Context-free and context-sensitive dynamics in recurrent neural networks

Continuous-valued recurrent neural networks can learn mechanisms for processing context-free languages. The dynamics of such networks is usually based on damped oscillation around fixed points in state space and requires that the dynamical components are arranged in certain ways. It is shown that qualitatively similar dynamics with similar constraints hold for a(n)b(n)c(n), a context-sensitive language. The additional difficulty with a(n)b(n)c(n), compared with the context-free language a(n)b(n), consists of 'counting up' and 'counting down' letters simultaneously. The network solution is to oscillate in two principal dimensions, one for counting up and one for counting down. This study focuses on the dynamics employed by the sequential cascaded network, in contrast to the simple recurrent network, and the use of backpropagation through time. Found solutions generalize well beyond training data, however, learning is not reliable. The contribution of this study lies in demonstrating how the dynamics in recurrent neural networks that process context-free languages can also be employed in processing some context-sensitive languages (traditionally thought of as requiring additional computation resources). This continuity of mechanism between language classes contributes to our understanding of neural networks in modelling language learning and processing.

Geospatial clustering in data-rich environments: Features and issues

Geospatial clustering must be designed in such a way that it takes into account the special features of geoinformation and the peculiar nature of geographical environments in order to successfully derive geospatially interesting global concentrations and localized excesses. This paper examines families of geospaital clustering recently proposed in the data mining community and identifies several features and issues especially important to geospatial clustering in data-rich environments.

Maximum Likelihood Estimation of Mixture Densities for Binned and Truncated Multivariate Data

Binning and truncation of data are common in data analysis and machine learning. This paper addresses the problem of fitting mixture densities to multivariate binned and truncated data. The EM approach proposed by McLachlan and Jones (Biometrics, 44: 2, 571-578, 1988) for the univariate case is generalized to multivariate measurements. The multivariate solution requires the evaluation of multidimensional integrals over each bin at each iteration of the EM procedure. Naive implementation of the procedure can lead to computationally inefficient results. To reduce the computational cost a number of straightforward numerical techniques are proposed. Results on simulated data indicate that the proposed methods can achieve significant computational gains with no loss in the accuracy of the final parameter estimates. Furthermore, experimental results suggest that with a sufficient number of bins and data points it is possible to estimate the true underlying density almost as well as if the data were not binned. The paper concludes with a brief description of an application of this approach to diagnosis of iron deficiency anemia, in the context of binned and truncated bivariate measurements of volume and hemoglobin concentration from an individual's red blood cells.

Model-based clustering in gene expression microarrays: An application to breast cancer data

In microarray studies, the application of clustering techniques is often used to derive meaningful insights into the data. In the past, hierarchical methods have been the primary clustering tool employed to perform this task. The hierarchical algorithms have been mainly applied heuristically to these cluster analysis problems. Further, a major limitation of these methods is their inability to determine the number of clusters. Thus there is a need for a model-based approach to these. clustering problems. To this end, McLachlan et al. [7] developed a mixture model-based algorithm (EMMIX-GENE) for the clustering of tissue samples. To further investigate the EMMIX-GENE procedure as a model-based -approach, we present a case study involving the application of EMMIX-GENE to the breast cancer data as studied recently in van 't Veer et al. [10]. Our analysis considers the problem of clustering the tissue samples on the basis of the genes which is a non-standard problem because the number of genes greatly exceed the number of tissue samples. We demonstrate how EMMIX-GENE can be useful in reducing the initial set of genes down to a more computationally manageable size. The results from this analysis also emphasise the difficulty associated with the task of separating two tissue groups on the basis of a particular subset of genes. These results also shed light on why supervised methods have such a high misallocation error rate for the breast cancer data.

Fatty liver characterization and classification by ultrasound

Steatosis, also known as fatty liver, corresponds to an abnormal retention of lipids within the hepatic cells and reflects an impairment of the normal processes of synthesis and elimination of fat. Several causes may lead to this condition, namely obesity, diabetes, or alcoholism. In this paper an automatic classification algorithm is proposed for the diagnosis of the liver steatosis from ultrasound images. The features are selected in order to catch the same characteristics used by the physicians in the diagnosis of the disease based on visual inspection of the ultrasound images. The algorithm, designed in a Bayesian framework, computes two images: i) a despeckled one, containing the anatomic and echogenic information of the liver, and ii) an image containing only the speckle used to compute the textural features. These images are computed from the estimated RF signal generated by the ultrasound probe where the dynamic range compression performed by the equipment is taken into account. A Bayes classifier, trained with data manually classified by expert clinicians and used as ground truth, reaches an overall accuracy of 95% and a 100% of sensitivity. The main novelties of the method are the estimations of the RF and speckle images which make it possible to accurately compute textural features of the liver parenchyma relevant for the diagnosis.

