By Topic

Grammatical Inference: Theory, Applications and Alternatives, IEE Colloquium on

Date 22-23 April 1993

Filter Results

Displaying Results 1 - 25 of 33
  • New directions in grammatical inference

    Publication Year: 1993 , Page(s): 1/1 - 1/7
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (2045 KB)  

    An introduction is given to grammatical inference, theory, application and alternatives. Grammatical interference is an important field of research, and it has suffered from the lack of a focused research community. The author discusses what sort of grammars there are and what they can do, then provides a discussion of the types of grammatical inference algorithm that have been successfully employ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Parse tree n-grams for spoken language modelling

    Publication Year: 1993 , Page(s): 26/1 - 26/6
    Cited by:  Patents (2)
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (300 KB)  

    A method is described for modelling natural language for speech recognition. Its aim is to incorporate the advantages of two previous types of approach; the statistical approach and the formal linguistic approach. The n-gram model (J.K. Baker, 1975), is a statistical model based on corpus data, and is used to estimate the probability of an unseen sequence of words based on this data. A phrase stru... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Adaptive statistical and grammar models of language for application to speech recognition

    Publication Year: 1993 , Page(s): 25/1 - 25/8
    Cited by:  Patents (1)
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (372 KB)  

    The statistical and syntactic approaches to the modelling of language are consolidated in order to improve performance in speech recognition. The authors also aim to minimise the need for human intervention in the training of the language model from a corpus. Hybrid speech recognition systems using both bigram and grammar models can yield improved performance compared with the use of either model ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Grammatical inference based on hyperedge replacement: a summary

    Publication Year: 1993 , Page(s): 7/1 - 7/6
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (388 KB)  

    As a fundamental goal of pattern generation and recognition, one wants to find syntactic descriptions for certain types or sets of patterns in such a way that automatic generation or recognition of the patterns is provided. The types or sets of patterns, that are of interest, may not be known completely, but only by some samples. A grammatical-inference algorithm is introduced which constructs hyp... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Integrating segmentation and recognition in on-line cursive handwriting using error-correcting grammars

    Publication Year: 1993 , Page(s): 23/1 - 23/6
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (388 KB)  

    A grammar-based approach is given to recognizing online cursive handwriting that overcomes the inherent problems of variability in input size and the need to integrate segmentation and recognition. The input stream is represented as a sequence of uniform stroke descriptions that are processed by a mixture of neural-networks, each designed to recognize letters of different sizes. Words are then rec... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Can grammar and complexity bridge the gap between psychology and biologically plausible neural networks?

    Publication Year: 1993 , Page(s): P1/1 - P1/7
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (548 KB)  

    There are two approaches to neural network modelling in biology. One can take an existing relatively well understood artificial neural network and determine how its behaviour is changed if it is made more consistent with the known anatomy and physiology of the brain. On the other hand, one can start from known biology and consider what types of useful behaviour such networks might produce. However... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Automatically acquiring and evaluating a classification of words

    Publication Year: 1993 , Page(s): P8/1 - P8/7
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (356 KB)  

    An investigation is made into how to automatically derive a meaningful classification of words from ordinary text on the assumption that words of similar role usually appear in similar contexts. The authors propose novel patterns in text which derive good contextual information. They describe methods by which the contextual information can be used to classify like words together. Methods to descri... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Language understanding and subsequential transducer learning

    Publication Year: 1993 , Page(s): 11/1 - 1110
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (720 KB)  

    The application of the Onward Subsequential Transducer Inference Algorithm (OSTIA) recently introduced by J. Oncina et al. (1993) to (pseudo-) natural language understanding is considered. For this purpose, a task proposed by J.A. Feldman et al. (1990), as a touchstone for comparing the capabilities of language learning systems has been adopted and three increasingly difficult semantic coding sche... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The application of k-testable languages in the strict sense to phone recognition in automatic speech recognition

    Publication Year: 1993 , Page(s): 22/1 - 22/7
    Cited by:  Papers (1)
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (564 KB)  

    The aim of the work is to show the feasibility and the appeal of characterizable methods of grammatical inference in establishing the required structural models for particular syntactic pattern recognition problems, so as to allow a satisfactory performance to be achieved. In particular, the authors have chosen the application area of automatic speech recognition. They describe two applications in... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A syntax based grammar of stress sequences

    Publication Year: 1993 , Page(s): P7/1 - P7/7
    Cited by:  Patents (1)
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (404 KB)  

    The Spoken English Corpus (SEC) is a speech corpus containing prosodic and syntactic annotations and provides an ideal source for the study of the relationship between prosody and syntax. It is shown that the placement of stresses on words in an utterance may be largely predicted from word classifications. The language model used to make these predictions is inferred from the SEC training set. How... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Context free grammar induction using genetic algorithms

    Publication Year: 1993 , Page(s): P11/1 - P11/5
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (388 KB)  

    A genetic algorithm was developed for the purpose of inferring context free grammars. Results are reported on the inference of two grammars in this class. Various forms of the grammar to generate the language of correctly balanced and nested brackets were successfully inferred, but more complex grammars were not learnt with the resources available. The author also discusses various issues such as ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Connectionism and natural language

    Publication Year: 1993 , Page(s): 20/1 - 2010
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (640 KB)  

    Neural nets, have been successfully applied to various aspects of grammatical inference. The authors consider the relationship between neural nets and grammatical inference whilst taking account of some of the constraints involved in modelling natural language acquisition. These constraints are engendered by attempts to pay attention to the characteristics of the actual data upon which language ac... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A new regular language learning algorithm from lexicographically ordered complete samples

    Publication Year: 1993 , Page(s): 6/1 - 6/7
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (400 KB)  

