-=- to parse sequences into actions. A run-time incremental parsing is implemented in which the states (in Earley Model) whose probabilities drop below a certain value are removed. Also, only sequences o... |

381 |
The estimation of stochastic context-free grammars using the inside-outside algorithm. Computer Speech and Language
- Lari, Young
- 1990
(Show Context)
Citation Context ...mmar. In that case, the problem is to estimate the probabilities for the rules from the training data. The so-called inside-outside algorithm serves as the basis for the general solution of this task =-=[12]-=-, [13]. Constructing PCFGs from scratch, however, requires alternative learning techniques. There are two different approaches to this problem. The first one uses two operators: merge and chunk [2], [... |

274 |
Trainable Grammars for Speech Recognition
- Baker
- 1979
(Show Context)
Citation Context ...In that case, the problem is to estimate the probabilities for the rules from the training data. The so-called inside-outside algorithm serves as the basis for the general solution of this task [12], =-=[13]-=-. Constructing PCFGs from scratch, however, requires alternative learning techniques. There are two different approaches to this problem. The first one uses two operators: merge and chunk [2], [14]. T... |

256 | Recognition of visual activities and interactions by stochastic parsing
- Ivanov, Bobick
(Show Context)
Citation Context ...f 1.0 ∗ 0.8 ∗ 0.8 ∗ 0.2 =0.128. Probabilistic Context Free Grammars have many uses in speech recognition [15], [16], natural language processing [18], computational biology [17], sensor networks [4], =-=[8]-=-, [10], [11] etc. III. PREVIOUS WORK In this section, we present a brief overview of literature on PCFG inference and application of PCFGs to sensor data. A. Event Recognition using PCFGs There are se... |

133 | Stochastic context-free grammars for tRNA modeling
- Sakakibara, Brown, et al.
- 1994
(Show Context)
Citation Context ...aab bbhas the probability of 1.0 ∗ 0.8 ∗ 0.8 ∗ 0.2 =0.128. Probabilistic Context Free Grammars have many uses in speech recognition [15], [16], natural language processing [18], computational biology =-=[17]-=-, sensor networks [4], [8], [10], [11] etc. III. PREVIOUS WORK In this section, we present a brief overview of literature on PCFG inference and application of PCFGs to sensor data. A. Event Recognitio... |

132 | Inducing probabilistic grammars by Bayesian model merging
- Stolcke, Omohundro
- 1994
(Show Context)
Citation Context ...], [13]. Constructing PCFGs from scratch, however, requires alternative learning techniques. There are two different approaches to this problem. The first one uses two operators: merge and chunk [2], =-=[14]-=-. This is the approach that forms the basis for our inference algorithm. This scheme utilizes Bayesian formulation to evaluate the grammar. The solution is obtained by maximizing a priori values for t... |

129 | Bayesian Learning of Probabilistic Language Models
- Stolcke
- 1994
(Show Context)
Citation Context ...y string) or under-generalization (a grammar accepting only the training data). To address this problem, we are presenting here a grammar induction method inspired by the scheme introduced by Stolcke =-=[2]-=-. There are two differences between his approach and ours. First, each method uses the different metric for evaluating the grammar at each step of its construction. Second, we are using a novel method... |

122 |
An Efficient Recognition and Syntax-analysis Algorithm for Contextfree Languages
- Kasami
- 1965
(Show Context)
Citation Context ...d different parsing possibilities and remove the ones that fail as the parsing progresses. Please note that parsing is a quite light task that can be done in O(l 3 ) (Earley-Stolcke[2], [9], CYK[19], =-=[20]-=-, [21]) where l is the length of the string to be parsed and thus much smaller than D. Fig.12 presents a simple version of the deployment scheme proposed above. Sense Event Data Parse Events Sensor No... |

79 |
Programming Languages and Their Compilers: Preliminary Notes
- Cocke, Schwartz
- 1970
(Show Context)
Citation Context ...ld hold different parsing possibilities and remove the ones that fail as the parsing progresses. Please note that parsing is a quite light task that can be done in O(l 3 ) (Earley-Stolcke[2], [9], CYK=-=[19]-=-, [20], [21]) where l is the length of the string to be parsed and thus much smaller than D. Fig.12 presents a simple version of the deployment scheme proposed above. Sense Event Data Parse Events Sen... |

75 | Probabilistic Top-Down Parsing and Language Modeling
- Roark
- 2001
(Show Context)
Citation Context ... of theparsing tree of that sentence. In Fig. 1, the string aaab bbhas the probability of 1.0 ∗ 0.8 ∗ 0.8 ∗ 0.2 =0.128. Probabilistic Context Free Grammars have many uses in speech recognition [15], =-=[16]-=-, natural language processing [18], computational biology [17], sensor networks [4], [8], [10], [11] etc. III. PREVIOUS WORK In this section, we present a brief overview of literature on PCFG inferenc... |