Recomendação de conteúdos multimédia

A quantidade e variedade de conteúdos multimédia actualmente disponíveis cons- tituem um desafio para os utilizadores dado que o espaço de procura e escolha de fontes e conteúdos excede o tempo e a capacidade de processamento dos utilizado- res. Este problema da selecção, em função do perfil do utilizador, de informação em grandes conjuntos heterogéneos de dados é complexo e requer ferramentas específicas. Os Sistemas de Recomendação surgem neste contexto e são capazes de sugerir ao utilizador itens que se coadunam com os seus gostos, interesses ou necessidades, i.e., o seu perfil, recorrendo a metodologias de inteligência artificial. O principal objectivo desta tese é demonstrar que é possível recomendar em tempo útil conteúdos multimédia a partir do perfil pessoal e social do utilizador, recorrendo exclusivamente a fontes públicas e heterogéneas de dados. Neste sen- tido, concebeu-se e desenvolveu-se um Sistema de Recomendação de conteúdos multimédia baseado no conteúdo, i.e., nas características dos itens, no historial e preferências pessoais e nas interacções sociais do utilizador. Os conteúdos mul- timédia recomendados, i.e., os itens sugeridos ao utilizador, são provenientes da estação televisiva britânica, British Broadcasting Corporation (BBC), e estão classificados de acordo com as categorias dos programas da BBC. O perfil do utilizador é construído levando em conta o historial, o contexto, as preferências pessoais e as actividades sociais. O YouTube é a fonte do histo- rial pessoal utilizada, permitindo simular a principal fonte deste tipo de dados - a Set-Top Box (STB). O historial do utilizador é constituído pelo conjunto de vídeos YouTube e programas da BBC vistos pelo utilizador. O conteúdo dos vídeos do YouTube está classificado segundo as categorias de vídeo do próprio YouTube, sendo efectuado o mapeamento para as categorias dos programas da BBC. A informação social, que é proveniente das redes sociais Facebook e Twit- ter, é recolhida através da plataforma Beancounter. As actividades sociais do utilizador obtidas são filtradas para extrair os filmes e séries que são, por sua vez, enriquecidos semanticamente através do recurso a repositórios abertos de dados interligados. Neste caso, os filmes e séries são classificados através dos géneros da IMDb e, posteriormente, mapeados para as categorias de programas da BBC. Por último, a informação do contexto e das preferências explícitas, através da classificação dos itens recomendados, do utilizador são também contempladas. O sistema desenvolvido efectua recomendações em tempo real baseado nas actividades das redes sociais Facebook e Twitter, no historial de vídeos Youtube e de programas da BBC vistos e preferências explícitas. Foram realizados testes com cinco utilizadores e o tempo médio de resposta do sistema para criar o conjunto inicial de recomendações foi 30 s. As recomendações personalizadas são geradas e actualizadas mediante pedido expresso do utilizador.

Scenarios Generation for Multi-Agent simulation of Electricity Markets based on Intelligent Data Analysis

This document presents a tool able to automatically gather data provided by real energy markets and to generate scenarios, capture and improve market players’ profiles and strategies by using knowledge discovery processes in databases supported by artificial intelligence techniques, data mining algorithms and machine learning methods. It provides the means for generating scenarios with different dimensions and characteristics, ensuring the representation of real and adapted markets, and their participating entities. The scenarios generator module enhances the MASCEM (Multi-Agent Simulator of Competitive Electricity Markets) simulator, endowing a more effective tool for decision support. The achievements from the implementation of the proposed module enables researchers and electricity markets’ participating entities to analyze data, create real scenarios and make experiments with them. On the other hand, applying knowledge discovery techniques to real data also allows the improvement of MASCEM agents’ profiles and strategies resulting in a better representation of real market players’ behavior. This work aims to improve the comprehension of electricity markets and the interactions among the involved entities through adequate multi-agent simulation.

Data Mining Approach to support the Generation of Realistic Scenarios for Multi-Agent simulation of Electricity Markets

This paper presents the Realistic Scenarios Generator (RealScen), a tool that processes data from real electricity markets to generate realistic scenarios that enable the modeling of electricity market players’ characteristics and strategic behavior. The proposed tool provides significant advantages to the decision making process in an electricity market environment, especially when coupled with a multi-agent electricity markets simulator. The generation of realistic scenarios is performed using mechanisms for intelligent data analysis, which are based on artificial intelligence and data mining algorithms. These techniques allow the study of realistic scenarios, adapted to the existing markets, and improve the representation of market entities as software agents, enabling a detailed modeling of their profiles and strategies. This work contributes significantly to the understanding of the interactions between the entities acting in electricity markets by increasing the capability and realism of market simulations.