    A regular language learning algorithm is presented to obtain descriptions which consist of deterministic finite automata (DFAs). The process is an identification in the limit process. The main characteristic is that the DFAs are conjectured using a constructive strategy which does not use a large data space. The total time used is polynomial in the size of the minimum-state DFA and the data seen s... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Inference of stochastic regular languages through simple recurrent networks

    Publication Year: 1993 , Page(s): 16/1 - 16/6
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (392 KB)  

    Grammatical inference has been recently approached through artificial neural networks. Recurrent connectionist architectures were trained to accept or reject strings belonging to a number of specific regular languages, or to predict the possible successor(s) for each character in the string. On the other hand, for static (non-string) data, M.D. Richard et al. (1991), showed that a nonrecurrent arc... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Deriving a probabilistic grammar of semantic markers from unrestricted English text

    Publication Year: 1993 , Page(s): P9/1 - P9/7
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (368 KB)  

    The derivation is described of a probabilistic grammar for main subject field codes from the machine readable version of the Longman Dictionary of Contemporary English (LDOCE) (P. Procter, 1978). These codes are used in the dictionary to mark the subject area to which a certain sense of a word belongs. The grammar consists of the dictionary itself and a matrix that describes how closely two main s... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Learning context-free grammar with enhanced neural network pushdown automaton

    Publication Year: 1993 , Page(s): P6/1 - P613
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (904 KB)  

    Previously (C. Sreerupa Das et al., 1992, 1993; C.L. Giles et al., 1990; G.Z. Sun et al., 1990, 1991), a model was developed of neural network pushdown automata (NNPDA). NNPDA is a hybrid system which couples the neural network finite controller with an external continuous stack memory. It learns context-free grammars from examples by minimizing the properly defined objective function. In the orig... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Application of the error-correcting grammatical inference algorithm (ECGI) to planar shape recognition

    Publication Year: 1993 , Page(s): 24/1 - 2410
    Cited by:  Papers (2)
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (748 KB)  

    ECGI is an error-correcting-based learning technique that aims at obtaining structural finite-state models of (unidimensional) objects from samples of these objects. The learning procedure captures certain useful regularities of the training data in the object-models, while also obtaining appropriate models of the `irregularities' (errors and distortions) that these data tend to exhibit with respe... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Probabilistic dependency grammar, and its application in constructing language models for speech applications

    Publication Year: 1993 , Page(s): 12/1 - 1211
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (748 KB)  

    Dependency grammar defines grammatical notions in terms of direct links between words at a lexical level. These lexical dependency relations are considered more fundamental than other representations, such as phrase structure analyses, which can be derived from them. In simple cases there is a direct correspondence between dependencies and phrase structures. The dependencies for a given word can b... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Regular inference with maximal valid grammar method

    Publication Year: 1993 , Page(s): 5/1 - 5/9
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (496 KB)  

    The aim of grammatical inference is defined from a new point of view. The maximal valid grammar method is proposed to infer context free grammars from structural positive samples using negative samples. A polynomial algorithm is given. The MVG method is applied on the class of regular grammars View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Learning a class of regular languages in the probably approximately correct learnability framework of Valiant

    Publication Year: 1993 , Page(s): 2/1 - 216
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (780 KB)  

    Results are given, relating to the probably approximately correct (M.A. Harrison, 1978) learning of a class of regular languages called terminal distinguishable regular languages. The authors prove that the VC-dimension of this concept class is infinite. However, when further restriction is imposed on the length and the structure of strings this class is found to have finite VC-dimension that grow... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Learning, extracting, inserting and verifying grammatical information in recurrent neural networks

    Publication Year: 1993
    Cited by:  Patents (1)
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (96 KB)  

    Recurrent neural networks can be trained from string examples to behave like deterministic finite-state automata (DFA's) and pushdown automata (adapts) i.e. they recognize respectively deterministic regular and context-free grammars (DCFG's). The author discusses some of the successes and failures of this type of `recurrent neural network' grammatical inference engine, as well as some of the issue... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Viewing grammar induction as an adaptive control problem

    Publication Year: 1993 , Page(s): P2/1 - P2/6
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (484 KB)  

    Adaptive control is the design of systems that control an autonomous agent in a changing and uncertain environment. The author argues that many types of real world tasks in which grammar induction is a goal may be viewed in the framework of existing adaptive control techniques and that much is to be gained by doing so. Background on adaptive control is presented, focusing on one algorithm, C.J.C.H... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Analysis of a simple bipos language model-attempt at a strategy to improve language models for speech recognition

    Publication Year: 1993 , Page(s): P10/1 - P10/8
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (540 KB)  

    A speech recognizer has to choose, at each point in the utterance, the words among all the words in the vocabulary, that are the most likely. To that end, it uses an acoustic model and a language model and the author focuses on the language model. The bipos model is presented and analysed. A method is introduced called probability decomposition to measure which part of the model is performing part... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Algebraic grammatical inference

    Publication Year: 1993 , Page(s): P12/1 - P1210
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (656 KB)  

    A description is given of an approach to grammatical inference. Given both a positive and negative sample of the language to be learned, each string is mapped into a multi-variate polynomial expression. These expressions are each given values depending on their membership of the language. If we are inferring a non-stochastic grammar, then each expression derived from a positive string must equate ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Inductive and deductive learning of grammar: dealing with incomplete theories

    Publication Year: 1993 , Page(s): 13/1 - 1310
    Save to Project icon | Click to expandAbstract | PDF file iconPDF (424 KB)  

    A framework is given for learning plausible unification-based natural language grammars. The authors assume that the system will have some initial unification-based grammar and they use learning to overcome the incompleteness of this grammar (its undergeneration, i.e. where it fails to generate strings that humans would regard as grammatical). The authors use both model-driven (deductive) and data... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.