Forming of Learning Set for Neural Networks in Problems of Losless Data Compression

questions of forming of learning sets for artificial neural networks in problems of lossless data compression are considered. Methods of construction and use of learning sets are studied. The way of forming of learning set during training an artificial neural network on the data stream is offered.

Convex analysis for minimizing and learning submodular set functions

The connections between convexity and submodularity are explored, for purposes of minimizing and learning submodular set functions.

First, we develop a novel method for minimizing a particular class of submodular functions, which can be expressed as a sum of concave functions composed with modular functions. The basic algorithm uses an accelerated first order method applied to a smoothed version of its convex extension. The smoothing algorithm is particularly novel as it allows us to treat general concave potentials without needing to construct a piecewise linear approximation as with graph-based techniques.

Second, we derive the general conditions under which it is possible to find a minimizer of a submodular function via a convex problem. This provides a framework for developing submodular minimization algorithms. The framework is then used to develop several algorithms that can be run in a distributed fashion. This is particularly useful for applications where the submodular objective function consists of a sum of many terms, each term dependent on a small part of a large data set.

Lastly, we approach the problem of learning set functions from an unorthodox perspective---sparse reconstruction. We demonstrate an explicit connection between the problem of learning set functions from random evaluations and that of sparse signals. Based on the observation that the Fourier transform for set functions satisfies exactly the conditions needed for sparse reconstruction algorithms to work, we examine some different function classes under which uniform reconstruction is possible.

The impact of context on pattern category learning and representation

Context traditionally has been regarded in vision research as a determinant for the interpretation of sensory information on the basis of previously acquired knowledge. Here we propose a novel, complementary perspective by showing that context also specifically affects visual category learning. In two experiments involving sets of Compound Gabor patterns we explored how context, as given by the stimulus set to be learned, affects the internal representation of pattern categories. In Experiment 1, we changed the (local) context of the individual signal classes by changing the configuration of the learning set. In Experiment 2, we varied the (global) context of a fixed class configuration by changing the degree of signal accentuation. Generalization performance was assessed in terms of the ability to recognize contrast-inverted versions of the learning patterns. Both contextual variations yielded distinct effects on learning and generalization thus indicating a change in internal category representation. Computer simulations suggest that the latter is related to changes in the set of attributes underlying the production rules of the categories. The implications of these findings for phenomena of contrast (in)variance in visual perception are discussed.

Classification of valence changes of trivalent rare earth ions in alkaline earth borates using artificial neural networks

The investigations of classification on the valence changes from RE3+ to RE2+ (RE = Eu, Sm, Yb, Tm) in host compounds of alkaline earth berate were performed using artificial neural networks (ANNs). For comparison, the common methods of pattern recognition, such as SIMCA, KNN, Fisher discriminant analysis and stepwise discriminant analysis were adopted. A learning set consisting of 24 host compounds and a test set consisting of 12 host compounds were characterized by eight crystal structure parameters. These parameters were reduced from 8 to 4 by leaps and bounds algorithm. The recognition rates from 87.5 to 95.8% and prediction capabilities from 75.0 to 91.7% were obtained. The results provided by ANN method were better than that achieved by the other four methods. (C) 1999 Elsevier Science B.V. All rights reserved.

SEARCH FOR THE BASIC REGULARITIES OF THE TRANSITION EMISSION OF EUROPIUM(II) ION IN COMPLEX FLUORIDES USING PATTERN-RECOGNITION

The relationship between structures of complex fluorides and spectral structure of Eu(II) ion in complex fluorides (AB(m)F(n)) is investigated by means of pattern recognition methods, such as KNN, ALKNN, BAYES, LLM, SIMCA and PCA. A learning set consisting of 32 f-f transition emission host compounds and 31 d-f transition emission host compounds and a test set consisting of 27 host compounds were characterized by 12 crystal structural parameters. These parameters, i.e. features, were reduced from 12 to 6 by multiple criteria for the classification of these host compounds as f-f transition emission or d-f transition emission. A recognition rate from 79.4 to 96.8% and prediction capabilities from 85.2 to 92.6% were obtained. According to the above results, the spectral structures of Eu(II) ion in seven unknown host lattices were predicted.

