id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
cmp-lg/9807006
A Maximum-Entropy Partial Parser for Unrestricted Text
cmp-lg cs.CL
This paper describes a partial parser that assigns syntactic structures to sequences of part-of-speech tags. The program uses the maximum entropy parameter estimation method, which allows a flexible combination of different knowledge sources: the hierarchical structure, parts of speech and phrasal categories. In effect, the parser goes beyond simple bracketing and recognises even fairly complex structures. We give accuracy figures for different applications of the parser.
cmp-lg/9807007
Chunk Tagger - Statistical Recognition of Noun Phrases
cmp-lg cs.CL
We describe a stochastic approach to partial parsing, i.e., the recognition of syntactic structures of limited depth. The technique utilises Markov Models, but goes beyond usual bracketing approaches, since it is capable of recognising not only the boundaries, but also the internal structure and syntactic category of simple as well as complex NP's, PP's, AP's and adverbials. We compare tagging accuracy for different applications and encoding schemes.
cmp-lg/9807008
A Linguistically Interpreted Corpus of German Newspaper Text
cmp-lg cs.CL
In this paper, we report on the development of an annotation scheme and annotation tools for unrestricted German text. Our representation format is based on argument structure, but also permits the extraction of other kinds of representations. We discuss several methodological issues and the analysis of some phenomena. Additional focus is on the tools developed in our project and their applications.
cmp-lg/9807009
A Projection Architecture for Dependency Grammar and How it Compares to LFG
cmp-lg cs.CL
This paper explores commonalities and differences between \dachs, a variant of Dependency Grammar, and Lexical-Functional Grammar. \dachs\ is based on traditional linguistic insights, but on modern mathematical tools, aiming to integrate different knowledge systems (from syntax and semantics) via their coupling to an abstract syntactic primitive, the dependency relation. These knowledge systems correspond rather closely to projections in LFG. We will investigate commonalities arising from the usage of the projection approach in both theories, and point out differences due to the incompatible linguistic premises. The main difference to LFG lies in the motivation and status of the dimensions, and the information coded there. We will argue that LFG confounds different information in one projection, preventing it to achieve a good separation of alternatives and calling the motivation of the projection into question.
cmp-lg/9807010
Automatically Creating Bilingual Lexicons for Machine Translation from Bilingual Text
cmp-lg cs.CL
A method is presented for automatically augmenting the bilingual lexicon of an existing Machine Translation system, by extracting bilingual entries from aligned bilingual text. The proposed method only relies on the resources already available in the MT system itself. It is based on the use of bilingual lexical templates to match the terminal symbols in the parses of the aligned sentences.
cmp-lg/9807011
Statistical Models for Unsupervised Prepositional Phrase Attachment
cmp-lg cs.CL
We present several unsupervised statistical models for the prepositional phrase attachment task that approach the accuracy of the best supervised methods for this task. Our unsupervised approach uses a heuristic based on attachment proximity and trains from raw text that is annotated with only part-of-speech tags and morphological base forms, as opposed to attachment information. It is therefore less resource-intensive and more portable than previous corpus-based algorithms proposed for this task. We present results for prepositional phrase attachment in both English and Spanish.
cmp-lg/9807012
Character design for soccer commmentary
cmp-lg cs.CL
In this paper we present early work on an animated talking head commentary system called {\bf Byrne}\footnote{David Byrne is the lead singer of the Talking Heads.}. The goal of this project is to develop a system which can take the output from the RoboCup soccer simulator, and generate appropriate affective speech and facial expressions, based on the character's personality, emotional state, and the state of play. Here we describe a system which takes pre-analysed simulator output as input, and which generates text marked-up for use by a speech generator and a face animation system. We make heavy use of inter-system standards, so that future versions of Byrne will be able to take advantage of advances in the technologies that it incorporates.
cmp-lg/9807013
Improving Data Driven Wordclass Tagging by System Combination
cmp-lg cs.CL
In this paper we examine how the differences in modelling between different data driven systems performing the same NLP task can be exploited to yield a higher accuracy than the best individual system. We do this by means of an experiment involving the task of morpho-syntactic wordclass tagging. Four well-known tagger generators (Hidden Markov Model, Memory-Based, Transformation Rules and Maximum Entropy) are trained on the same corpus data. After comparison, their outputs are combined using several voting strategies and second stage classifiers. All combination taggers outperform their best component, with the best combination showing a 19.1% lower error rate than the best individual tagger.
cmp-lg/9808001
An Empirical Evaluation of Probabilistic Lexicalized Tree Insertion Grammars
cmp-lg cs.CL
We present an empirical study of the applicability of Probabilistic Lexicalized Tree Insertion Grammars (PLTIG), a lexicalized counterpart to Probabilistic Context-Free Grammars (PCFG), to problems in stochastic natural-language processing. Comparing the performance of PLTIGs with non-hierarchical N-gram models and PCFGs, we show that PLTIG combines the best aspects of both, with language modeling capability comparable to N-grams, and improved parsing performance over its non-lexicalized counterpart. Furthermore, training of PLTIGs displays faster convergence than PCFGs.
cmp-lg/9808002
Indexing with WordNet synsets can improve Text Retrieval
cmp-lg cs.CL
The classical, vector space model for text retrieval is shown to give better results (up to 29% better in our experiments) if WordNet synsets are chosen as the indexing space, instead of word forms. This result is obtained for a manually disambiguated test collection (of queries and documents) derived from the Semcor semantic concordance. The sensitivity of retrieval performance to (automatic) disambiguation errors when indexing documents is also measured. Finally, it is observed that if queries are not disambiguated, indexing by synsets performs (at best) only as good as standard word indexing.
cmp-lg/9808003
Parallel Strands: A Preliminary Investigation into Mining the Web for Bilingual Text
cmp-lg cs.CL
Parallel corpora are a valuable resource for machine translation, but at present their availability and utility is limited by genre- and domain-specificity, licensing restrictions, and the basic difficulty of locating parallel texts in all but the most dominant of the world's languages. A parallel corpus resource not yet explored is the World Wide Web, which hosts an abundance of pages in parallel translation, offering a potential solution to some of these problems and unique opportunities of its own. This paper presents the necessary first step in that exploration: a method for automatically finding parallel translated documents on the Web. The technique is conceptually simple, fully language independent, and scalable, and preliminary evaluation results indicate that the method may be accurate enough to apply without human intervention.
cmp-lg/9808004
Word Length Frequency and Distribution in English: Observations, Theory, and Implications for the Construction of Verse Lines
cmp-lg cs.CL
Recent observations in the theory of verse and empirical metrics have suggested that constructing a verse line involves a pattern-matching search through a source text, and that the number of found elements (complete words totaling a specified number of syllables) is given by dividing the total number of words by the mean number of syllables per word in the source text. This paper makes this latter point explicit mathematically, and in the course of this demonstration shows that the word length frequency totals in English output are distributed geometrically (previous researchers reported an adjusted Poisson distribution), and that the sequential distribution is random at the global level, with significant non-randomness in the fine structure. Data from a corpus of just under two million words, and a syllable-count lexicon of 71,000 word-forms is reported. The pattern-matching theory is shown to be internally coherent, and it is observed that some of the analytic techniques described here form a satisfactory test for regular (isometric) lineation in a text.
cmp-lg/9808005
Combining Expression and Content in Domains for Dialog Managers
cmp-lg cs.CL
We present work in progress on abstracting dialog managers from their domain in order to implement a dialog manager development tool which takes (among other data) a domain description as input and delivers a new dialog manager for the described domain as output. Thereby we will focus on two topics; firstly, the construction of domain descriptions with description logics and secondly, the interpretation of utterances in a given domain.
cmp-lg/9808006
Isometric Lineation in English Texts: An Empirical and Mathematical Examination of its Character and Consequences
cmp-lg cs.CL
In this paper we build on earlier observations and theory regarding word length frequency and sequential distribution to develop a mathematical characterization of some of the language features distinguishing isometrically lineated text from unlineated text, in other words the features distinguishing isometrical verse from prose. It is shown that the frequency of syllables making complete words produces a flat distribution for prose, while that for verse exhibits peaks at the line length position and subsequent multiples of that position. Data from several verse authors is presented, including a detailed mathematical analysis of the dynamics underlying peak creation, and comments are offered on the processes by which authors construct lines. We note that the word-length sequence of prose is random, whereas lineation necessitates non-random word-length sequencing, and that this has the probable consequence of introducing a degree of randomness into the otherwise highly ordered grammatical sequence. In addition we observe that this effect can be ameliorated by a reduction in the mean word length of the text (confirming earlier observations that verse tends to use shorter words) and the use of lines varying from the core isometrical set. The frequency of variant lines is shown to be coincident with the frequency of polysyllables, suggesting that the use of variant lines is motivated by polysyllabic word placement. The restrictive effects of different line lengths, the relationship between metrical restriction and poetic effect, and the general character of metrical rules are also discussed.
