Search

Gateways científicos trazem enormes benefícios para usuários finais, simplificando o acesso e ocultando a complexidade da infraestrutura de computação distribuída subjacente. O gateway científico de bioinformática, BioinfoPortal, por meio do seu middleware CSGrid, usufrui dos recursos heterogêneos do Santos Dumont. No entanto, a submissão de tarefas ainda exige um esforço significativo, no que tange à decisão da melhor configuração que leve a uma execução eficiente. O framework de aprendizado de máquina, em desenvolvimento, ao ser integrado ao gateway, viabilizará essa decisão. No presente trabalho apresentamos um estudo de desempenho com caso de estudo da bioinformática visando analisar o comportamento de variáveis de saída do slurm/sacct dado valores das variáveis de entrada obtidas da configuração de tarefas do SDumont, o que pôde ser realizado pela modelagem deste cenário como uma tarefa de classificação binária. Os nossos resultados indicam ser possível extrair regras e avaliar a influência das variáveis de entrada Bootstrap, Nó e Thread, sendo Bootstrap a mais significativa e aquela com mais peso para o sistema de recomendação de alocação de recursos no BioinfoPortal.

Person reidentification, i.e., retrieving a person of interest across several non-overlapping cameras, is a task that is far from trivial. Despite its great commercial value and wide range of applications (e.g., surveillance, intelligent environments, forensics, service robotics, marketing), it remains unsolved, even when the individuals do not change clothes during the recognition period. This paper provides an outlook on long-term person reidentification, an emerging research topic regarding when consecutive acquisitions of an individual can be found apart for days or even months, making such a task even more challenging. A long-term reidentification system using face recognition is presented to emphasize current techniques’ limitations. FCT (Fundação para a Ciência e a Tecnologia), under the project UIDB/05567/2020. info:eu-repo/semantics/publishedVersion

There has been a significant effort by the research community to address the problem of providing methods to organize documentation, with the help of Information Retrieval methods. In this paper, we present several experiments with stream analysis methods to explore streams of text documents. This paper also presents possible architectures of the Text Document Stream Organization, with the use of incremental algorithms like Incremental Sparse TF-IDF and Incremental Similarity. Our results show that with this architecture, significant improvements are achieved, regarding efficiency in grouping of similar documents. These improvements are important since it is of general knowledge that great amounts of text analysis are a high dimensional and complex subject of study, in the data analysis area.

Motivated by the challenges of Big Data, this paper presents an approximative algorithm to assess the Kolmogorov-Smirnov test. This goodness of fit statistical test is extensively used because it is non-parametric. This work focuses on the one-sample test, which considers the hypothesis that a given univariate sample follows some reference distribution. The method allows to evaluate the departure from such a distribution of a input stream, being space and time efficient. We show the accuracy of our algorithm by making several experiments in different scenarios: varying reference distribution and its parameters, sample size, and available memory. The performance of rival methods, some of which are considered the state-of-the-art, were compared. It is demonstrated that our algorithm is superior in most of the cases, considering the absolute error of the test statistic.

A popular computer puzzle, the game of Minesweeper requires its human players to have a mix of both luck and strategy to succeed. Analyzing these aspects more formally, in our research we assessed the feasibility of a novel methodology based on Reinforcement Learning as an adequate approach to tackle the problem presented by this game. For this purpose we employed Multi-Armed Bandit algorithms which were carefully adapted in order to enable their use to define autonomous computational players, targeting to make the best use of some game peculiarities. After experimental evaluation, results showed that this approach was indeed successful, especially in smaller game boards, such as the standard beginner level. Despite this fact the main contribution of this work is a detailed examination of Minesweeper from a learning perspective, which led to various original insights which are thoroughly discussed., Comment: To be published in IEEE Transactions on Games (ISSN 2475-1510 / 2475-1502)

This paper presents a framework for dealing with the problem of GPS trajectory classification in the context of the Rio de Janeiro’s public transit system (with hundreds or more classes). Such framework combines the versatile WiSARD classifier with a set of rules defined a priori, resulting in a neuro-symbolic learning system with very interesting characteristics and cutting-edge performance. We also verified the influence of different binarization methods in order to adapt raw data to WiSARD, which feeds from binary data only. These ideas were tested against a large data set of trajectories of buses from the city of Rio de Janeiro. The results confirm the practical applicability of those, since the accomplished performance was as good as that of other state-of-the-art rival methods in most test scenarios.

