Ricardo Bastos Cavalcante Prudêncio

Meta-learning approaches to selecting time series models

We present here an original work that applies meta-learning approaches to select models for time-series forecasting. In our work, we investigated two meta-learning approaches, each one used in a different case study. Initially, we used a single machine learning algorithm to select among two models to forecast stationary time series (case study I). Following, we used the NOEMON approach, a more recent work in the meta-learning area, to rank three models used to forecast time series of the M3-Competition (case study II). The experiments performed in both case studies revealed encouraging results.

Combining meta-learning and search techniques to select parameters for support vector machines

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.

Supervised link prediction in weighted networks

Link prediction is an important task in Social Network Analysis. This problem refers to predicting the emergence of future relationships between nodes in a social network. Our work focuses on a supervised machine learning strategy for link prediction. Here, the target attribute is a class label indicating the existence or absence of a link between a node pair. The predictor attributes are metrics computed from the network structure, describing the given pair. The majority of works for supervised prediction only considers unweighted networks. In this light, our aim is to investigate the relevance of using weights to improve supervised link prediction. Link weights express the ‘strength’ of relationships and could bring useful information for prediction. However, the relevance of weights for unsupervised strategies of link prediction was not always verified (in some cases, the performance was even harmed). Our preliminary results on supervised prediction on a co-authorship network revealed satisfactory results when weights were considered, which encourage us for further analysis.

Ranking and selecting clustering algorithms using a meta-learning approach

We present a novel framework that applies a meta-learning approach to clustering algorithms. Given a dataset, our meta-learning approach provides a ranking for the candidate algorithms that could be used with that dataset. This ranking could, among other things, support non-expert users in the algorithm selection task. In order to evaluate the framework proposed, we implement a prototype that employs regression support vector machines as the meta-learner. Our case study is developed in the context of cancer gene expression micro-array datasets.

Time Series Based Link Prediction

Link prediction is a task in Social Network Analysis that consists of predicting connections that are most likely to appear considering previous observed links in a social network. The majority of works in this area only performs the task by exploring the state of the network at a specific moment to make the prediction of new links, without considering the behavior of links as time goes by. In this light, we investigate if temporal information can bring any performance gain to the link prediction task. A traditional approach for link prediction uses a chosen topological similarity metric on non-connected pairs of nodes of the network at present time to obtain a score that is going to be used by an unsupervised or a supervised method for link prediction. Our approach initially consists of building time series for each pair of non-connected nodes by computing their similarity scores at different past times. Then, we deploy a forecasting model on these time series and use their forecasts as the final scores of the pairs. Our preliminary results using two link prediction methods (unsupervised and supervised) on co-authorship networks revealed satisfactory results when temporal information was considered.

A Modal Symbolic Classifier for selecting time series models

The selection of a good model for forecasting a time series is a task that involves experience and knowledge. Employing machine learning algorithms is a promising approach to acquiring knowledge in regards to this task. A supervised classification method originating from the symbolic data analysis field is proposed for the model selection problem. This method was applied in the task of selecting between two widespread models, and compared to other learning algorithms. To date, it has obtained the lowest classification errors among all the tested algorithms.

Proximity measures for link prediction based on temporal events

Link prediction is a well-known task from the Social Network Analysis field that deals with the occurrence of connections in a network. It consists of using the network structure up to a given time in order to predict the appearance of links in a close future. The majority of previous work in link prediction is focused on the application of proximity measures (e.g., path distance, common neighbors) to non-connected pairs of nodes at present time in order to predict new connections in the future. New links can be predicted for instance by ordering the pairs of nodes according to their proximity scores. A limitation usually observed in previous work is that only the current state of the network is used to compute the proximity scores, without taking any temporal information into account (i.e., a static graph representation is adopted). In this work, we propose a new proximity measure for link prediction based on the concept of temporal events. In our work, we defined a temporal event related to a pair of nodes according to the creation, maintenance or interruption of the relationship between the nodes in consecutive periods of time. We proposed an event-based score which is updated along time by rewarding the temporal events observed between the pair of nodes under analysis and their neighborhood. The assigned rewards depend on the type of temporal event observed (e.g., if a link is conserved along time, a positive reward is assigned). Hence, the dynamics of links as the network evolves is used to update representative scores to pairs of nodes, rewarding pairs which formed or preserved a link and penalizing the ones that are no longer connected. In the performed experiments, we evaluated the proposed event-based measure in different scenarios for link prediction using co-authorship networks. Promising results were observed when the proposed measure was compared to both static proximity measures and a time series approach (a more competitive method) that also deploys temporal information for link prediction.