cmp-lg/9808007
Some Properties of Preposition and Subordinate Conjunction Attachments
cmp-lg cs.CL
Determining the attachments of prepositions and subordinate conjunctions is a key problem in parsing natural language. This paper presents a trainable approach to making these attachments through transformation sequences and error-driven learning. Our approach is broad coverage, and accounts for roughly three times the attachment cases that have previously been handled by corpus-based techniques. In addition, our approach is based on a simplified model of syntax that is more consistent with the practice in current state-of-the-art language processing systems. This paper sketches syntactic and algorithmic details, and presents experimental results on data sets derived from the Penn Treebank. We obtain an attachment accuracy of 75.4% for the general case, the first such corpus-based result to be reported. For the restricted cases previously studied with corpus-based methods, our approach yields an accuracy comparable to current work (83.1%).
cmp-lg/9808008
Deriving the Predicate-Argument Structure for a Free Word Order Language
cmp-lg cs.CL
In relatively free word order languages, grammatical functions are intricately related to case marking. Assuming an ordered representation of the predicate-argument structure, this work proposes a Combinatory Categorial Grammar formulation of relating surface case cues to categories and types for correctly placing the arguments in the predicate-argument structure. This is achieved by assigning case markers GF-encoding categories. Unlike other CG formulations, type shifting does not proliferate or cause spurious ambiguity. Categories of all argument-encoding grammatical functions follow from the same principle of category assignment. Normal order evaluation of the combinatory form reveals the predicate-argument structure. Application of the method to Turkish is shown.
cmp-lg/9808009
How to define a context-free backbone for DGs: Implementing a DG in the LFG formalism
cmp-lg cs.CL
This paper presents a multidimensional Dependency Grammar (DG), which decouples the dependency tree from word order, such that surface ordering is not determined by traversing the dependency tree. We develop the notion of a \emph{word order domain structure}, which is linked but structurally dissimilar to the syntactic dependency tree. We then discuss the implementation of such a DG using constructs from a unification-based phrase-structure approach, namely Lexical-Functional Grammar (LFG). Particular attention is given to the analysis of discontinuities in DG in terms of LFG's functional uncertainty.
cmp-lg/9808010
Letter to Sound Rules for Accented Lexicon Compression
cmp-lg cs.CL
This paper presents trainable methods for generating letter to sound rules from a given lexicon for use in pronouncing out-of-vocabulary words and as a method for lexicon compression. As the relationship between a string of letters and a string of phonemes representing its pronunciation for many languages is not trivial, we discuss two alignment procedures, one fully automatic and one hand-seeded which produce reasonable alignments of letters to phones. Top Down Induction Tree models are trained on the aligned entries. We show how combined phoneme/stress prediction is better than separate prediction processes, and still better when including in the model the last phonemes transcribed and part of speech information. For the lexicons we have tested, our models have a word accuracy (including stress) of 78% for OALD, 62% for CMU and 94% for BRULEX. The extremely high scores on the training sets allow substantial size reductions (more than 1/20). WWW site: http://tcts.fpms.ac.be/synthesis/mbrdico
cmp-lg/9808011
Primitive Part-of-Speech Tagging using Word Length and Sentential Structure
cmp-lg cs.CL
It has been argued that, when learning a first language, babies use a series of small clues to aid recognition and comprehension, and that one of these clues is word length. In this paper we present a statistical part of speech tagger which trains itself solely on the number of letters in each word in a sentence.
cmp-lg/9808012
Separating Surface Order and Syntactic Relations in a Dependency Grammar
cmp-lg cs.CL
This paper proposes decoupling the dependency tree from word order, such that surface ordering is not determined by traversing the dependency tree. We develop the notion of a \emph{word order domain structure}, which is linked but structurally dissimilar to the syntactic dependency tree. The proposal results in a lexicalized, declarative, and formally precise description of word order; features which lack previous proposals for dependency grammars. Contrary to other lexicalized approaches to word order, our proposal does not require lexical ambiguities for ordering alternatives.
cmp-lg/9808013
Partial Evaluation for Efficient Access to Inheritance Lexicons
cmp-lg cs.CL
Multiple default inheritance formalisms for lexicons have attracted much interest in recent years. I propose a new efficient method to access such lexicons. After showing two basic strategies for lookup in inheritance lexicons, a compromise is developed which combines to a large degree (from a practical point of view) the advantages of both strategies and avoids their disadvantages. The method is a kind of (off-line) partial evaluation that makes a subset of inherited information explicit before using the lexicon. I identify the parts of a lexicon which should be evaluated, and show how partial evaluation works for inheritance lexicons. Finally, the theoretical results are confirmed by a complete implementation. Speedups by a factor of 10-100 are reached.
cmp-lg/9808014
Spotting Prosodic Boundaries in Continuous Speech in French
cmp-lg cs.CL
A radio speech corpus of 9mn has been prosodically marked by a phonetician expert, and non expert listeners. this corpus is large enough to train and test an automatic boundary spotting system, namely a time delay neural network fed with F0 values, vowels and pseudo-syllable durations. Results validate both prosodic marking and automatic spotting of prosodic events.
cmp-lg/9808015
Error-Driven Pruning of Treebank Grammars for Base Noun Phrase Identification
cmp-lg cs.CL
Finding simple, non-recursive, base noun phrases is an important subtask for many natural language processing applications. While previous empirical methods for base NP identification have been rather complex, this paper instead proposes a very simple algorithm that is tailored to the relative simplicity of the task. In particular, we present a corpus-based approach for finding base NPs by matching part-of-speech tag sequences. The training phase of the algorithm is based on two successful techniques: first the base NP grammar is read from a ``treebank'' corpus; then the grammar is improved by selecting rules with high ``benefit'' scores. Using this simple algorithm with a naive heuristic for matching rules, we achieve surprising accuracy in an evaluation on the Penn Treebank Wall Street Journal.
cmp-lg/9808016
Segregatory Coordination and Ellipsis in Text Generation
cmp-lg cs.CL
In this paper, we provide an account of how to generate sentences with coordination constructions from clause-sized semantic representations. An algorithm is developed to generate sentences with ellipsis, gapping, right-node-raising, and non-constituent coordination constructions. Various examples from linguistic literature will be used to demonstrate that the algorithm does its job well.
cmp-lg/9808017
A Variant of Earley Parsing
cmp-lg cs.CL
The Earley algorithm is a widely used parsing method in natural language processing applications. We introduce a variant of Earley parsing that is based on a ``delayed'' recognition of constituents. This allows us to start the recognition of a constituent only in cases in which all of its subconstituents have been found within the input string. This is particularly advantageous in several cases in which partial analysis of a constituent cannot be completed and in general in all cases of productions sharing some suffix of their right-hand sides (even for different left-hand side nonterminals). Although the two algorithms result in the same asymptotic time and space complexity, from a practical perspective our algorithm improves the time and space requirements of the original method, as shown by reported experimental results.
cmp-lg/9809001
Towards an implementable dependency grammar
cmp-lg cs.CL
The aim of this paper is to define a dependency grammar framework which is both linguistically motivated and computationally parsable. See the demo at http://www.conexor.fi/analysers.html#testing
cmp-lg/9809002
Some Ontological Principles for Designing Upper Level Lexical Resources
cmp-lg cs.CL
The purpose of this paper is to explore some semantic problems related to the use of linguistic ontologies in information systems, and to suggest some organizing principles aimed to solve such problems. The taxonomic structure of current ontologies is unfortunately quite complicated and hard to understand, especially for what concerns the upper levels. I will focus here on the problem of ISA overloading, which I believe is the main responsible of these difficulties. To this purpose, I will carefully analyze the ontological nature of the categories used in current upper-level structures, considering the necessity of splitting them according to more subtle distinctions or the opportunity of excluding them because of their limited organizational role.
cmp-lg/9809003
A Comparison of WordNet and Roget's Taxonomy for Measuring Semantic Similarity
cmp-lg cs.CL
This paper presents the results of using Roget's International Thesaurus as the taxonomy in a semantic similarity measurement task. Four similarity metrics were taken from the literature and applied to Roget's The experimental evaluation suggests that the traditional edge counting approach does surprisingly well (a correlation of r=0.88 with a benchmark set of human similarity judgements, with an upper bound of r=0.90 for human subjects performing the same task.)
cond-mat/0002331
From naive to sophisticated behavior in multiagents based financial market models
cond-mat.stat-mech cond-mat.dis-nn cs.CE nlin.AO physics.data-an q-fin.TR
We discuss the behavior of two magnitudes, physical complexity and mutual information function of the outcome of a model of heterogeneous, inductive rational agents inspired in the El Farol Bar problem and the Minority Game. The first is a measure rooted in Kolmogorov-Chaitin theory and the second one a measure related with information entropy of Shannon. We make extensive computer simulations, as result of which, we propose an ansatz for physical complexity and establish the dependence of exponent of that ansatz from the parameters of the model. We discuss the accuracy of our results and the relationship with the behavior of mutual information function as a measure of time correlations of agents choice.
cond-mat/0009165
Occam factors and model-independent Bayesian learning of continuous distributions
cond-mat cs.LG nlin.AO physics.data-an
Learning of a smooth but nonparametric probability density can be regularized using methods of Quantum Field Theory. We implement a field theoretic prior numerically, test its efficacy, and show that the data and the phase space factors arising from the integration over the model space determine the free parameter of the theory ("smoothness scale") self-consistently. This persists even for distributions that are atypical in the prior and is a step towards a model-independent theory for learning continuous distributions. Finally, we point out that a wrong parameterization of a model family may sometimes be advantageous for small data sets.