Open set recognition is a classification-like task. It is accomplished not only by the identification of observations which belong to targeted classes (i.e., the classes among those represented in the training sample which should be later recognized) but also by the rejection of inputs from other classes in the problem domain. The need for proper handling of elements of classes beyond those of interest is frequently ignored, even in works found in the literature. This leads to the improper development of learning systems, which may obtain misleading results when evaluated in their test beds, consequently failing to keep the performance level while facing some real challenge. The adaptation of a classifier for open set recognition is not always possible: the probabilistic premises most of them are built upon are not valid in a open-set setting. Still, this paper details how this was realized for WiSARD a weightless artificial neural network model. Such achievement was based on an elaborate distance-like computation this model provides and the definition of rejection thresholds during training. The proposed methodology was tested through a collection of experiments, with distinct backgrounds and goals. The results obtained confirm the usefulness of this tool for open set recognition.

Clustering is a powerful and versatile tool for knowledge discovery, able to provide a valuable information for data analysis in various domains. To perform this task based on streaming data is quite challenging: outdated knowledge needs to be disposed while the current knowledge is obtained from fresh data; since data are continuously flowing, strict efficiency constraints have to be met. This paper presents WCDS, an approach to this problem based on the WiSARD artificial neural network model. This model already had useful characteristics as inherent incremental learning capability and patent functioning speed. These were combined with novel features as an adaptive countermeasure to cluster imbalance, a mechanism to discard expired data, and offline clustering based on a pairwise similarity measure for WiSARD discriminators. In an insightful experimental evaluation, the proposed system had an excellent performance according to multiple quality standards. This supports its applicability for the analysis of data streams.

Credit analysis is a real-world classification problem where it is quite common to find datasets with a large amount of noisy data. State-of-the-art classifiers that employ error minimisation techniques, on the other hand, require a long time to converge, in order to achieve robustness. This paper explores ClusWiSARD, a clustering customisation of the WiSARD weightless neural network model, applied to two different credit analysis real-world problems. Experimental evidence shows that ClusWiSARD is very competitive with Support Vector Machine (SVM) w.r.t. accuracy, with the advantage of being capable of online learning. ClusWiSARD outperforms SVM in training time, by two orders of magnitude, and is slightly faster in test time.

To cluster a data stream is a more challenging task than its regular batch version, having stricter performance constraints. In this paper an approach to this problem is presented, based on WiSARD, a memory-based artificial neural network (ANN) model. This model functioning was reviewed and improved, in order to adapt it to this task. The experimental results obtained support the use of this system for the analysis of data streams in an informative way.

Open set recognition is, more than an interesting research subject, a component of various machine learning applications which is sometimes neglected: it is not unusual the existence of learning systems developed on the top of closed-set assumptions, ignoring the error risk involved in a prediction. This risk is strictly related to the location in feature space where the prediction has to be made, compared to the location of the training data: the more distant the training observations are, less is known, higher is the risk. Proper handling of this risk can be necessary in various situation where classification and its variants are employed. This paper presents an approach to open set recognition based on an elaborate distance-like computation provided by a weightless neural network model. The results obtained in the proposed test scenarios are quite interesting, placing the proposed method among the current best ones.

This work investigates the effect of different data structures on the performance and accuracy of VG-RAM-based classifiers. This weightless neural model is based on RAM nodes having very large address input, what suggests the use of special data structures in order to deal with space and time computational costs. Four different data structures are explored, including the classical one used in recent VG-RAM related literature, resulting in a novel and accurate yet fast setup.

One of the major data mining tasks is to cluster similar data, because of its usefulness, providing means of summarizing large ammounts of raw data into handy information. Clustering data streams is particularly challenging, because of the constraints imposed when dealing with this kind of input. Here we report our work, in which it was investigated the use of WiSARD discriminators as primary data synthesizing units. An analysis of StreamWiSARD, a new sliding-window stream data clustering system, the benefits and the drawbacks of its use and a comparison to other approaches are all presented.