New prognostic markers, determined using gene expression analyses, reveal two distinct subtypes of chronic myelomonocytic leukaemia patients

Chronic myelomonocytic leukaemia (CMML) is a heterogeneous haematopoietic disorder characterized by myeloproliferative or myelodysplastic features. At present, the pathogenesis of this malignancy is not completely understood. In this study, we sought to analyse gene expression profiles of CMML in order to characterize new molecular outcome predictors. A learning set of 32 untreated CMML patients at diagnosis was available for TaqMan low-density array gene expression analysis. From 93 selected genes related to cancer and cell cycle, we built a five-gene prognostic index after multiplicity correction. Using this index, we characterized two categories of patients with distinct overall survival (94% vs. 19% for good and poor overall survival, respectively; P = 0.007) and we successfully validated its strength on an independent cohort of 21 CMML patients with Affymetrix gene expression data. We found no specific patterns of association with traditional prognostic stratification parameters in the learning cohort. However, the poor survival group strongly correlated with high-risk treated patients and transformation to acute myeloid leukaemia. We report here a new multigene prognostic index for CMML, independent of the gene expression measurement method, which could be used as a powerful tool to predict clinical outcome and help physicians to evaluate criteria for treatments.

Discriminações condicionais sem conseqüências diferenciais em crianças: efeitos da história de treino precoce de simetria

O presente estudo avaliou a formação de classes de estímulos através de um treino de duas discriminações condicionais (AB e AC) com pareamento consistente modelo-comparação correta, sem conseqüências diferenciais imediatas e fading, aplicando-se testes de simetria e de equivalência após cada bloco de treino. Participaram quatro crianças do pré-escolar que foram expostas ao procedimento de ensino, em duas etapas: na Etapa 1, com estímulos usuais e na Etapa 2 com estímulos não usuais Para cada modelo, três estímulos de comparação foram apresentados simultaneamente. Cada modelo foi emparelhado consistentemente, com os estímulos de comparação, sendo que o estímulo de comparação correto e o modelo apareceram em fading ao longo do treino. Relações simétricas foram demonstradas com dois participantes na Etapa 1, mas não ocorreram relações emergentes na Etapa 2. Dois participantes transferiram o desempenho obtido de uma etapa para outra com o treino discriminativo, como uma espécie de learning set arbitrário. Os resultados indicam que a sequência de treino com estímulos usuais e não usuais e a natureza dos estímulos na Etapa 1 foram variáveis relevantes.

Tranferência de aprendizagem no uso de ferramentas por macacos-prego (Cebus cf. apella)

Estudiosos questionam se a proficiência de macacos-prego (Cebus spp.) no uso de objetos como ferramentas, seria fruto de descobertas arbitrárias decorrentes dos frequentes comportamentos exploratórios desses primatas ou se seria devida à compreensão da função das ferramentas. Considerando que tais animais são capazes de modificar, transportar e fabricar ferramentas é possível propor que algum nível de compreensão esteja envolvido, ainda que não seja desvinculado de sua história de vida e sim construído a partir de uma série de interações com situações relevantes para a aquisição de um repertório generalizado de uso de ferramentas. A fim de investigar tal proposta foi realizada uma série de experimentos com dois grupos de macacos-prego (Cebus cf. apella) em que foi manipulada a história experimental desses animais. Todos os sujeitos passaram por repetidas exposições a um problema em que deveriam encaixar seis blocos de brinquedo para construir uma torre, usá-la para ter acesso a uma vareta distante, com essa vareta chegar a uma segunda vareta e encaixá-las formando uma haste longa o suficiente para permitir a recolha de pelotas de alimento em um equipamento. Enquanto dois sujeitos foram repetidamente expostos ao referido problema sem receber nenhum treino adicional, outros dois sujeitos tiveram uma rica história experimental construída, passando pelo treino em tarefas indiretamente relacionadas ao problema final entre as reapresentações do mesmo. Os sujeitos do primeiro grupo não foram capazes de resolver o problema, ao passo que os do segundo grupo o fizeram, ainda que não tenham sido diretamente treinados a isso. Concluiu-se que uma história relevante é fundamental para a chamada compreensão da solução de um problema e que essa compreensão ou insight são processos comportamentais adaptativos, em que habilidades aprendidas em um contexto específicosão transferidas para novos contextos a partir de processos básicos como a Generalização de Estímulos, a Generalização Funcional e o Learning Set.

