Language model cross adaptation for LVCSR system combination

State-of-the-art large vocabulary continuous speech recognition (LVCSR) systems often combine outputs from multiple sub-systems that may even be developed at different sites. Cross system adaptation, in which model adaptation is performed using the outputs from another sub-system, can be used as an alternative to hypothesis level combination schemes such as ROVER. Normally cross adaptation is only performed on the acoustic models. However, there are many other levels in LVCSR systems' modelling hierarchy where complimentary features may be exploited, for example, the sub-word and the word level, to further improve cross adaptation based system combination. It is thus interesting to also cross adapt language models (LMs) to capture these additional useful features. In this paper cross adaptation is applied to three forms of language models, a multi-level LM that models both syllable and word sequences, a word level neural network LM, and the linear combination of the two. Significant error rate reductions of 4.0-7.1% relative were obtained over ROVER and acoustic model only cross adaptation when combining a range of Chinese LVCSR sub-systems used in the 2010 and 2011 DARPA GALE evaluations. © 2012 Elsevier Ltd. All rights reserved.

On equations for regular languages, finite automata, and sequential networks

We consider systems of equations of the form where A is the underlying alphabet, the Xi are variables, the Pi,a are boolean functions in the variables Xi, and each δi is either the empty word or the empty set. The symbols υ and denote concatenation and union of languages over A. We show that any such system has a unique solution which, moreover, is regular. These equations correspond to a type of automation, called boolean automation, which is a generalization of a nondeterministic automation. The equations are then used to determine the language accepted by a sequential network; they are obtainable directly from the network.

左侧梭状回内文字的字形加工：中文正常阅读者，中文文盲、非中文母语被试的对比研究

Reading is an important human-specific skill obtained through extensive learning experience and is reliance on the ability to rapidly recognize single words. According to the behavioral studies, the most important stage of reading is the representation of “visual word form”, which is independent on surface visual features of the reading materials. The prelexical visual word form representation is characterized by the abstractive and highly effective and precise processing. Neuroimaging and neuropsychological studies have investigated the neural basis underlying the visual word form processing. On the basis of summary of the existing literature, the current thesis aimed to address three fundamental questions involving neural basis of word recognition. First, is there a dedicated neural network that is specialized for word recognition? Second, is the orthographic information represented in the putative word/character selective region (VWFA)? Third, what is the role of reading experience in the genesis of the VWFA, is experience a main driver to shape VWFA instead of evolutionary selectivity? Nineteen Chinese literate volunteers, 5 Chinese illiterates and 4 native English speakers participated in this study, and performed perceptual tasks during fMRI scanning. To address the first question, we compared the differential responses to three categories of visual objects, i.e., faces, line drawings of objects and Chinese characters, and defined the region of interesting (ROI) for the next experiment. To address the second question, Chinese character orthography was manipulated to reveal possible differential responses to real characters, false characters, radical combinations, and stroke combinations in the regions defined by the first experiment. To examine the role of reading experience in genesis of specialization for character, the responses for unfamiliar Chinese characters in Chinese illiterates and native English speakers were compared with that in the Chinese literates, and tracked the change in cortical activation after a short-term reading training in the illiterates. Data were analyzed in two dimensions. Both BOLD signal amplitude and spatial distribution pattern among multi-voxels were used to systematically investigate the responsiveness of the left fusiform gyrus to Chinese characters. Our results provide strong and clear evidence for the existence of functionally specialized regions in the human ventral occipital-temporal cortex. In the skilled readers a region specialized for written words could be consistently found in the lateral part of the left fusiform gyrus, line drawings in the median part and faces in the middle. Our results further show that spatial distribution analysis, a method that was not commonly used in neuroimaging of reading, appears to be a more effective measurement for category specialization for visual objects processing. Although we failed to provide evidence that VWFA processes orthographic information in terms of signal intensitiy, we do show that response pattern of real characters and radical collections in this area is different from that of false characters and random stroke combinations. Our last set of experiments suggests that the selective bias to reading material is clearly experience dependent. The response to unknown characters in both English speakers/readers and Chinese illiterates is fundamentally different from that of the skilled Chinese readers. The response pattern for unknown characters is more similar to that for line drawings rather as a weak version of character in skilled Chinese readers. Short-term training is not sufficient to produce VWFA bias even when tested with learned characters, rather the learned characters generated a overall upward shift of the activation of the left fusiform region. Formation of a dedicated region specialized for visual word/character might depend on long-term extensive reading experience, or there might be a critical period for reading acquisition.