cond-mat/0010337
Optimization with Extremal Dynamics
cond-mat.stat-mech cs.NE math.OC
We explore a new general-purpose heuristic for finding high-quality solutions to hard optimization problems. The method, called extremal optimization, is inspired by self-organized criticality, a concept introduced to describe emergent complexity in physical systems. Extremal optimization successively replaces extremely undesirable variables of a single sub-optimal solution with new, random ones. Large fluctuations ensue, that efficiently explore many local optima. With only one adjustable parameter, the heuristic's performance has proven competitive with more elaborate methods, especially near phase transitions which are believed to coincide with the hardest instances. We use extremal optimization to elucidate the phase transition in the 3-coloring problem, and we provide independent confirmation of previously reported extrapolations for the ground-state energy of +-J spin glasses in d=3 and 4.
cond-mat/0104066
Beyond the Zipf-Mandelbrot law in quantitative linguistics
cond-mat.stat-mech cs.CL nlin.AO
In this paper the Zipf-Mandelbrot law is revisited in the context of linguistics. Despite its widespread popularity the Zipf--Mandelbrot law can only describe the statistical behaviour of a rather restricted fraction of the total number of words contained in some given corpus. In particular, we focus our attention on the important deviations that become statistically relevant as larger corpora are considered and that ultimately could be understood as salient features of the underlying complex process of language generation. Finally, it is shown that all the different observed regimes can be accurately encompassed within a single mathematical framework recently introduced by C. Tsallis.
cond-mat/0104214
Extremal Optimization for Graph Partitioning
cond-mat.stat-mech cs.NE math.OC
Extremal optimization is a new general-purpose method for approximating solutions to hard optimization problems. We study the method in detail by way of the NP-hard graph partitioning problem. We discuss the scaling behavior of extremal optimization, focusing on the convergence of the average run as a function of runtime and system size. The method has a single free parameter, which we determine numerically and justify using a simple argument. Our numerical results demonstrate that on random graphs, extremal optimization maintains consistent accuracy for increasing system sizes, with an approximation error decreasing over runtime roughly as a power law t^(-0.4). On geometrically structured graphs, the scaling of results from the average run suggests that these are far from optimal, with large fluctuations between individual trials. But when only the best runs are considered, results consistent with theoretical arguments are recovered.
cond-mat/0109121
Coordination of Decisions in a Spatial Agent Model
cond-mat.stat-mech cs.MA
For a binary choice problem, the spatial coordination of decisions in an agent community is investigated both analytically and by means of stochastic computer simulations. The individual decisions are based on different local information generated by the agents with a finite lifetime and disseminated in the system with a finite velocity. We derive critical parameters for the emergence of minorities and majorities of agents making opposite decisions and investigate their spatial organization. We find that dependent on two essential parameters describing the local impact and the spatial dissemination of information, either a definite stable minority/majority relation (single-attractor regime) or a broad range of possible values (multi-attractor regime) occurs. In the latter case, the outcome of the decision process becomes rather diverse and hard to predict, both with respect to the share of the majority and their spatial distribution. We further investigate how a dissemination of information on different time scales affects the outcome of the decision process. We find that a more ``efficient'' information exchange within a subpopulation provides a suitable way to stabilize their majority status and to reduce ``diversity'' and uncertainty in the decision process.
cond-mat/0109218
Entropic analysis of the role of words in literary texts
cond-mat.stat-mech cs.CL
Beyond the local constraints imposed by grammar, words concatenated in long sequences carrying a complex message show statistical regularities that may reflect their linguistic role in the message. In this paper, we perform a systematic statistical analysis of the use of words in literary English corpora. We show that there is a quantitative relation between the role of content words in literary English and the Shannon information entropy defined over an appropriate probability distribution. Without assuming any previous knowledge about the syntactic structure of language, we are able to cluster certain groups of words according to their specific role in the text.
cond-mat/0110165
Jamming Model for the Extremal Optimization Heuristic
cond-mat.stat-mech cs.NE physics.comp-ph
Extremal Optimization, a recently introduced meta-heuristic for hard optimization problems, is analyzed on a simple model of jamming. The model is motivated first by the problem of finding lowest energy configurations for a disordered spin system on a fixed-valence graph. The numerical results for the spin system exhibit the same phenomena found in all earlier studies of extremal optimization, and our analytical results for the model reproduce many of these features.
cond-mat/0201139
Long-range fractal correlations in literary corpora
cond-mat.stat-mech cs.CL nlin.AO
In this paper we analyse the fractal structure of long human-language records by mapping large samples of texts onto time series. The particular mapping set up in this work is inspired on linguistic basis in the sense that is retains {\em the word} as the fundamental unit of communication. The results confirm that beyond the short-range correlations resulting from syntactic rules acting at sentence level, long-range structures emerge in large written language samples that give rise to long-range correlations in the use of words.
cond-mat/0202190
Threshold Disorder as a Source of Diverse and Complex Behavior in Random Nets
cond-mat.dis-nn cs.NE q-bio.NC
We study the diversity of complex spatio-temporal patterns in the behavior of random synchronous asymmetric neural networks (RSANNs). Special attention is given to the impact of disordered threshold values on limit-cycle diversity and limit-cycle complexity in RSANNs which have `normal' thresholds by default. Surprisingly, RSANNs exhibit only a small repertoire of rather complex limit-cycle patterns when all parameters are fixed. This repertoire of complex patterns is also rather stable with respect to small parameter changes. These two unexpected results may generalize to the study of other complex systems. In order to reach beyond this seemingly-disabling `stable and small' aspect of the limit-cycle repertoire of RSANNs, we have found that if an RSANN has threshold disorder above a critical level, then there is a rapid increase of the size of the repertoire of patterns. The repertoire size initially follows a power-law function of the magnitude of the threshold disorder. As the disorder increases further, the limit-cycle patterns themselves become simpler until at a second critical level most of the limit cycles become simple fixed points. Nonetheless, for moderate changes in the threshold parameters, RSANNs are found to display specific features of behavior desired for rapidly-responding processing systems: accessibility to a large set of complex patterns.
cond-mat/0202383
Extended Comment on Language Trees and Zipping
cond-mat.stat-mech cs.CL cs.LG
This is the extended version of a Comment submitted to Physical Review Letters. I first point out the inappropriateness of publishing a Letter unrelated to physics. Next, I give experimental results showing that the technique used in the Letter is 3 times worse and 17 times slower than a simple baseline. And finally, I review the literature, showing that the ideas of the Letter are not novel. I conclude by suggesting that Physical Review Letters should not publish Letters unrelated to physics.
cond-mat/0203436
Entropy estimation of symbol sequences
cond-mat.stat-mech cs.CL cs.IT math.IT physics.data-an stat.ML
We discuss algorithms for estimating the Shannon entropy h of finite symbol sequences with long range correlations. In particular, we consider algorithms which estimate h from the code lengths produced by some compression algorithm. Our interest is in describing their convergence with sequence length, assuming no limits for the space and time complexities of the compression algorithms. A scaling law is proposed for extrapolation from finite sample lengths. This is applied to sequences of dynamical systems in non-trivial chaotic regimes, a 1-D cellular automaton, and to written English texts.
cond-mat/0203591
Anticorrelations and subdiffusion in financial systems
cond-mat.dis-nn cond-mat.stat-mech cs.CE q-fin.ST
Statistical dynamics of financial systems is investigated, based on a model of a randomly coupled equation system driven by a stochastic Langevin force. Anticorrelations of price returns, and subdiffusion of prices is found from the model, and and compared with those calculated from historical $/EURO exchange rates.
cond-mat/0208414
Winner-Relaxing Self-Organizing Maps
cond-mat.dis-nn cs.NE nlin.AO q-bio.NC
A new family of self-organizing maps, the Winner-Relaxing Kohonen Algorithm, is introduced as a generalization of a variant given by Kohonen in 1991. The magnification behaviour is calculated analytically. For the original variant a magnification exponent of 4/7 is derived; the generalized version allows to steer the magnification in the wide range from exponent 1/2 to 1 in the one-dimensional case, thus provides optimal mapping in the sense of information theory. The Winner Relaxing Algorithm requires minimal extra computations per learning step and is conveniently easy to implement.
