2012
Autores
Gomes, TAF; Prudencio, RBC; Soares, C; Rossi, ALD; Carvalho, A;
Publicação
NEUROCOMPUTING
Abstract
Support Vector Machines (SVMs) have achieved very good performance on different learning problems. However, the success of SVMs depends on the adequate choice of the values of a number of parameters (e.g., the kernel and regularization parameters). In the current work, we propose the combination of meta-learning and search algorithms to deal with the problem of SVM parameter selection. In this combination, given a new problem to be solved, meta-learning is employed to recommend SVM parameter values based on parameter configurations that have been successfully adopted in previous similar problems. The parameter values returned by meta-learning are then used as initial search points by a search technique, which will further explore the parameter space. In this proposal, we envisioned that the initial solutions provided by meta-learning are located in good regions of the search space (i.e. they are closer to optimum solutions). Hence, the search algorithm would need to evaluate a lower number of candidate solutions when looking for an adequate solution. In this work, we investigate the combination of meta-learning with two search algorithms: Particle Swarm Optimization and Tabu Search. The implemented hybrid algorithms were used to select the values of two SVM parameters in the regression domain. These combinations were compared with the use of the search algorithms without meta-learning. The experimental results on a set of 40 regression problems showed that, on average, the proposed hybrid methods obtained lower error rates when compared to their components applied in isolation.
2012
Autores
Gomes, C; Almada Lobo, B; Borges, J; Soares, C;
Publicação
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Abstract
This paper presents a combination of optimization and data mining techniques to address the surgery scheduling problem. In this approach, we first develop a model to predict the duration of the surgeries using a data mining algorithm. The prediction model outcomes are then used by a mathematical optimization model to schedule surgeries in an optimal way. In this paper, we present the results of using three different data mining algorithms to predict the duration of surgeries and compare them with the estimates made by surgeons. The results obtained by the data mining models show an improvement in estimation accuracy of 36%.We also compare the schedules generated by the optimization model based on the estimates made by the prediction models against reality. Our approach enables an increase in the number of surgeries performed in the operating theater, thus allowing a reduction on the average waiting time for surgery and a reduction in the overtime and undertime per surgery performed. These results indicate that the proposed approach can help the hospital improve significantly the efficiency of resource usage and increase the service levels. © Springer-Verlag 2012.
2012
Autores
Ribeiro, G; Duivesteijn, W; Soares, C; Knobbe, A;
Publicação
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Abstract
Label Ranking problems are receiving increasing attention in machine learning. The goal is to predict not just a single value from a finite set of labels, but rather the permutation of that set that applies to a new example (e.g., the ranking of a set of financial analysts in terms of the quality of their recommendations). In this paper, we adapt a multilayer perceptron algorithm for label ranking. We focus on the adaptation of the Back-Propagation (BP) mechanism. Six approaches are proposed to estimate the error signal that is propagated by BP. The methods are discussed and empirically evaluated on a set of benchmark problems. © 2012 Springer-Verlag.
2012
Autores
Ferreira, CA; Gama, J; Costa, VS;
Publicação
COMPUTER AND INFORMATION SCIENCES II
Abstract
In this work we present XmuSer, a multi-relational framework suitable to explore temporal patterns available in multi-relational databases. xMuS er's main idea consists of exploiting frequent sequence mining, using an efficient and direct method to learn temporal patterns in the form of sequences. Grounded on a coding methodology and on the efficiency of sequence miners, we find the most interesting sequential patterns available and then map these findings into a new table, which encodes the multi-relational timed data using sequential patterns. In the last step of our framework, we use an ILP algorithm to learn a theory on the enlarged relational database that consists on the original multi-relational database and the new sequence relation. We evaluate our framework by addressing three classification problems.
2012
Autores
Moreira Matias, L; Ferreira, C; Gama, J; Mendes Moreira, J; De Sousa, JF;
Publicação
CEUR Workshop Proceedings
Abstract
Mining public transportation networks is a growing and explosive challenge due to the increasing number of information available. In highly populated urban zones, the vehicles can often fail the schedule. Such fails cause headway deviations (HD) between high-frequency bus pairs. In this paper, we propose to identify systematic HD which usually provokes the phenomenon known as Bus Bunching (BB). We use the PrefixSpan algorithm to accurately mine sequences of bus stops where multiple HD frequently emerges, forcing two or more buses to clump. Our results are promising: 1) we demonstrated that the BB origin can be modeled like a sequence mining problem where 2) the discovered patterns can easily identify the route schedule points to adjust in order to mitigate such events.
2012
Autores
Tork, HF; Oliveira, M; Gama, J; Malinowski, S; Morla, R;
Publicação
CEUR Workshop Proceedings
Abstract
Failure detection in telecommunication networks is a vital task. So far, several supervised and unsupervised solutions have been provided for discovering failures in such networks. Among them unsupervised approaches has attracted more attention since no label data is required [1]. Often, network devices are not able to provide information about the type of failure. In such cases, unsupervised setting is more appropriate for diagnosis. Among unsupervised approaches, Principal Component Analysis (PCA) has been widely used for anomaly detection literature and can be applied to matrix data (e.g. Users-Features). However, one of the important properties of network data is their temporal sequential nature. So considering the interaction of dimensions over a third dimension, such as time, may provide us better insights into the nature of network failures. In this paper we demonstrate the power of three-way analysis to detect events and anomalies in time-evolving network data.
The access to the final selection minute is only available to applicants.
Please check the confirmation e-mail of your application to obtain the access code.