汉语文本语篇语句焦点确定及焦点与重音关系研究

This study investigated the method of the focus identification in Chinese text discourse and the relationship between accent and focus, large corpus analysis and decision tree were used in the research. The main results are: 1. Based on the concept of the Focus and understanding of the discourse, Foci identification is consistent and steady; 2. Special Focus markers and specific Focus constructions have greater influence than special constituent order on identifying Focus in Chinese discourse; while information states also have great influence on focus identifying; part of speech，information state, the relative position in the sentence, focus-sensitive operator, specific Focus constructions, contrast relations, relations between the sentences are important factors to focus identifying; 3. Using multi-dimensional tagging and knowledge discovery, it is a feasible way to construct and employ decision trees by computing tagging results to identify Focus; 4. Focus predicting also depends on literal types and styles of the discourse, several types of decision trees should be constructed for different literal types; 5. In the monologue discourse, the most prominent accent is located on the Focus word or in the scope of the Focus; there are some kinds of rules on accent assignment in broad Focus; it is necessary to analyze and classify focus structure for the research of relations between accent and Focus.

Odnowienie jedności Metropolii Kijowskiej ze Stolicą Apostolską według Jana Pawła II

Wydział Teologiczny: Zakład Teologii Fundamentalnej i Ekumenicznej

O antropomorfizowaniu i potrzebie przedstawień figuralnych w sztuce greckiej

Estetyka w archeologii. Antropomorfizacje w pradziejach i starożytności, eds. E. Bugaj, A. P. Kowalski, Poznań: Wydawnictwo Poznańskie.

Conversion as transformative experience: a sociological study of identity formation and transformation processes

This thesis contributes to the understanding of the processes involved in the formation and transformation of identities. It achieves this goal by establishing the critical importance of ‘background’ and ‘liminality’ in the shaping of identity. Drawing mainly from the work of cultural anthropology and philosophical hermeneutics a theoretical framework is constructed from which transformative experiences can be analysed. The particular experience at the heart of this study is the phenomenon of conversion and the dynamics involved in the construction of that process. Establishing the axial age as the horizon from which the process of conversion emerged will be the main theme of the first part of the study. Identifying the ‘birth’ of conversion allows a deeper understanding of the historical dynamics that make up the process. From these fundamental dynamics a theoretical framework is constructed in order to analyse the conversion process. Applying this theoretical framework to a number of case-studies will be the central focus of this study. The transformative experiences of Saint Augustine, the fourteenth century nun Margaret Ebner, the communist revolutionary Karl Marx and the literary figure of Arthur Koestler will provide the material onto which the theoretical framework can be applied. A synthesis of the Judaic religious and the Greek philosophical traditions will be the main findings for the shaping of Augustine’s conversion experience. The dissolution of political order coupled with the institutionalisation of the conversion process will illuminate the mystical experiences of Margaret Ebner at a time when empathetic conversion reached its fullest expression. The final case-studies examine two modern ‘conversions’ that seem to have an ideological rather than a religious basis to them. On closer examination it will be found that the German tradition of Biblical Criticism played a most influential role in the ‘conversion’ of Marx and mythology the best medium to understand the experiences of Koestler. The main ideas emerging from this study highlight the fluidity of identity and the important role of ‘background’ in its transformation. The theoretical framework, as constructed for this study, is found to be a useful methodological tool that can offer insights into experiences, such as conversion, that otherwise would remain hidden from our enquiries.