cond-mat/0301307
Nonextensive statistical mechanics and economics
cond-mat.stat-mech cs.CE q-fin.ST
Ergodicity, this is to say, dynamics whose time averages coincide with ensemble averages, naturally leads to Boltzmann-Gibbs (BG) statistical mechanics, hence to standard thermodynamics. This formalism has been at the basis of an enormous success in describing, among others, the particular stationary state corresponding to thermal equilibrium. There are, however, vast classes of complex systems which accomodate quite badly, or even not at all, within the BG formalism. Such dynamical systems exhibit, in one way or another, nonergodic aspects. In order to be able to theoretically study at least some of these systems, a formalism was proposed 14 years ago, which is sometimes referred to as nonextensive statistical mechanics. We briefly introduce this formalism, its foundations and applications. Furthermore, we provide some bridging to important economical phenomena, such as option pricing, return and volume distributions observed in the financial markets, and the fascinating and ubiquitous concept of risk aversion. One may summarize the whole approach by saying that BG statistical mechanics is based on the entropy $S_{BG}=-k \sum_i p_i \ln p_i$, and typically provides {\it exponential laws} for describing stationary states and basic time-dependent phenomena, while nonextensive statistical mechanics is instead based on the entropic form $S_q=k(1-\sum_ip_i^q)/(q-1)$ (with $S_1=S_{BG}$), and typically provides, for the same type of description, (asymptotic) {\it power laws}.
cond-mat/0301459
Collectives for the Optimal Combination of Imperfect Objects
cond-mat.dis-nn cond-mat.stat-mech cs.MA nlin.AO
In this letter we summarize some recent theoretical work on the design of collectives, i.e., of systems containing many agents, each of which can be viewed as trying to maximize an associated private utility, where there is also a world utility rating the behavior of that overall system that the designer of the collective wishes to optimize. We then apply algorithms based on that work on a recently suggested testbed for such optimization problems (Challet & Johnson, PRL, vol 89, 028701 2002). This is the problem of finding the combination of imperfect nano-scale objects that results in the best aggregate object. We present experimental results showing that these algorithms outperform conventional methods by more than an order of magnitude in this domain.
cond-mat/0303089
Multiplicative point process as a model of trading activity
cond-mat.stat-mech cs.CE math.SP nlin.AO nlin.CD q-fin.TR
Signals consisting of a sequence of pulses show that inherent origin of the 1/f noise is a Brownian fluctuation of the average interevent time between subsequent pulses of the pulse sequence. In this paper we generalize the model of interevent time to reproduce a variety of self-affine time series exhibiting power spectral density S(f) scaling as a power of the frequency f. Furthermore, we analyze the relation between the power-law correlations and the origin of the power-law probability distribution of the signal intensity. We introduce a stochastic multiplicative model for the time intervals between point events and analyze the statistical properties of the signal analytically and numerically. Such model system exhibits power-law spectral density S(f)~1/f**beta for various values of beta, including beta=1/2, 1 and 3/2. Explicit expressions for the power spectra in the low frequency limit and for the distribution density of the interevent time are obtained. The counting statistics of the events is analyzed analytically and numerically, as well. The specific interest of our analysis is related with the financial markets, where long-range correlations of price fluctuations largely depend on the number of transactions. We analyze the spectral density and counting statistics of the number of transactions. The model reproduces spectral properties of the real markets and explains the mechanism of power-law distribution of trading activity. The study provides evidence that the statistical properties of the financial markets are enclosed in the statistics of the time interval between trades. A multiplicative point process serves as a consistent model generating this statistics.
cond-mat/0304132
Causalities of the Taiwan Stock Market
cond-mat.stat-mech cs.CE q-fin.ST
Volatility, fitting with first order Landau expansion, stationarity, and causality of the Taiwan stock market (TAIEX) are investigated based on daily records. Instead of consensuses that consider stock market index change as a random time series we propose the market change as a dual time series consists of the index and the corresponding volume. Therefore, causalities between these two time series are investigated.
cond-mat/0305508
Neural network modeling of data with gaps: method of principal curves, Carleman's formula, and other
cond-mat.dis-nn cs.NE physics.data-an
A method of modeling data with gaps by a sequence of curves has been developed. The new method is a generalization of iterative construction of singular expansion of matrices with gaps. Under discussion are three versions of the method featuring clear physical interpretation: linear - modeling the data by a sequence of linear manifolds of small dimension; quasilinear - constructing "principal curves: (or "principal surfaces"), univalently projected on the linear principal components; essentially non-linear - based on constructing "principal curves": (principal strings and beams) employing the variation principle; the iteration implementation of this method is close to Kohonen self-organizing maps. The derived dependencies are extrapolated by Carleman's formulas. The method is interpreted as a construction of neural network conveyor designed to solve the following problems: to fill gaps in data; to repair data - to correct initial data values in such a way as to make the constructed models work best; to construct a calculator to fill gaps in the data line fed to the input.
cond-mat/0305527
Back-propagation of accuracy
cond-mat.dis-nn cs.NA cs.NE math.NA
In this paper we solve the problem: how to determine maximal allowable errors, possible for signals and parameters of each element of a network proceeding from the condition that the vector of output signals of the network should be calculated with given accuracy? "Back-propagation of accuracy" is developed to solve this problem. The calculation of allowable errors for each element of network by back-propagation of accuracy is surprisingly similar to a back-propagation of error, because it is the backward signals motion, but at the same time it is very different because the new rules of signals transformation in the passing back through the elements are different. The method allows us to formulate the requirements to the accuracy of calculations and to the realization of technical devices, if the requirements to the accuracy of output signals of the network are known.
cond-mat/0305681
Seven clusters in genomic triplet distributions
cond-mat.dis-nn cs.CV physics.bio-ph physics.data-an q-bio.GN
In several recent papers new gene-detection algorithms were proposed for detecting protein-coding regions without requiring learning dataset of already known genes. The fact that unsupervised gene-detection is possible closely connected to existence of a cluster structure in oligomer frequency distributions. In this paper we study cluster structure of several genomes in the space of their triplet frequencies, using pure data exploration strategy. Several complete genomic sequences were analyzed, using visualization of tables of triplet frequencies in a sliding window. The distribution of 64-dimensional vectors of triplet frequencies displays a well-detectable cluster structure. The structure was found to consist of seven clusters, corresponding to protein-coding information in three possible phases in one of the two complementary strands and in the non-coding regions with high accuracy (higher than 90% on the nucleotide level). Visualizing and understanding the structure allows to analyze effectively performance of different gene-prediction tools. Since the method does not require extraction of ORFs, it can be applied even for unassembled genomes. The information content of the triplet distributions and the validity of the mean-field models are analysed.
cond-mat/0307083
Generation of Explicit Knowledge from Empirical Data through Pruning of Trainable Neural Networks
cond-mat cs.NE physics.data-an
This paper presents a generalized technology of extraction of explicit knowledge from data. The main ideas are 1) maximal reduction of network complexity (not only removal of neurons or synapses, but removal all the unnecessary elements and signals and reduction of the complexity of elements), 2) using of adjustable and flexible pruning process (the pruning sequence shouldn't be predetermined - the user should have a possibility to prune network on his own way in order to achieve a desired network structure for the purpose of extraction of rules of desired type and form), and 3) extraction of rules not in predetermined but any desired form. Some considerations and notes about network architecture and training process and applicability of currently developed pruning techniques and rule extraction algorithms are discussed. This technology, being developed by us for more than 10 years, allowed us to create dozens of knowledge-based expert systems. In this paper we present a generalized three-step technology of extraction of explicit knowledge from empirical data.
cond-mat/0307630
Product Distribution Field Theory
cond-mat.stat-mech cond-mat.dis-nn cs.MA nlin.AO
This paper presents a novel way to approximate a distribution governing a system of coupled particles with a product of independent distributions. The approach is an extension of mean field theory that allows the independent distributions to live in a different space from the system, and thereby capture statistical dependencies in that system. It also allows different Hamiltonians for each independent distribution, to facilitate Monte Carlo estimation of those distributions. The approach leads to a novel energy-minimization algorithm in which each coordinate Monte Carlo estimates an associated spectrum, and then independently sets its state by sampling a Boltzmann distribution across that spectrum. It can also be used for high-dimensional numerical integration, (constrained) combinatorial optimization, and adaptive distributed control. This approach also provides a simple, physics-based derivation of the powerful approximate energy-minimization algorithms semi-formally derived in \cite{wowh00, wotu02c, wolp03a}. In addition it suggests many improvements to those algorithms, and motivates a new (bounded rationality) game theory equilibrium concept.
cond-mat/0312019
A theoretical investigation of ferromagnetic tunnel junctions with 4-valued conductances
cond-mat.mes-hall cond-mat.mtrl-sci cs.CE physics.ins-det quant-ph
In considering a novel function in ferromagnetic tunnel junctions consisting of ferromagnet(FM)/barrier/FM junctions, we theoretically investigate multiple valued (or multi-level) cell property, which is in principle realized by sensing conductances of four states recorded with magnetization configurations of two FMs; that is, (up,up), (up,down), (down,up), (down,down). To obtain such 4-valued conductances, we propose FM1/spin-polarized barrier/FM2 junctions, where the FM1 and FM2 are different ferromagnets, and the barrier has spin dependence. The proposed idea is applied to the case of the barrier having localized spins. Assuming that all the localized spins are pinned parallel to magnetization axes of the FM1 and FM2, 4-valued conductances are explicitly obtained for the case of many localized spins. Furthermore, objectives for an ideal spin-polarized barrier are discussed.
