Murillo G. Carneiro

Network-based data classification: combining K-associated optimal graphs and high-level prediction

BackgroundTraditional data classification techniques usually divide the data space into sub-spaces, each representing a class. Such a division is carried out considering only physical attributes of the training data (e.g., distance, similarity, or distribution). This approach is called low-level classification. On the other hand, network or graph-based approach is able to capture spacial, functional, and topological relations among data, providing a so-called high-level classification. Usually, network-based algorithms consist of two steps: network construction and classification. Despite that complex network measures are employed in the classification to capture patterns of the input data, the network formation step is critical and is not well explored. Some of them, such as K-nearest neighbors algorithm (KNN) and ε-radius, consider strict local information of the data and, moreover, depend on some parameters, which are not easy to be set.MethodsWe propose a network-based classification technique, named high-level classification on K-associated optimal graph (HL-KAOG), combining the K-associated optimal graph and high-level prediction. In this way, the network construction algorithm is non-parametric, and it considers both local and global information of the training data. In addition, since the proposed technique combines low-level and high-level terms, it classifies data not only by physical features but also by checking conformation of the test instance to formation pattern of each class component. Computer simulations are conducted to assess the effectiveness of the proposed technique.ResultsThe results show that a larger portion of the high-level term is required to get correct classification when there is a complex-formed and well-defined pattern in the data set. In this case, we also show that traditional classification algorithms are unable to identify those data patterns. Moreover, computer simulations on real-world data sets show that HL-KAOG and support vector machines provide similar results and they outperform well-known techniques, such as decision trees and K-nearest neighbors.ConclusionsThe proposed technique works with a very reduced number of parameters and it is able to obtain good predictive performance in comparison with traditional techniques. In addition, the combination of high level and low level algorithms based on network components can allow greater exploration of patterns in data sets.

Dimensionality reduction with the k-associated optimal graph applied to image classification

In this paper, we aim to study the usage of different network formation methods into a graph embedding framework to perform supervised dimensionality reduction. Images are often high-dimensional patterns, and dimensionality reduction can enhance processing and also increase classification accuracy. Specifically, our technique maps images into networks and constructs two network adjacency matrices to convey information about intra-class components and inter-class penalty connections. Both matrices are inserted into an optimization framework in order to achieve a projection vector that is used to project high-dimension data samples into a low-dimensional space. One advantage of the technique is that no parameter is required, that is, there is no need to select a model for the input data. Applications on handwritten digits recognition are performed, and the proposed technique is compared to some classical network formation methods. Numerical results show the approach is promising.

Network-based supervised data classification by using an heuristic of ease of access

We propose a new supervised classification technique which considers the ease of access of unlabeled instances to training classes through an underlying network. The training data set is used to construct a network, in which instances (nodes) represent the states that a random walker visits, and the network link structure is modified by performing a link weight composition between the unlabeled instance bias and the initial network link weights. Different from traditional classification heuristics, which divide the training data set into subspaces, the proposed scheme uses random walk limiting probabilities to measure the limiting state transitions among training nodes. An unlabeled instance receives the label of the class that is most easily reached by the random walker, that is, the limiting transition to that class is large. Simulation results suggest that the proposed technique is comparable to some well-known classification techniques.

K-associated optimal network for graph embedding dimensionality reduction

In machine learning, dimensionality reduction aims at reducing the dimension of the input data in order to achieve a small set of features that keeps the most important original relationships among data samples. In this paper, we investigate the usage of a non-parametric network formation algorithm into a graph embedding framework to perform supervised dimensionality reduction. Specifically, our technique maps data into networks and constructs two network adjacency matrices which convey information about intra-class components and inter-class penalty connections. Both matrices are inserted into an optimization framework in order to achieve a projection vector that is used to project high-dimension data samples into a low-dimensional space. One advantage of the technique is that no parameter is required, that is, there is no need to select a model for the input data. Computer simulations on real-world data sets have been performed to compare the proposed technique to some classical network formation methods such as k-NN and e-radius, and to well-known dimensionality reduction algorithms such as PCA and LDA. Statistical tests have shown that our approach outperforms those algorithms.