Sikh Sabad Kirtan as a Musical Construction of Memory

The performance of devotional music in India has been an active, sonic conduit where spiritual identities are shaped and forged, and both history and mythology lived out and remembered daily. For the followers of Sikhism, congregational hymn singing has been the vehicle through which text, melody and ritual act as repositories of memory, elevating memory to a place where historical and social events can be reenacted and memorialized on levels of spiritual significance. This dissertation investigates the musical process of Shabad Kirtan, Sikh hymn singing, in a Sikh musical service as a powerful vehicle to forge a sense of identification between individual and the group. As an intimate part of Sikh life from birth to death, the repertoire of Shabad Kirtan draws from a rich mosaic of classical and folk genres as well as performance styles, acting as a musical and cognitive archive. Through a detailed analysis of the Asa Di Var service, Shabad Kirtan is explored as a phenomenological experience where time, place and occasion interact as a meaningful unit through which the congregation creates and recreates themselves, invoking deep memories and emotional experiences. Supported by explanatory tables, diagrams and musical transcriptions, the sonic movements of the service show how the Divine Word as Shabad is not only embodied through the Guru Granth Sahib, but also encountered through the human enactment of the service, aurally, viscerally and phenomenologically.

Assessing segmentation processes by click detection: online measure of statistical learning, or simple interference?

Statistical learning can be used to extract the words from continuous speech. Gómez, Bion, and Mehler (Language and Cognitive Processes, 26, 212–223, 2011) proposed an online measure of statistical learning: They superimposed auditory clicks on a continuous artificial speech stream made up of a random succession of trisyllabic nonwords. Participants were instructed to detect these clicks, which could be located either within or between words. The results showed that, over the length of exposure, reaction times (RTs) increased more for within-word than for between-word clicks. This result has been accounted for by means of statistical learning of the between-word boundaries. However, even though statistical learning occurs without an intention to learn, it nevertheless requires attentional resources. Therefore, this process could be affected by a concurrent task such as click detection. In the present study, we evaluated the extent to which the click detection task indeed reflects successful statistical learning. Our results suggest that the emergence of RT differences between within- and between-word click detection is neither systematic nor related to the successful segmentation of the artificial language. Therefore, instead of being an online measure of learning, the click detection task seems to interfere with the extraction of statistical regularities.

Investigation into a modular rule-based testing method for testing business rules in scheduling applications

Rule testing in transport scheduling is a complex and potentially costly business problem. This paper proposes an automated method for the rule-based testing of business rules using the extensible Markup Language for rule representation and transportation. A compiled approach to rule execution is also proposed for performance-critical scheduling systems.

Assistive software tools for secondary-level students with literacy difficulties.

Study conducted to evaluate the effectiveness of four assistive technology (AT) tools on literacy: (1) speech synthesis, (2) spellchecker, (3) homophone tool, and (4) dictionary. All four of these programs are featured in TextHelp’s Read&Write Gold software package. A total of 93 secondary-level students with reading disabilities participated in the study. The participants completed a number of computer-based literacy tests after being assigned to a Read&Write group or a control group that utilized Microsoft Word. The results indicated that improvements in the following areas for the Read&Write group: (1) reading comprehension, (2) homophone error detection, (3) spelling error detection, and (4) word meanings. The Microsoft Word group also improved in the areas of word meanings and error detection, though performed worse on homophone error detection. The authors contend that these results indicate that speech synthesis, spell checkers, homophone tools, and dictionary programs have a positive effect on literacy among students with reading disabilities. This study was conducted by researchers at the Queen’s University in Belfast, Ireland.