cond-mat/0402508
Information Theory - The Bridge Connecting Bounded Rational Game Theory and Statistical Physics
cond-mat.stat-mech cond-mat.dis-nn cs.GT cs.MA nlin.AO
A long-running difficulty with conventional game theory has been how to modify it to accommodate the bounded rationality of all real-world players. A recurring issue in statistical physics is how best to approximate joint probability distributions with decoupled (and therefore far more tractable) distributions. This paper shows that the same information theoretic mathematical structure, known as Product Distribution (PD) theory, addresses both issues. In this, PD theory not only provides a principled formulation of bounded rationality and a set of new types of mean field theory in statistical physics. It also shows that those topics are fundamentally one and the same.
cond-mat/0402581
Dictionary based methods for information extraction
cond-mat.stat-mech cond-mat.other cs.IR q-bio.GN q-bio.OT
In this paper we present a general method for information extraction that exploits the features of data compression techniques. We first define and focus our attention on the so-called "dictionary" of a sequence. Dictionaries are intrinsically interesting and a study of their features can be of great usefulness to investigate the properties of the sequences they have been extracted from (e.g. DNA strings). We then describe a procedure of string comparison between dictionary-created sequences (or "artificial texts") that gives very good results in several contexts. We finally present some results on self-consistent classification problems.
cond-mat/0403233
Artificial Sequences and Complexity Measures
cond-mat.stat-mech cs.CL cs.IR cs.IT math.IT
In this paper we exploit concepts of information theory to address the fundamental problem of identifying and defining the most suitable tools to extract, in a automatic and agnostic way, information from a generic string of characters. We introduce in particular a class of methods which use in a crucial way data compression techniques in order to define a measure of remoteness and distance between pairs of sequences of characters (e.g. texts) based on their relative information content. We also discuss in detail how specific features of data compression techniques could be used to introduce the notion of dictionary of a given sequence and of Artificial Text and we show how these new tools can be used for information extraction purposes. We point out the versatility and generality of our method that applies to any kind of corpora of character strings independently of the type of coding behind them. We consider as a case study linguistic motivated problems and we present results for automatic language recognition, authorship attribution and self consistent-classification.
cond-mat/0407439
A Theoretical Study on Spin-Dependent Transport of "Ferromagnet/Carbon Nanotube Encapsulating Magnetic Atoms/Ferromagnet" Junctions with 4-Valued Conductances
cond-mat.mes-hall cond-mat.mtrl-sci cs.CE physics.chem-ph
As a novel function of ferromagnet (FM)/spacer/FM junctions, we theoretically investigate multiple-valued (or multi-level) cell property, which is in principle realized by sensing conductances of four states recorded with magnetization configurations of two FMs; (up,up), (up,down), (down,up), (down,down). In order to sense all the states, 4-valued conductances corresponding to the respective states are necessary. We previously proposed that 4-valued conductances are obtained in FM1/spin-polarized spacer (SPS)/FM2 junctions, where FM1 and FM2 have different spin polarizations, and the spacer depends on spin [J. Phys.: Condens. Matter 15, 8797 (2003)]. In this paper, an ideal SPS is considered as a single-wall armchair carbon nanotube encapsulating magnetic atoms, where the nanotube shows on-resonance or off-resonance at the Fermi level according to its length. The magnitude of the obtained 4-valued conductances has an opposite order between the on-resonant nanotube and the off-resonant one, and this property can be understood by considering electronic states of the nanotube. Also, the magnetoresistance ratio between (up,up) and (down,down) can be larger than the conventional one between parallel and anti-parallel configurations.
cond-mat/0408190
From spin glasses to hard satisfiable formulas
cond-mat.stat-mech cond-mat.dis-nn cs.AI
We introduce a highly structured family of hard satisfiable 3-SAT formulas corresponding to an ordered spin-glass model from statistical physics. This model has provably "glassy" behavior; that is, it has many local optima with large energy barriers between them, so that local search algorithms get stuck and have difficulty finding the true ground state, i.e., the unique satisfying assignment. We test the hardness of our formulas with two Davis-Putnam solvers, Satz and zChaff, the recently introduced Survey Propagation (SP), and two local search algorithms, Walksat and Record-to-Record Travel (RRT). We compare our formulas to random 3-XOR-SAT formulas and to two other generators of hard satisfiable instances, the minimum disagreement parity formulas of Crawford et al., and Hirsch's hgen. For the complete solvers the running time of our formulas grows exponentially in sqrt(n), and exceeds that of random 3-XOR-SAT formulas for small problem sizes. SP is unable to solve our formulas with as few as 25 variables. For Walksat, our formulas appear to be harder than any other known generator of satisfiable instances. Finally, our formulas can be solved efficiently by RRT but only if the parameter d is tuned to the height of the barriers between local minima, and we use this parameter to measure the barrier heights in random 3-XOR-SAT formulas as well.
cond-mat/0410270
On uniqueness theorems for Tsallis entropy and Tsallis relative entropy
cond-mat.stat-mech cs.IT math.IT
The uniqueness theorem for Tsallis entropy was presented in {\it H.Suyari, IEEE Trans. Inform. Theory, Vol.50, pp.1783-1787 (2004)} by introducing the generalized Shannon-Khinchin's axiom. In the present paper, this result is generalized and simplified as follows: {\it Generalization}: The uniqueness theorem for Tsallis relative entropy is shown by means of the generalized Hobson's axiom. {\it Simplification}: The uniqueness theorem for Tsallis entropy is shown by means of the generalized Faddeev's axiom.
cond-mat/0410271
A generalized Faddeev's axiom and the uniqueness theorem for Tsallis entropy
cond-mat.stat-mech cs.IT math.IT
The uniequness theorem for the Tsallis entropy by introducing the generalized Faddeev's axiom is proven. Our result improves the recent result, the uniqueness theorem for Tsallis entropy by the generalized Shannon-Khinchin's axiom in \cite{Suy}, in the sence that our axiom is simpler than his one, as similar that Faddeev's axiom is simpler than Shannon-Khinchin's one.
cond-mat/0410460
A Computational Study of Rotating Spiral Waves and Spatio-Temporal Transient Chaos in a Deterministic Three-Level Active System
cond-mat.other cs.NE nlin.CG
Spatio-temporal dynamics of a deterministic three-level cellular automaton (TLCA) of Zykov-Mikhailov type (Sov. Phys. - Dokl., 1986, Vol.31, No.1, P.51) is studied numerically. Evolution of spatial structures is investigated both for the original Zykov-Mikhailov model (which is applicable to, for example, Belousov-Zhabotinskii chemical reactions) and for proposed by us TLCA, which is a generalization of Zykov-Mikhailov model for the case of two-channel diffusion. Such the TLCA is a minimal model for an excitable medium of microwave phonon laser, called phaser (D. N. Makovetskii, Tech. Phys., 2004, Vol.49, No.2, P.224; cond-mat/0402640). The most interesting observed forms of TLCA dynamics are as follows: (a) spatio-temporal transient chaos in form of highly bottlenecked collective evolution of excitations by rotating spiral waves (RSW) with variable topological charges; (b) competition of left-handed and right-handed RSW with unexpected features, including self-induced alteration of integral effective topological charge; (c) transient chimera states, i.e. coexistence of regular and chaotic domains in TLCA patterns; (d) branching of TLCA states with different symmetry which may lead to full restoring of symmetry of imperfect starting pattern. Phenomena (a) and (c) are directly related to phaser dynamics features observed earlier in real experiments at liquid helium temperatures on corundum crystals doped by iron-group ions. ACM: F.1.1, I.6, J.2; PACS:05.65.+b, 07.05.Tp, 82.20.Wt
cond-mat/0410594
A model of student's dilemma
cond-mat.other cond-mat.stat-mech cs.MA physics.soc-ph
Each year perhaps millions of young people face the following dilemma: should I continue my education or rather start working with already acquired skills. Right decision must take into account somebody's own abilities, accessibility to education institutions, competition, and potential benefits. A multi-agent, evolutionary model of this dilemma predicts a transition between stratified and homogeneous phases, evolution that diminishes fitness, fewer applicants per seat for decreased capacity of the university, and presence of poor students at \'elite universities.
cond-mat/0412460
Exact Maxwell-Boltzmann, Bose-Einstein and Fermi-Dirac Statistics
cond-mat.stat-mech cs.IT math.IT quant-ph
The exact Maxwell-Boltzmann (MB), Bose-Einstein (BE) and Fermi-Dirac (FD) entropies and probabilistic distributions are derived by the combinatorial method of Boltzmann, without Stirling's approximation. The new entropy measures are explicit functions of the probability and degeneracy of each state, and the total number of entities, N. By analysis of the cost of a "binary decision", exact BE and FD statistics are shown to have profound consequences for the behaviour of quantum mechanical systems.
cond-mat/0412587
Spin dependent transport of ``nonmagnetic metal/zigzag nanotube encapsulating magnetic atoms/nonmagnetic metal'' junctions
cond-mat.mes-hall cond-mat.mtrl-sci cs.CE physics.chem-ph quant-ph
Towards a novel magnetoresistance (MR) device with a carbon nanotube, we propose ``nonmagnetic metal/zigzag nanotube encapsulating magnetic atoms/nonmagnetic metal'' junctions. We theoretically investigate how spin-polarized edges of the nanotube and the encapsulated magnetic atoms influence on transport. When the on-site Coulomb energy divided by the magnitude of transfer integral, $U/|t|$, is larger than 0.8, large MR effect due to the direction of spins of magnetic atoms, which has the magnitude of the MR ratio of about 100%, appears reflecting such spin-polarized edges.