Identificação automática das primeiras quebras em traços sísmicos por meio de uma rede neural direta

Apesar do avanço tecnológico ocorrido na prospecção sísmica, com a rotina dos levantamentos 2D e 3D, e o significativo aumento na quantidade de dados, a identificação dos tempos de chegada da onda sísmica direta (primeira quebra), que se propaga diretamente do ponto de tiro até a posição dos arranjos de geofones, permanece ainda dependente da avaliação visual do intérprete sísmico. O objetivo desta dissertação, insere-se no processamento sísmico com o intuito de buscar um método eficiente, tal que possibilite a simulação computacional do comportamento visual do intérprete sísmico, através da automação dos processos de tomada de decisão envolvidos na identificação das primeiras quebras em um traço sísmico. Visando, em última análise, preservar o conhecimento intuitivo do intérprete para os casos complexos, nos quais o seu conhecimento será, efetivamente, melhor aproveitado. Recentes descobertas na tecnologia neurocomputacional produziram técnicas que possibilitam a simulação dos aspectos qualitativos envolvidos nos processos visuais de identificação ou interpretação sísmica, com qualidade e aceitabilidade dos resultados. As redes neurais artificiais são uma implementação da tecnologia neurocomputacional e foram, inicialmente, desenvolvidas por neurobiologistas como modelos computacionais do sistema nervoso humano. Elas diferem das técnicas computacionais convencionais pela sua habilidade em adaptar-se ou aprender através de uma repetitiva exposição a exemplos, pela sua tolerância à falta de alguns dos componentes dos dados e pela sua robustez no tratamento com dados contaminados por ruído. O método aqui apresentado baseia-se na aplicação da técnica das redes neurais artificiais para a identificação das primeiras quebras nos traços sísmicos, a partir do estabelecimento de uma conveniente arquitetura para a rede neural artificial do tipo direta, treinada com o algoritmo da retro-propagação do erro. A rede neural artificial é entendida aqui como uma simulação computacional do processo intuitivo de tomada de decisão realizado pelo intérprete sísmico para a identificação das primeiras quebras nos traços sísmicos. A aplicabilidade, eficiência e limitações desta abordagem serão avaliadas em dados sintéticos obtidos a partir da teoria do raio.

Four-gene expression model predictive of lymph node metastases in oral squamous cell carcinoma

Background. Previous knowledge of cervical lymph node compromise may be crucial to choose the best treatment strategy in oral squamous cell carcinoma (OSCC). Here we propose a set four genes, whose mRNA expression in the primary tumor predicts nodal status in OSCC, excluding tongue. Material and methods. We identified differentially expressed genes in OSCC with and without compromised lymph nodes using Differential Display RT-PCR. Known genes were chosen to be validated by means of Northern blotting or real time RT-PCR (qRT-PCR). Thereafter we constructed a Nodal Index (NI) using discriminant analysis in a learning set of 35 patients, which was further validated in a second independent group of 20 patients. Results. Of the 63 differentially expressed known genes identified comparing three lymph node positive (pN+) and three negative (pN0) primary tumors, 23 were analyzed by Northern analysis or RT-PCR in 49 primary tumors. Six genes confirmed as differentially expressed were used to construct a NI, as the best set predictive of lymph nodal status, with the final result including four genes. The NI was able to correctly classify 32 of 35 patients comprising the learning group (88.6%; p = 0.009). Casein kinase 1alpha1 and scavenger receptor class B, member 2 were found to be up regulated in pN + group in contrast to small proline-rich protein 2B and Ras-GTPase activating protein SH3 domain-binding protein 2 which were upregulated in the pN0 group. We validated further our NI in an independent set of 20 primary tumors, 11 of them pN0 and nine pN+ with an accuracy of 80.0% (p = 0.012). Conclusions. The NI was an independent predictor of compromised lymph nodes, taking into the consideration tumor size and histological grade. The genes identified here that integrate our "Nodal Index" model are predictive of lymph node metastasis in OSCC.

Functional error modeling for uncertainty quantification in hydrogeology

Approximate models (proxies) can be employed to reduce the computational costs of estimating uncertainty. The price to pay is that the approximations introduced by the proxy model can lead to a biased estimation. To avoid this problem and ensure a reliable uncertainty quantification, we propose to combine functional data analysis and machine learning to build error models that allow us to obtain an accurate prediction of the exact response without solving the exact model for all realizations. We build the relationship between proxy and exact model on a learning set of geostatistical realizations for which both exact and approximate solvers are run. Functional principal components analysis (FPCA) is used to investigate the variability in the two sets of curves and reduce the dimensionality of the problem while maximizing the retained information. Once obtained, the error model can be used to predict the exact response of any realization on the basis of the sole proxy response. This methodology is purpose-oriented as the error model is constructed directly for the quantity of interest, rather than for the state of the system. Also, the dimensionality reduction performed by FPCA allows a diagnostic of the quality of the error model to assess the informativeness of the learning set and the fidelity of the proxy to the exact model. The possibility of obtaining a prediction of the exact response for any newly generated realization suggests that the methodology can be effectively used beyond the context of uncertainty quantification, in particular for Bayesian inference and optimization.