66 | Building Probabilistic Models for Natural Language
- Chen
- 1996
(Show Context)
Citation Context .... Evaluation Metric for the PCFG Our goal is to find a grammar G that maximizes a posteriori probability given the training data O = {o1,o2,...}. Hence, we want to compute the grammar defined as [2], =-=[3]-=-: G = argmaxGP (G|O). (2) Using Bayes’ formula, we obtain the following expansion of Eq. (2). P (G)P (O|G) G = argmaxG = argmaxGP (G)P (O|G) P (O) where P (G) is the grammar a priori probability which... |

58 |
Recognition and parsing of context-free languages
- Younger
- 1967
(Show Context)
Citation Context ...erent parsing possibilities and remove the ones that fail as the parsing progresses. Please note that parsing is a quite light task that can be done in O(l 3 ) (Earley-Stolcke[2], [9], CYK[19], [20], =-=[21]-=-) where l is the length of the string to be parsed and thus much smaller than D. Fig.12 presents a simple version of the deployment scheme proposed above. Sense Event Data Parse Events Sensor Node Sen... |

44 | Recognizing Multitasked Activities from Video Using Stochastic Context-Free Grammar
- Moore, Essa
(Show Context)
Citation Context ... ∗ 0.8 ∗ 0.8 ∗ 0.2 =0.128. Probabilistic Context Free Grammars have many uses in speech recognition [15], [16], natural language processing [18], computational biology [17], sensor networks [4], [8], =-=[10]-=-, [11] etc. III. PREVIOUS WORK In this section, we present a brief overview of literature on PCFG inference and application of PCFGs to sensor data. A. Event Recognition using PCFGs There are several ... |

35 | Using a stochastic context-free grammar as a language model for speech recognition
- Jurafsky, Wooters, et al.
- 1995
(Show Context)
Citation Context ...anches of theparsing tree of that sentence. In Fig. 1, the string aaab bbhas the probability of 1.0 ∗ 0.8 ∗ 0.8 ∗ 0.2 =0.128. Probabilistic Context Free Grammars have many uses in speech recognition =-=[15]-=-, [16], natural language processing [18], computational biology [17], sensor networks [4], [8], [10], [11] etc. III. PREVIOUS WORK In this section, we present a brief overview of literature on PCFG in... |

21 | AddressEvent Imagers for Sensor Networks: Evaluation and Modeling
- Teixeira, Culurciello, et al.
- 2006
(Show Context)
Citation Context ...work assigns a probability to different combinations of events constituting certain behavior. The paper also presents a way to calculate the probability of an event for a given set of actions. Papers =-=[5]-=- and [6] complement each other. The first one introduces Address-Event Imagers for sensor nodes. These are image sensors, capabilities of which are limited to achieve efficiency in power, computation ... |

20 | A Lightweight Camera Sensor Network Operating on Symbolic Information
- Teixeira, Lymberopoulos, et al.
- 2006
(Show Context)
Citation Context ...igns a probability to different combinations of events constituting certain behavior. The paper also presents a way to calculate the probability of an event for a given set of actions. Papers [5] and =-=[6]-=- complement each other. The first one introduces Address-Event Imagers for sensor nodes. These are image sensors, capabilities of which are limited to achieve efficiency in power, computation cost, st... |

9 | Deleted interpolation using a hierarchical bayesian grammar network for recognizing human activity
- Kitani, Sato, et al.
- 2005
(Show Context)
Citation Context ... ∗ 0.8 ∗ 0.2 =0.128. Probabilistic Context Free Grammars have many uses in speech recognition [15], [16], natural language processing [18], computational biology [17], sensor networks [4], [8], [10], =-=[11]-=- etc. III. PREVIOUS WORK In this section, we present a brief overview of literature on PCFG inference and application of PCFGs to sensor data. A. Event Recognition using PCFGs There are several papers... |

3 |
A Sensory Grammar for Inferring Behaviors
- Lymberopoulos, Ogale, et al.
- 2006
(Show Context)
Citation Context ...ity of 1.0 ∗ 0.8 ∗ 0.8 ∗ 0.2 =0.128. Probabilistic Context Free Grammars have many uses in speech recognition [15], [16], natural language processing [18], computational biology [17], sensor networks =-=[4]-=-, [8], [10], [11] etc. III. PREVIOUS WORK In this section, we present a brief overview of literature on PCFG inference and application of PCFGs to sensor data. A. Event Recognition using PCFGs There a... |

1 |
An Easy to Program Sensor System for Parsing Human Activities, ENALAB
- Lymberopoulos, Barton-Sweeney, et al.
- 2006
(Show Context)
Citation Context ... cooking differs from cleaning the dishes in the kitchen in terms of positions of the agent and how this difference can be captured by PCFGs. Following the methodology of [6], the authors describe in =-=[7]-=- a human behavior parsing sensor system from start to end. The paper discusses a context free grammar that distinguishes cooking from cleaning and from other kitchen activities. Experimental results a... |