cond-mat/0412723
Modelling financial markets by the multiplicative sequence of trades
cond-mat.stat-mech cs.CE math.SP physics.data-an q-fin.ST
We introduce the stochastic multiplicative point process modelling trading activity of financial markets. Such a model system exhibits power-law spectral density S(f) ~ 1/f**beta, scaled as power of frequency for various values of beta between 0.5 and 2. Furthermore, we analyze the relation between the power-law autocorrelations and the origin of the power-law probability distribution of the trading activity. The model reproduces the spectral properties of trading activity and explains the mechanism of power-law distribution in real markets.
cond-mat/0504025
Point process model of 1/f noise versus a sum of Lorentzians
cond-mat.stat-mech astro-ph cond-mat.dis-nn cs.CE math.ST nlin.AO physics.data-an q-bio.NC stat.TH
We present a simple point process model of $1/f^{\beta}$ noise, covering different values of the exponent $\beta$. The signal of the model consists of pulses or events. The interpulse, interevent, interarrival, recurrence or waiting times of the signal are described by the general Langevin equation with the multiplicative noise and stochastically diffuse in some interval resulting in the power-law distribution. Our model is free from the requirement of a wide distribution of relaxation times and from the power-law forms of the pulses. It contains only one relaxation rate and yields $1/f^ {\beta}$ spectra in a wide range of frequency. We obtain explicit expressions for the power spectra and present numerical illustrations of the model. Further we analyze the relation of the point process model of $1/f$ noise with the Bernamont-Surdin-McWhorter model, representing the signals as a sum of the uncorrelated components. We show that the point process model is complementary to the model based on the sum of signals with a wide-range distribution of the relaxation times. In contrast to the Gaussian distribution of the signal intensity of the sum of the uncorrelated components, the point process exhibits asymptotically a power-law distribution of the signal intensity. The developed multiplicative point process model of $1/f^{\beta}$ noise may be used for modeling and analysis of stochastic processes in different systems with the power-law distribution of the intensity of pulsing signals.
cond-mat/0506037
Diagnosis of weaknesses in modern error correction codes: a physics approach
cond-mat.stat-mech cond-mat.dis-nn cs.IT math.IT
One of the main obstacles to the wider use of the modern error-correction codes is that, due to the complex behavior of their decoding algorithms, no systematic method which would allow characterization of the Bit-Error-Rate (BER) is known. This is especially true at the weak noise where many systems operate and where coding performance is difficult to estimate because of the diminishingly small number of errors. We show how the instanton method of physics allows one to solve the problem of BER analysis in the weak noise range by recasting it as a computationally tractable minimization problem.
cond-mat/0506652
The theoretical capacity of the Parity Source Coder
cond-mat.dis-nn cond-mat.stat-mech cs.IT math.IT
The Parity Source Coder is a protocol for data compression which is based on a set of parity checks organized in a sparse random network. We consider here the case of memoryless unbiased binary sources. We show that the theoretical capacity saturate the Shannon limit at large K. We also find that the first corrections to the leading behavior are exponentially small, so that the behavior at finite K is very close to the optimal one.
cond-mat/0508216
Cluster Variation Method in Statistical Physics and Probabilistic Graphical Models
cond-mat.stat-mech cs.IT math.IT
The cluster variation method (CVM) is a hierarchy of approximate variational techniques for discrete (Ising--like) models in equilibrium statistical mechanics, improving on the mean--field approximation and the Bethe--Peierls approximation, which can be regarded as the lowest level of the CVM. In recent years it has been applied both in statistical physics and to inference and optimization problems formulated in terms of probabilistic graphical models. The foundations of the CVM are briefly reviewed, and the relations with similar techniques are discussed. The main properties of the method are considered, with emphasis on its exactness for particular models and on its asymptotic properties. The problem of the minimization of the variational free energy, which arises in the CVM, is also addressed, and recent results about both provably convergent and message-passing algorithms are discussed.
cond-mat/0511159
Learning by message-passing in networks of discrete synapses
cond-mat.dis-nn cs.LG q-bio.NC
We show that a message-passing process allows to store in binary "material" synapses a number of random patterns which almost saturates the information theoretic bounds. We apply the learning algorithm to networks characterized by a wide range of different connection topologies and of size comparable with that of biological systems (e.g. $n\simeq10^{5}-10^{6}$). The algorithm can be turned into an on-line --fault tolerant-- learning protocol of potential interest in modeling aspects of synaptic plasticity and in building neuromorphic devices.
cond-mat/0512017
Combinatorial Information Theory: I. Philosophical Basis of Cross-Entropy and Entropy
cond-mat.stat-mech cs.IT math-ph math.IT math.MP physics.data-an
This study critically analyses the information-theoretic, axiomatic and combinatorial philosophical bases of the entropy and cross-entropy concepts. The combinatorial basis is shown to be the most fundamental (most primitive) of these three bases, since it gives (i) a derivation for the Kullback-Leibler cross-entropy and Shannon entropy functions, as simplified forms of the multinomial distribution subject to the Stirling approximation; (ii) an explanation for the need to maximize entropy (or minimize cross-entropy) to find the most probable realization; and (iii) new, generalized definitions of entropy and cross-entropy - supersets of the Boltzmann principle - applicable to non-multinomial systems. The combinatorial basis is therefore of much broader scope, with far greater power of application, than the information-theoretic and axiomatic bases. The generalized definitions underpin a new discipline of ``{\it combinatorial information theory}'', for the analysis of probabilistic systems of any type. Jaynes' generic formulation of statistical mechanics for multinomial systems is re-examined in light of the combinatorial approach. (abbreviated abstract)
cond-mat/0601021
Characterizing correlations of flow oscillations at bottlenecks
cond-mat.stat-mech cs.MA
"Oscillations" occur in quite different kinds of many-particle-systems when two groups of particles with different directions of motion meet or intersect at a certain spot. We present a model of pedestrian motion that is able to reproduce oscillations with different characteristics. The Wald-Wolfowitz test and Gillis' correlated random walk are shown to hold observables that can be used to characterize different kinds of oscillations.
cond-mat/0601487
Loop Calculus in Statistical Physics and Information Science
cond-mat.stat-mech cond-mat.dis-nn cs.IT math.IT
Considering a discrete and finite statistical model of a general position we introduce an exact expression for the partition function in terms of a finite series. The leading term in the series is the Bethe-Peierls (Belief Propagation)-BP contribution, the rest are expressed as loop-contributions on the factor graph and calculated directly using the BP solution. The series unveils a small parameter that often makes the BP approximation so successful. Applications of the loop calculus in statistical physics and information science are discussed.
cond-mat/0601573
Amorphous packings of hard spheres in large space dimension
cond-mat.stat-mech cond-mat.dis-nn cs.IT math.GM math.IT
In a recent paper (cond-mat/0506445) we derived an expression for the replicated free energy of a liquid of hard spheres based on the HNC free energy functional. An approximate equation of state for the glass and an estimate of the random close packing density were obtained in d=3. Here we show that the HNC approximation is not needed: the same expression can be obtained from the full diagrammatic expansion of the replicated free energy. Then, we consider the asymptotics of this expression when the space dimension d is very large. In this limit, the entropy of the hard sphere liquid has been computed exactly. Using this solution, we derive asymptotic expressions for the glass transition density and for the random close packing density for hard spheres in large space dimension.
cond-mat/0602183
Nonlinear parametric model for Granger causality of time series
cond-mat.dis-nn cond-mat.stat-mech cs.LG physics.med-ph q-bio.QM
We generalize a previously proposed approach for nonlinear Granger causality of time series, based on radial basis function. The proposed model is not constrained to be additive in variables from the two time series and can approximate any function of these variables, still being suitable to evaluate causality. Usefulness of this measure of causality is shown in a physiological example and in the study of the feed-back loop in a model of excitatory and inhibitory neurons.