A multiple kernel learning algorithm for drug-target interaction prediction

BackgroundDrug-target networks are receiving a lot of attention in late years, given its relevance for pharmaceutical innovation and drug lead discovery. Different in silico approaches have been proposed for the identification of new drug-target interactions, many of which are based on kernel methods. Despite technical advances in the latest years, these methods are not able to cope with large drug-target interaction spaces and to integrate multiple sources of biological information.ResultsWe propose KronRLS-MKL, which models the drug-target interaction problem as a link prediction task on bipartite networks. This method allows the integration of multiple heterogeneous information sources for the identification of new interactions, and can also work with networks of arbitrary size. Moreover, it automatically selects the more relevant kernels by returning weights indicating their importance in the drug-target prediction at hand. Empirical analysis on four data sets using twenty distinct kernels indicates that our method has higher or comparable predictive performance than 18 competing methods in all prediction tasks. Moreover, the predicted weights reflect the predictive quality of each kernel on exhaustive pairwise experiments, which indicates the success of the method to automatically reveal relevant biological sources.ConclusionsOur analysis show that the proposed data integration strategy is able to improve the quality of the predicted interactions, and can speed up the identification of new drug-target interactions as well as identify relevant information for the task.AvailabilityThe source code and data sets are available at www.cin.ufpe.br/~acan/kronrlsmkl/.

Mining Rules for the Automatic Selection Process of Clustering Methods Applied to Cancer Gene Expression Data

Different algorithms have been proposed in the literature to cluster gene expression data, however there is no single algorithm that can be considered the best one independently on the data. In this work, we applied the concepts of Meta-Learning to relate features of gene expression data sets to the performance of clustering algorithms. In our context, each meta-example represents descriptive features of a gene expression data set and a label indicating the best clustering algorithm when applied to the data. A set of such meta-examples is given as input to a learning technique (the meta-learner ) which is responsible to acquire knowledge relating the descriptive features and the best algorithms. In our work, we performed experiments on a case study in which a meta-learner was applied to discriminate among three competing algorithms for clustering gene expression data of cancer. In this case study, a set of meta-examples was generated from the application of the algorithms to 30 different cancer data sets. The knowledge extracted by the meta-learner was useful to understanding the suitability of each clustering algorithm for specific problems.

Search based constrained test case selection using execution effort

Software testing is essential to guarantee high quality products. However, it is a very expensive activity, particularly when manually performed. One way to cut down costs is by reducing the input test suites, which are usually large in order to fully satisfy the test goals. Yet, since large test suites usually contain redundancies (i.e., two or more test cases (TC) covering the same requirement/piece of code), it is possible to reduce them in order to respect time/people constraints without severely compromising coverage. In this light, we formulated the TC selection problem as a constrained search based optimization task, using requirements coverage as the fitness function to be maximized (quality of the resultant suite), and the execution effort (time) of the selected TCs as a constraint in the search process. Our work is based on the Particle Swarm Optimization (PSO) algorithm, which is simple and efficient when compared to other widespread search techniques. Despite that, besides our previous works, we did not find any other proposals using PSO for TC selection, neither we found solutions treating this task as a constrained optimization problem. We implemented a Binary Constrained PSO (BCPSO) for functional TC selection, and two hybrid algorithms integrating BCPSO with local search mechanisms, in order to refine the solutions provided by BCPSO. These algorithms were evaluated using two different real-world test suites of functional TCs related to the mobile devices domain. In the performed experiments, the BCPSO obtained promising results for the optimization tasks considered. Also, the hybrid algorithms obtained statistically better results than the individual search techniques.