Analogies between linguistics and information theory

An analogy is established between the syntagm and paradigm from Saussurean linguistics and the message and messages for selection from the information theory initiated by Claude Shannon. The analogy is pursued both as an end itself and for its analytic value in understanding patterns of retrieval from full text systems. The multivalency of individual words when isolated from their syntagm is contrasted with the relative stability of meaning of multi-word sequences, when searching ordinary written discourse. The syntagm is understood as the linear sequence of oral and written language. Saussureâ??s understanding of the word, as a unit which compels recognition by the mind, is endorsed, although not regarded as final. The lesser multivalency of multi-word sequences is understood as the greater determination of signification by the extended syntagm. The paradigm is primarily understood as the network of associations a word acquires when considered apart from the syntagm. The restriction of information theory to expression or signals, and its focus on the combinatorial aspects of the message, is sustained. The message in the model of communication in information theory can include sequences of written language. Shannonâ??s understanding of the written word, as a cohesive group of letters, with strong internal statistical influences, is added to the Saussurean conception. Sequences of more than one word are regarded as weakly correlated concatenations of cohesive units.

Notas sobre el "primum cognitum" en Cayetano y su vínculo con la metafísica

Cayetano tuvo un papel protagónico en los debates intelectuales de su época (s. XVI), destacándose como un auténtico adalid del tomismo; inclusive muchas de sus tesis pasaron a formar parte de tal doctrina, las cuales –a veces matizadas y otras no tanto– perviven hasta nuestros días. Entre los diversos temas que afrontó se destaca su particular noción de primum cognitum. Aquí, partimos del hecho de que la tradición tomista no ha visto mayores inconvenientes en equiparar lo que entienden Cayetano y el Aquinate por el ens primo cadit; empero nosotros estamos en condiciones de afirmar que tal equiparación al menos es problemática. En este sentido, nos ocuparemos de realizar un primer acercamiento al tema del ente primer conocido cayetaniano para intentar mostrar su real significado. Asimismo, procuraremos ver, ayudados por los aportes de algunos estudiosos contemporáneos –en especial el de Lawrence Dewan–, hasta qué punto tal noción se corresponde o no con los desarrollos teóricos de Tomás de Aquino, sobre todo con el vínculo entre el primer conocido y el ente como objeto de la metafísica.

Genetic diversity of liver flukes (Fasciola hepatica) from Eastern Europe.

The genetic diversity of liver fluke populations in three different countries from Eastern Europe (Greece, Bulgaria, and Poland) in comparison with available data from other countries was determined. Specifically, SNPs from regions of two nuclear genes, 28S rDNA, ß-tubulin 3 and an informative region of the mitochondrial genome were examined. Two major lineages for the 28S rDNA gene based on the highly polymorphic 105th nucleotide position were found. These lineages were widely and almost equally spread not only through the countries studied but also in other investigated geographical areas. Two basic lineages and additional haplotypes were defined for the mtDNA gene region, consisting of the cytochrome c oxidase subunit III gene, transfer RNA histidine gene and cytochome b gene. The basic lineages were observed within Greek, Bulgarian, and Polish Fasciola hepatica populations but the distribution of additional haplotypes differed between the populations from the three countries. For the ß-tubulin 3 gene multiple polymorphic sites were revealed but no explicit clades. The SNPs were spread unequally in all studied geographical regions with an evident distinction between the Greek and Polish specimens. Additional genotypes for the 28S rDNA region as well as haplotypes of the mtDNA region that were typical for the Greek or Polish populations were observed. Significant polymorphisms for ß-tubulin 3 gene were displayed with decreasing percentage of presence within populations from Greece to Poland. There was an amino acid substitution in ß-tubulin 3 protein found only among Polish specimens. It is hypothesized that genotypic differences between Greek, Bulgarian, and Polish liver fluke populations are due to territorial division and genetic drift in past epochs.

A new method for estimating liquidity risk: Insights from a liquidity-adjusted CAPM framework

This paper proposes a new non-parametric method for estimating model-free, time-varying liquidity betas which builds on realized covariance and volatility theory. Working under a liquidity-adjusted CAPM framework we provide evidence that liquidity risk is a factor priced in the Greek stock market, mainly arising from the covariation of individual liquidity with local market liquidity, however, the level of liquidity seems to be an irrelevant variable in asset pricing. Our findings provide support to the notion that liquidity shocks transmitted across securities can cause market-wide effects and can have important implications for portfolio diversification strategies. ©2012 Elsevier B.V. All rights reserved.