cond-mat/0602345
Numerical Modeling of Coexistence, Competition and Collapse of Rotating Spiral Waves in Three-Level Excitable Media with Discrete Active Centers and Absorbing Boundaries
cond-mat.other cs.NE nlin.CG
Spatio-temporal dynamics of excitable media with discrete three-level active centers (ACs) and absorbing boundaries is studied numerically by means of a deterministic three-level model (see S. D. Makovetskiy and D. N. Makovetskii, on-line preprint cond-mat/0410460 ), which is a generalization of Zykov- Mikhailov model (see Sov. Phys. -- Doklady, 1986, Vol.31, No.1, P.51) for the case of two-channel diffusion of excitations. In particular, we revealed some qualitatively new features of coexistence, competition and collapse of rotating spiral waves (RSWs) in three-level excitable media under conditions of strong influence of the second channel of diffusion. Part of these features are caused by unusual mechanism of RSWs evolution when RSW's cores get into the surface layer of an active medium (i.~e. the layer of ACs resided at the absorbing boundary). Instead of well known scenario of RSW collapse, which takes place after collision of RSW's core with absorbing boundary, we observed complicated transformations of the core leading to nonlinear ''reflection'' of the RSW from the boundary or even to birth of several new RSWs in the surface layer. To our knowledge, such nonlinear ''reflections'' of RSWs and resulting die hard vorticity in excitable media with absorbing boundaries were unknown earlier. ACM classes: F.1.1, I.6, J.2; PACS numbers: 05.65.+b, 07.05.Tp, 82.20.Wt
cond-mat/0602661
On the high density behavior of Hamming codes with fixed minimum distance
cond-mat.stat-mech cond-mat.dis-nn cs.IT math.IT
We discuss the high density behavior of a system of hard spheres of diameter d on the hypercubic lattice of dimension n, in the limit n -> oo, d -> oo, d/n=delta. The problem is relevant for coding theory. We find a solution to the equations describing the liquid up to very large values of the density, but we show that this solution gives a negative entropy for the liquid phase when the density is large enough. We then conjecture that a phase transition towards a different phase might take place, and we discuss possible scenarios for this transition. Finally we discuss the relation between our results and known rigorous bounds on the maximal density of the system.
cond-mat/0603189
Loop series for discrete statistical models on graphs
cond-mat.stat-mech cond-mat.dis-nn cs.IT math.IT
In this paper we present derivation details, logic, and motivation for the loop calculus introduced in \cite{06CCa}. Generating functions for three inter-related discrete statistical models are each expressed in terms of a finite series. The first term in the series corresponds to the Bethe-Peierls (Belief Propagation)-BP contribution, the other terms are labeled by loops on the factor graph. All loop contributions are simple rational functions of spin correlation functions calculated within the BP approach. We discuss two alternative derivations of the loop series. One approach implements a set of local auxiliary integrations over continuous fields with the BP contribution corresponding to an integrand saddle-point value. The integrals are replaced by sums in the complimentary approach, briefly explained in \cite{06CCa}. A local gauge symmetry transformation that clarifies an important invariant feature of the BP solution, is revealed in both approaches. The partition function remains invariant while individual terms change under the gauge transformation. The requirement for all individual terms to be non-zero only for closed loops in the factor graph (as opposed to paths with loose ends) is equivalent to fixing the first term in the series to be exactly equal to the BP contribution. Further applications of the loop calculus to problems in statistical physics, computer and information sciences are discussed.
cond-mat/0604267
Survey propagation for the cascading Sourlas code
cond-mat.stat-mech cs.IT math.IT
We investigate how insights from statistical physics, namely survey propagation, can improve decoding of a particular class of sparse error correcting codes. We show that a recently proposed algorithm, time averaged belief propagation, is in fact intimately linked to a specific survey propagation for which Parisi's replica symmetry breaking parameter is set to zero, and that the latter is always superior to belief propagation in the high connectivity limit. We briefly look at further improvements available by going to the second level of replica symmetry breaking.
cond-mat/0606125
Microscopic activity patterns in the Naming Game
cond-mat.dis-nn cs.MA physics.soc-ph
The models of statistical physics used to study collective phenomena in some interdisciplinary contexts, such as social dynamics and opinion spreading, do not consider the effects of the memory on individual decision processes. On the contrary, in the Naming Game, a recently proposed model of Language formation, each agent chooses a particular state, or opinion, by means of a memory-based negotiation process, during which a variable number of states is collected and kept in memory. In this perspective, the statistical features of the number of states collected by the agents becomes a relevant quantity to understand the dynamics of the model, and the influence of topological properties on memory-based models. By means of a master equation approach, we analyze the internal agent dynamics of Naming Game in populations embedded on networks, finding that it strongly depends on very general topological properties of the system (e.g. average and fluctuations of the degree). However, the influence of topological properties on the microscopic individual dynamics is a general phenomenon that should characterize all those social interactions that can be modeled by memory-based negotiation processes.
cond-mat/0606696
Statistical mechanics of error exponents for error-correcting codes
cond-mat.dis-nn cond-mat.stat-mech cs.IT math.IT
Error exponents characterize the exponential decay, when increasing message length, of the probability of error of many error-correcting codes. To tackle the long standing problem of computing them exactly, we introduce a general, thermodynamic, formalism that we illustrate with maximum-likelihood decoding of low-density parity-check (LDPC) codes on the binary erasure channel (BEC) and the binary symmetric channel (BSC). In this formalism, we apply the cavity method for large deviations to derive expressions for both the average and typical error exponents, which differ by the procedure used to select the codes from specified ensembles. When decreasing the noise intensity, we find that two phase transitions take place, at two different levels: a glass to ferromagnetic transition in the space of codewords, and a paramagnetic to glass transition in the space of codes.
cond-mat/0608312
On Cavity Approximations for Graphical Models
cond-mat.stat-mech cond-mat.dis-nn cs.IT math.IT
We reformulate the Cavity Approximation (CA), a class of algorithms recently introduced for improving the Bethe approximation estimates of marginals in graphical models. In our new formulation, which allows for the treatment of multivalued variables, a further generalization to factor graphs with arbitrary order of interaction factors is explicitly carried out, and a message passing algorithm that implements the first order correction to the Bethe approximation is described. Furthermore we investigate an implementation of the CA for pairwise interactions. In all cases considered we could confirm that CA[k] with increasing $k$ provides a sequence of approximations of markedly increasing precision. Furthermore in some cases we could also confirm the general expectation that the approximation of order $k$, whose computational complexity is $O(N^{k+1})$ has an error that scales as $1/N^{k+1}$ with the size of the system. We discuss the relation between this approach and some recent developments in the field.
cond-mat/0611567
Generalized Statistics Framework for Rate Distortion Theory
cond-mat.stat-mech cs.IT math.IT
Variational principles for the rate distortion (RD) theory in lossy compression are formulated within the ambit of the generalized nonextensive statistics of Tsallis, for values of the nonextensivity parameter satisfying $ 0 < q < 1 $ and $ q > 1 $. Alternating minimization numerical schemes to evaluate the nonextensive RD function, are derived. Numerical simulations demonstrate the efficacy of generalized statistics RD models.
cond-mat/0611717
Non-equilibrium phase transition in negotiation dynamics
cond-mat.stat-mech cs.MA physics.soc-ph q-bio.PE
We introduce a model of negotiation dynamics whose aim is that of mimicking the mechanisms leading to opinion and convention formation in a population of individuals. The negotiation process, as opposed to ``herding-like'' or ``bounded confidence'' driven processes, is based on a microscopic dynamics where memory and feedback play a central role. Our model displays a non-equilibrium phase transition from an absorbing state in which all agents reach a consensus to an active stationary state characterized either by polarization or fragmentation in clusters of agents with different opinions. We show the exystence of at least two different universality classes, one for the case with two possible opinions and one for the case with an unlimited number of opinions. The phase transition is studied analytically and numerically for various topologies of the agents' interaction network. In both cases the universality classes do not seem to depend on the specific interaction topology, the only relevant feature being the total number of different opinions ever present in the system.
cond-mat/0701218
Generalized Statistics Framework for Rate Distortion Theory with Bregman Divergences
cond-mat.stat-mech cs.IT math.IT
A variational principle for the rate distortion (RD) theory with Bregman divergences is formulated within the ambit of the generalized (nonextensive) statistics of Tsallis. The Tsallis-Bregman RD lower bound is established. Alternate minimization schemes for the generalized Bregman RD (GBRD) theory are derived. A computational strategy to implement the GBRD model is presented. The efficacy of the GBRD model is exemplified with the aid of numerical simulations.
cond-mat/0703351
Error Correction and Digitalization Concepts in Biochemical Computing
cond-mat.soft cond-mat.dis-nn cond-mat.mtrl-sci cs.CE q-bio.BM quant-ph
We offer a theoretical design of new systems that show promise for digital biochemical computing, including realizations of error correction by utilizing redundancy, as well as signal rectification. The approach includes information processing using encoded DNA sequences, DNAzyme biocatalyzed reactions and the use of DNA-functionalized magnetic nanoparticles. Digital XOR and NAND logic gates and copying (fanout) are designed using the same components.
cond-mat/9703183
Finite size scaling of the bayesian perceptron
cond-mat.stat-mech cond-mat.dis-nn cs.AI cs.LG
We study numerically the properties of the bayesian perceptron through a gradient descent on the optimal cost function. The theoretical distribution of stabilities is deduced. It predicts that the optimal generalizer lies close to the boundary of the space of (error-free) solutions. The numerical simulations are in good agreement with the theoretical distribution. The extrapolation of the generalization error to infinite input space size agrees with the theoretical results. Finite size corrections are negative and exhibit two different scaling regimes, depending on the training set size. The variance of the generalization error vanishes for $N \rightarrow \infty$ confirming the property of self-averaging.
cond-mat/9810144
Relaxation in graph coloring and satisfiability problems
cond-mat.dis-nn cs.AI
Using T=0 Monte Carlo simulation, we study the relaxation of graph coloring (K-COL) and satisfiability (K-SAT), two hard problems that have recently been shown to possess a phase transition in solvability as a parameter is varied. A change from exponentially fast to power law relaxation, and a transition to freezing behavior are found. These changes take place for smaller values of the parameter than the solvability transition. Results for the coloring problem for colorable and clustered graphs and for the fraction of persistent spins for satisfiability are also presented.