Semantic relations extraction from unstructured information for domain ontologies enrichment

Based in internet growth, through semantic web, together with communication speed improvement and fast development of storage device sizes, data and information volume rises considerably every day. Because of this, in the last few years there has been a growing interest in structures for formal representation with suitable characteristics, such as the possibility to organize data and information, as well as the reuse of its contents aimed for the generation of new knowledge. Controlled Vocabulary, specifically Ontologies, present themselves in the lead as one of such structures of representation with high potential. Not only allow for data representation, as well as the reuse of such data for knowledge extraction, coupled with its subsequent storage through not so complex formalisms. However, for the purpose of assuring that ontology knowledge is always up to date, they need maintenance. Ontology Learning is an area which studies the details of update and maintenance of ontologies. It is worth noting that relevant literature already presents first results on automatic maintenance of ontologies, but still in a very early stage. Human-based processes are still the current way to update and maintain an ontology, which turns this into a cumbersome task. The generation of new knowledge aimed for ontology growth can be done based in Data Mining techniques, which is an area that studies techniques for data processing, pattern discovery and knowledge extraction in IT systems. This work aims at proposing a novel semi-automatic method for knowledge extraction from unstructured data sources, using Data Mining techniques, namely through pattern discovery, focused in improving the precision of concept and its semantic relations present in an ontology. In order to verify the applicability of the proposed method, a proof of concept was developed, presenting its results, which were applied in building and construction sector.

An evolutionary multi-objective optimization system for earthworks

Earthworks involve the levelling or shaping of a target area through the moving or processing of the ground surface. Most construction projects require earthworks, which are heavily dependent on mechanical equipment (e.g., excavators, trucks and compactors). Often, earthworks are the most costly and time-consuming component of infrastructure constructions (e.g., road, railway and airports) and current pressure for higher productivity and safety highlights the need to optimize earthworks, which is a nontrivial task. Most previous attempts at tackling this problem focus on single-objective optimization of partial processes or aspects of earthworks, overlooking the advantages of a multi-objective and global optimization. This work describes a novel optimization system based on an evolutionary multi-objective approach, capable of globally optimizing several objectives simultaneously and dynamically. The proposed system views an earthwork construction as a production line, where the goal is to optimize resources under two crucial criteria (costs and duration) and focus the evolutionary search (non-dominated sorting genetic algorithm-II) on compaction allocation, using linear programming to distribute the remaining equipment (e.g., excavators). Several experiments were held using real-world data from a Portuguese construction site, showing that the proposed system is quite competitive when compared with current manual earthwork equipment allocation.

Combining data mining and evolutionary computation for multi-criteria optimization of earthworks

Earthworks tasks aim at levelling the ground surface at a target construction area and precede any kind of structural construction (e.g., road and railway construction). It is comprised of sequential tasks, such as excavation, transportation, spreading and compaction, and it is strongly based on heavy mechanical equipment and repetitive processes. Under this context, it is essential to optimize the usage of all available resources under two key criteria: the costs and duration of earthwork projects. In this paper, we present an integrated system that uses two artificial intelligence based techniques: data mining and evolutionary multi-objective optimization. The former is used to build data-driven models capable of providing realistic estimates of resource productivity, while the latter is used to optimize resource allocation considering the two main earthwork objectives (duration and cost). Experiments held using real-world data, from a construction site, have shown that the proposed system is competitive when compared with current manual earthwork design.

Poker learner: modelação de jogadores através de data mining

Dissertação de mestrado integrado em Engenharia e Gestão de Sistemas de Informação

Concise server-wide causality management for eventually consistent data stores

Large scale distributed data stores rely on optimistic replication to scale and remain highly available in the face of net work partitions. Managing data without coordination results in eventually consistent data stores that allow for concurrent data updates. These systems often use anti-entropy mechanisms (like Merkle Trees) to detect and repair divergent data versions across nodes. However, in practice hash-based data structures are too expensive for large amounts of data and create too many false conflicts. Another aspect of eventual consistency is detecting write conflicts. Logical clocks are often used to track data causality, necessary to detect causally concurrent writes on the same key. However, there is a nonnegligible metadata overhead per key, which also keeps growing with time, proportional with the node churn rate. Another challenge is deleting keys while respecting causality: while the values can be deleted, perkey metadata cannot be permanently removed without coordination. Weintroduceanewcausalitymanagementframeworkforeventuallyconsistentdatastores,thatleveragesnodelogicalclocks(BitmappedVersion Vectors) and a new key logical clock (Dotted Causal Container) to provides advantages on multiple fronts: 1) a new efficient and lightweight anti-entropy mechanism; 2) greatly reduced per-key causality metadata size; 3) accurate key deletes without permanent metadata.

Metaphorisms in programming

This paper introduces the metaphorism pattern of relational specification and addresses how specification following this pattern can be refined into recursive programs. Metaphorisms express input-output relationships which preserve relevant information while at the same time some intended optimization takes place. Text processing, sorting, representation changers, etc., are examples of metaphorisms. The kind of metaphorism refinement proposed in this paper is a strategy known as change of virtual data structure. It gives sufficient conditions for such implementations to be calculated using relation algebra and illustrates the strategy with the derivation of quicksort as example.