High Level Classification Totally Based on Complex Networks

Differently from traditional machine learning techniques applied to data classification, high level classification considers not only the physical features of the data (distance, similarity or distribution), but also the pattern formation of the data. In this latter case, a set of complex network measures are employed because of their abilities to capture spatial, functional and topological relations. Although high level techniques offer powerful features, good classification performance is usually obtained by combining them with some low level algorithms, which, in turn, reduces the efficiency of the overall technique. A priori, the reason is that low level and high level techniques provide different visions of classification. In this way, one cannot simply substitute another. This paper presents a data classification technique in which low level and high level classifications are embedded in a unique scheme, i.e., the proposed technique does not need a separated low level technique. The novelty is the use of a simple and recently proposed complex network measure, named component efficiency. Thus, our algorithm computes the efficiency of information exchanging among vertices in each component and the resulting values are used to drive the classification of the new instances i.e., a new instance will be classified into one of the components (class), in which their local features are in conformity with the insertion of the new instance. The experiments performed with artificial and real-world data sets show our approach totally based on complex networks is promising and it provides better results than some traditional classification techniques.

SCAS-IS: Knowledge Extraction and Reuse in Multiprocessor Task Scheduling Based on Cellular Automata

Static Task Scheduling Problem (STSP) in multiprocessors is a NP-Complete problem. Cellular Automata (CA) have been recently proposed to solve STSP. The main feature of CA-based models to STSP is the extraction of knowledge while scheduling an application and its subsequent reuse in other instances. Previous works showed this approach is promising. However some desirable features have not been successfully exploited yet, such as: (i) the usage of an arbitrary number of processors, (ii) the massive parallelism inherent to CA and (iii) the reuse of evolved rules with competitive results. This paper presents a new model called SCAS-IS (Synchronous Cellular Automata Scheduler with Initialization Strategies). Its major innovation is the employment of fixed initialization strategies to start up CA dynamics. Parallel program graphs found in literature and others randomly generated were used to test the new model. Results show SCAS-IS overcame related models both in make span obtained as computational performance. It is also competitive with meta-heuristics.

A Hybrid Strategy to Evolve Cellular Automata Rules with a Desired Dynamical Behavior Applied to the Task Scheduling Problem

Cellular automata (CA) are discrete dynamical systems that generate complex and unpredictable behaviors. CA can exhibit a rich variety of behaviors from ordered to chaotic dynamics. An important issue in several applications is to control this dynamic in order to extract the best performance of CA rules. In the CA-based task scheduling domain, a partial answer is given by recent works that investigate two approaches named µ and ρ to evolve CA rules through a standard genetic algorithm, avoiding an undesirable dynamical behavior denoted by long-cycle and chaotic rules. Both approaches have been shown able to find CA rules with adequate dynamical behavior. However, each one presented its particularities: µ was stronger to avoid long-cycle rules and ρ obtains more refined rules (fixed-point behavior). In the present work, we investigate a new mixed approach named µρ in which the good characteristics of µ and ρ are preserved.

A scheme for high level data classification using random walk and network measures

Abstract Supervised classification techniques are known to exploit physical information of the analysed data, such as similarity, distribution and other low level features. Despite the relevance of such features, recent works have showed that a higher variety of patterns can be detected by combining low level and high level features. In this paper, it is proposed a supervised classification technique which applies limiting probabilities of the random walk theory over underlying networks constructed from input labeled data. The appealing feature of the proposed approach is that the adjacency matrix which carries both physical and structural information about the data. Structural information are given by features extracted from network connections. The class of a given unlabeled sample is estimated by a heuristic called ease of access , which is measured by the random walk process over the adjacency matrix. Such approach makes the technique quite general as one can put distinct data measures of interest in the connection matrix of the underlying data network to guide the random walker. Specifically, we show examples of combining low and high level features in the proposed classification scheme. Simulation results using artificial and real data sets suggest that the proposed technique is not only competitive with current and established classification techniques, but it also can reveal intrinsic structural patterns formed by the input data.

Network structural optimization based on swarm intelligence for highlevel classification.

While most part of the complex network models are described in function of some growth mechanism, the optimization of a goal or certain characteristics can be desirable for some problems. This paper investigates structural optimization of networks in the highlevel classification context, where the classification produced by a traditional classifier is combined with the classification provided by complex network measures. Using the recently proposed social learning particle swarm optimization (SL-PSO), a bio-inspired optimization framework, which is responsible to build up the network and adjust the parameters of the hybrid model while conducting the optimization of a quality function, is proposed. Experiments on two real-world problems, the Handwritten Digits Recognition and the Semantic Role Labeling (SRL), were performed. In both problems, the optimization framework is able to improve the classification given by a state-of-the-art algorithm to SRL. Furthermore, the optimization framework proposed here can be extended to other machine learning tasks.