cond-mat/9902011
Cortical Potential Distributions and Cognitive Information Processing
cond-mat.dis-nn adap-org cond-mat.stat-mech cs.NE math-ph math.MP nlin.AO physics.bio-ph q-bio
The use of cortical field potentials rather than the details of spike trains as the basis for cognitive information processing is proposed. This results in a space of cognitive elements with natural metrics. Sets of spike trains may also be considered to be points in a multidimensional metric space. The closeness of sets of spike trains in such a space implies the closeness of points in the resulting function space of potential distributions.
cond-mat/9906206
Ocular dominance patterns in mammalian visual cortex: A wire length minimization approach
cond-mat.soft cond-mat.dis-nn cs.NE physics.bio-ph q-bio
We propose a theory for ocular dominance (OD) patterns in mammalian primary visual cortex. This theory is based on the premise that OD pattern is an adaptation to minimize the length of intra-cortical wiring. Thus we can understand the existing OD patterns by solving a wire length minimization problem. We divide all the neurons into two classes: left-eye dominated and right-eye dominated. We find that segregation of neurons into monocular regions reduces wire length if the number of connections with the neurons of the same class differs from that with the other class. The shape of the regions depends on the relative fraction of neurons in the two classes. If the numbers are close we find that the optimal OD pattern consists of interdigitating stripes. If one class is less numerous than the other, the optimal OD pattern consists of patches of the first class neurons in the sea of the other class neurons. We predict the transition from stripes to patches when the fraction of neurons dominated by the ipsilateral eye is about 40%. This prediction agrees with the data in macaque and Cebus monkeys. This theory can be applied to other binary cortical systems.
cs/0001002
Minimum Description Length and Compositionality
cs.CL cs.AI
We present a non-vacuous definition of compositionality. It is based on the idea of combining the minimum description length principle with the original definition of compositionality (that is, that the meaning of the whole is a function of the meaning of the parts). The new definition is intuitive and allows us to distinguish between compositional and non-compositional semantics, and between idiomatic and non-idiomatic expressions. It is not ad hoc, since it does not make any references to non-intrinsic properties of meaning functions (like being a polynomial). Moreover, it allows us to compare different meaning functions with respect to how compositional they are. It bridges linguistic and corpus-based, statistical approaches to natural language understanding.
cs/0001004
Multiplicative Algorithm for Orthgonal Groups and Independent Component Analysis
cs.LG
The multiplicative Newton-like method developed by the author et al. is extended to the situation where the dynamics is restricted to the orthogonal group. A general framework is constructed without specifying the cost function. Though the restriction to the orthogonal groups makes the problem somewhat complicated, an explicit expression for the amount of individual jumps is obtained. This algorithm is exactly second-order-convergent. The global instability inherent in the Newton method is remedied by a Levenberg-Marquardt-type variation. The method thus constructed can readily be applied to the independent component analysis. Its remarkable performance is illustrated by a numerical simulation.
cs/0001006
Compositionality, Synonymy, and the Systematic Representation of Meaning
cs.CL cs.LO
In a recent issue of Linguistics and Philosophy Kasmi and Pelletier (1998) (K&P), and Westerstahl (1998) criticize Zadrozny's (1994) argument that any semantics can be represented compositionally. The argument is based upon Zadrozny's theorem that every meaning function m can be encoded by a function \mu such that (i) for any expression E of a specified language L, m(E) can be recovered from \mu(E), and (ii) \mu is a homomorphism from the syntactic structures of L to interpretations of L. In both cases, the primary motivation for the objections brought against Zadrozny's argument is the view that his encoding of the original meaning function does not properly reflect the synonymy relations posited for the language. In this paper, we argue that these technical criticisms do not go through. In particular, we prove that \mu properly encodes synonymy relations, i.e. if two expressions are synonymous, then their compositional meanings are identical. This corrects some misconceptions about the function \mu, e.g. Janssen (1997). We suggest that the reason that semanticists have been anxious to preserve compositionality as a significant constraint on semantic theory is that it has been mistakenly regarded as a condition that must be satisfied by any theory that sustains a systematic connection between the meaning of an expression and the meanings of its parts. Recent developments in formal and computational semantics show that systematic theories of meanings need not be compositional.
cs/0001008
Predicting the expected behavior of agents that learn about agents: the CLRI framework
cs.MA cs.LG
We describe a framework and equations used to model and predict the behavior of multi-agent systems (MASs) with learning agents. A difference equation is used for calculating the progression of an agent's error in its decision function, thereby telling us how the agent is expected to fare in the MAS. The equation relies on parameters which capture the agent's learning abilities, such as its change rate, learning rate and retention rate, as well as relevant aspects of the MAS such as the impact that agents have on each other. We validate the framework with experimental results using reinforcement learning agents in a market system, as well as with other experimental results gathered from the AI literature. Finally, we use PAC-theory to show how to calculate bounds on the values of the learning parameters.
cs/0001010
A Real World Implementation of Answer Extraction
cs.CL
In this paper we describe ExtrAns, an answer extraction system. Answer extraction (AE) aims at retrieving those exact passages of a document that directly answer a given user question. AE is more ambitious than information retrieval and information extraction in that the retrieval results are phrases, not entire documents, and in that the queries may be arbitrarily specific. It is less ambitious than full-fledged question answering in that the answers are not generated from a knowledge base but looked up in the text of documents. The current version of ExtrAns is able to parse unedited Unix "man pages", and derive the logical form of their sentences. User queries are also translated into logical forms. A theorem prover then retrieves the relevant phrases, which are presented through selective highlighting in their context.
cs/0001012
Measures of Distributional Similarity
cs.CL
We study distributional similarity measures for the purpose of improving probability estimation for unseen cooccurrences. Our contributions are three-fold: an empirical comparison of a broad range of measures; a classification of similarity functions based on the information that they incorporate; and the introduction of a novel function that is superior at evaluating potential proxy distributions.
cs/0001015
Multi-Agent Only Knowing
cs.AI cs.LO
Levesque introduced a notion of ``only knowing'', with the goal of capturing certain types of nonmonotonic reasoning. Levesque's logic dealt with only the case of a single agent. Recently, both Halpern and Lakemeyer independently attempted to extend Levesque's logic to the multi-agent case. Although there are a number of similarities in their approaches, there are some significant differences. In this paper, we reexamine the notion of only knowing, going back to first principles. In the process, we simplify Levesque's completeness proof, and point out some problems with the earlier definitions. This leads us to reconsider what the properties of only knowing ought to be. We provide an axiom system that captures our desiderata, and show that it has a semantics that corresponds to it. The axiom system has an added feature of interest: it includes a modal operator for satisfiability, and thus provides a complete axiomatization for satisfiability in the logic K45.
cs/0001018
Adaptive simulated annealing (ASA): Lessons learned
cs.MS cs.CE
Adaptive simulated annealing (ASA) is a global optimization algorithm based on an associated proof that the parameter space can be sampled much more efficiently than by using other previous simulated annealing algorithms. The author's ASA code has been publicly available for over two years. During this time the author has volunteered to help people via e-mail, and the feedback obtained has been used to further develop the code. Some lessons learned, in particular some which are relevant to other simulated annealing algorithms, are described.
cs/0001020
Exploiting Syntactic Structure for Natural Language Modeling
cs.CL
The thesis presents an attempt at using the syntactic structure in natural language for improved language models for speech recognition. The structured language model merges techniques in automatic parsing and language modeling using an original probabilistic parameterization of a shift-reduce parser. A maximum likelihood reestimation procedure belonging to the class of expectation-maximization algorithms is employed for training the model. Experiments on the Wall Street Journal, Switchboard and Broadcast News corpora show improvement in both perplexity and word error rate - word lattice rescoring - over the standard 3-gram language model. The significance of the thesis lies in presenting an original approach to language modeling that uses the hierarchical - syntactic - structure in natural language to improve on current 3-gram modeling techniques for large vocabulary speech recognition.
cs/0001021
Refinement of a Structured Language Model
cs.CL
A new language model for speech recognition inspired by linguistic analysis is presented. The model develops hidden hierarchical structure incrementally and uses it to extract meaningful information from the word history - thus enabling the use of extended distance dependencies - in an attempt to complement the locality of currently used n-gram Markov models. The model, its probabilistic parametrization, a reestimation algorithm for the model parameters and a set of experiments meant to evaluate its potential for speech recognition are presented.
cs/0001022
Recognition Performance of a Structured Language Model
cs.CL
A new language model for speech recognition inspired by linguistic analysis is presented. The model develops hidden hierarchical structure incrementally and uses it to extract meaningful information from the word history - thus enabling the use of extended distance dependencies - in an attempt to complement the locality of currently used trigram models. The structured language model, its probabilistic parameterization and performance in a two-pass speech recognizer are presented. Experiments on the SWITCHBOARD corpus show an improvement in both perplexity and word error rate over conventional trigram models.