Young children's understanding of oddity: reducing complexity by simple oddity and "most different" strategies

Children aged between 3 and 7 years were taught simple and dimension-abstracted oddity discrimination using learning-set training techniques, in which isomorphic problems with varying content were presented with verbal explanation and feedback. Following the training phase, simple oddity (SO), dimension-abstracted oddity with one or two irrelevant dimensions, and non-oddity (NO) tasks were presented (without feedback) to determine the basis of solution. Although dimension-abstracted oddity requires discrimination based on a stimulus that is different from the others, which are all the same as each other on the relevant dimension, this was not the major strategy. The data were more consistent with use of a simple oddity strategy by 3- to 4-year-olds, and a most different strategy by 6- to 7-year-olds. These strategies are interpreted as reducing task complexity. (C) 2002 Elsevier Science Inc. All rights reserved.

Novel strategies in feedforward adaptation to a position-dependent perturbation

To investigate the control mechanisms used in adapting to position-dependent forces, subjects performed 150 horizontal reaching movements over 25 cm in the presence of a position-dependent parabolic force field (PF). The PF acted only over the first 10 cm of the movement. On every fifth trial, a virtual mechanical guide (double wall) constrained subjects to move along a straight-line path between the start and target positions. Its purpose was to register lateral force to track formation of an internal model of the force field, and to look for evidence of possible alternative adaptive strategies. The force field produced a force to the right, which initially caused subjects to deviate in that direction. They reacted by producing deviations to the left, into the force field, as early as the second trial. Further adaptation resulted in rapid exponential reduction of kinematic error in the latter portion of the movement, where the greatest perturbation to the handpath was initially observed, whereas there was little modification of the handpath in the region where the PF was active. Significant force directed to counteract the PF was measured on the first guided trial, and was modified during the first half of the learning set. The total force impulse in the region of the PF increased throughout the learning trials, but it always remained less than that produced by the PF. The force profile did not resemble a mirror image of the PF in that it tended to be more trapezoidal than parabolic in shape. As in previous studies of force-field adaptation, we found that changes in muscle activation involved a general increase in the activity of all muscles, which increased arm stiffness, and selectively-greater increases in the activation of muscles which counteracted the PF. With training, activation was exponentially reduced, albeit more slowly than kinematic error. Progressive changes in kinematics and EMG occurred predominantly in the region of the workspace beyond the force field. We suggest that constraints on muscle mechanics limit the ability of the central nervous system to employ an inverse dynamics model to nullify impulse-like forces by generating mirror-image forces. Consequently, subjects adopted a strategy of slightly overcompensating for the first half of the force field, then allowing the force field to push them in the opposite direction. Muscle activity patterns in the region beyond the boundary of the force field were subsequently adjusted because of the relatively-slow response of the second-order mechanics of muscle impedance to the force impulse.

Project Proposals Development – Entrepreneurship and Reflection in Teaching Information Technology

This work presents a model for development of project proposals by students as an approach to teaching information technology while promoting entrepreneurship and reflection. In teams of 3 to 5 participants, students elaborate a project proposal on a topic they have negotiated with each other and with the teacher. The project domain is related to the practical application of state-of-theart information technology in areas of substantial public interest or of immediate interest to the participants. This gives them ample opportunities for reflection not only on technical but also on social, economic, environmental and other dimensions of information technology. This approach has long been used with students of different years and programs of study at the Faculty of Mathematics and Informatics, Plovdiv University “Paisiy Hilendarski”. It has been found to develop all eight key competences for lifelong learning set forth in the Reference Framework and procedural skills required in real life.

Establishing Correspondences between Attribute Spaces and Complex Concept Spaces Using Meta-PGN Classifier

In this paper, we present one approach for extending the learning set of a classification algorithm with additional metadata. It is used as a base for giving appropriate names to found regularities. The analysis of correspondence between connections established in the attribute space and existing links between concepts can be used as a test for creation of an adequate model of the observed world. Meta-PGN classifier is suggested as a possible tool for establishing these connections. Applying this approach in the field of content-based image retrieval of art paintings provides a tool for extracting specific feature combinations, which represent different sides of artists' styles, periods and movements.