Search

Multi-label classification is a challenging problem which demands new knowledge discovery methods. This paper presents a Grammar-Guided Genetic Programming algorithm for solving multi-label classification problems using IF-THEN classification rules. This algorithm, called G3P-ML, is evaluated and compared to other multi-label classification techniques in different application domains. Computational experiments show that G3P-ML often obtains better results than other algorithms while achieving a lower number of rules than the other methods.

In the last years, the learning from multi-label data has attracted significant attention from a lot of researchers, motivated from an increasing number of modern applications that contain this type of data. Several methods have been proposed for solving this problem, however how to make feature weighting on multi-label data is still lacking in the literature. In multi-label data, each data point can be attributed to multiple labels simultaneously, thus a major difficulty lies in the determinations of the features useful for all multi-label concepts. In this paper, a new method for feature weighting in multi-label learning area is presented, based on the principles of the well-known ReliefF algorithm. The experimental stage shows the effectiveness of the proposal.

This paper deals with the problem of discovering subgroups in data by means of a grammar guided genetic programming algorithm, each subgroup including a set of related patterns. The proposed algorithm combines the requirements of discovering comprehensible rules with the ability of mining expressive and flexible solutions thanks to the use of a context-free grammar. A major characteristic of this algorithm is the small number of parameters required, so the mining process is easy for end-users. The algorithm proposed is compared with existing subgroup discovery evolutionary algorithms. The experimental results reveal the excellent behaviour of this algorithm, discovering comprehensible subgroups and behaving better than the other algorithms. The conclusions obtained were reinforced through a series of non-parametric tests.

WEKA has recently become a very referenced DM tool. In spite of all the functionality it provides, it does not include any framework for the development of evolutionary algorithms. An evolutionary computation framework is JCLEC, which has been successfully employed for developing several EAs. The combination of both may lead in a mutual benefit. Thus, this paper proposes an intermediate layer to connect WEKA with JCLEC. It also presents a study case which samples the process of including a JCLEC's EA into WEKA.

In this paper a Grammar Guided Genetic Programmingbased method for the learning of rule-based classification systems is proposed. The method learns disjunctive normal form rules generated by means of a context-free grammar. The individual constitutes a rule based decision list that represents the full classifier. To overcome the problem of computational time of this system, it parallelizes the evaluation phase reducing significantly the computation time. Moreover, different operator genetics are designed to maintain the diversity of the population and get a compact set of rules. The results obtained have been validated by the use of non-parametric statistical tests, showing a good performance in terms of accuracy and interpretability.

In this paper we present a novel algorithm, named GBAP, that jointly uses automatic programming with ant colony optimization for mining classification rules. GBAP is based on a context-free grammar that properly guides the search process of valid rules. Furthermore, its most important characteristics are also discussed, such as the use of two different heuristic measures for every transition rule, as well as the way it evaluates the mined rules. These features enhance the final rule compilation from the output classifier. Finally, the experiments over 17 diverse data sets prove that the accuracy values obtained by GBAP are pretty competitive and even better than those resulting from the top Ant-Miner algorithm.

Genetic Programming is very efficient in problem solving compared to other proposals but its performance is very slow when the size of the data increases This paper proposes a model for multi-threaded Genetic Programming classification evaluation using a NVIDIA CUDA GPUs programming model to parallelize the evaluation phase and reduce computational time Three different well-known Genetic Programming classification algorithms are evaluated using the parallel evaluation model proposed Experimental results using UCI Machine Learning data sets compare the performance of the three classification algorithms in single and multithreaded Java, C and CUDA GPU code Results show that our proposal is much more efficient.

The present work expounds a preliminary work of a genetic programming algorithm to deal with multi-label classification problems The algorithm uses Gene Expression Programming and codifies a classification rule into each individual A niching technique assures diversity in the population The final classifier is made up by a set of rules for each label that determines if a pattern belongs or not to the label The proposal have been tested over several domains and compared with other multi-label algorithms and the results shows that it is specially suitable to handle with nominal data sets.

An innovative technique based on multi-objective grammar guided genetic programming (MOG3P-MI) is proposed to detect the most relevant activities that a student needs to pass a course based on features extracted from logged data in an education web-based system. A more flexible representation of the available information based on multiple instance learning is used to prevent the appearance of a great number of missing values. Experimental results with the most relevant proposals in multiple instance learning in recent years demonstrate that MOG3P-MI successfully improves accuracy by finding a balance between specificity and sensitivity values. Moreover, simple and clear classification rules which are markedly useful to identify the number, type and time of activities that a student should do within the web system to pass a course are provided by our proposal.

This paper presents an evolutionary algorithm using G3P (Grammar Guided Genetic Programming) for mining association rules in different real-world databases This algorithm, called G3PARM, uses an auxiliary population made up of its best individuals that will then act as parents for the next generation The individuals are defined through a context-free grammar and it allows us to obtain datatype-generic and valid individuals We compare our approach to Apriori and FP-Growth algorithms and demonstrate that our proposal obtains rules with better support, confidence and coverage of the dataset instances Finally, a preliminary study is also introduced to compare the scalability of our algorithm Our experimental studies illustrate that this approach is highly promising for discovering association rules in databases.

This paper develops a first comparative study of multi- objective algorithms in Multiple Instance Learning (MIL) applications. These algorithms use grammar-guided genetic programming, a robust classification paradigm which is able to generate understandable rules that are adapted to work with the MIL framework. The algorithms obtained are based on the most widely used and compared multi-objective evolutionary algorithms. Thus, we design and implement SPG3P-MI based on the Strength Pareto Evolutionary Algorithm, NSG3P-MI based on the Non-dominated Sorting Genetic Algorithm and MOGLG3P-MI based on the Multi-objective genetic local search. These approaches are tested with different MIL applications and compared to a previous single-objective grammar-guided genetic programming proposal. The results demonstrate the excellent performance of multi-objective approaches in achieving accurate models and their ability to generate comprehensive rules in the knowledgable discovery process.

The aim of this paper is to present a new tool of multiple instance learning which is designed using a grammar based genetic programming (GGP) algorithm. We study its application in Web Mining framework to identify web pages interesting for the users. This new tool called GGP-MI algorithm is evaluated and compared with other available algorithms which extend a well-known neighborhood based algorithm (k-nearest neighbour algorithm) to multiple instance learning. Computational experiments show that, the GGP-MI algorithm obtains competitive results, solves problems of other algorithms, such as sparsity and scalability and adds comprehensibility and clarity in the knowledge discovery process.

This paper introduces the use of multi-objective evolutionary algorithms in multiple instance learning. In order to achieve this purpose, a multi-objective grammar-guided genetic programming algorithm (MOG3P-MI) has been designed. This algorithm has been evaluated and compared to other existing multiple instance learning algorithms. Research on the performance of our algorithm is carried out on two well-known drug activity prediction problems, Musk and Mutagenesis, both problems being considered typical benchmarks in multiple instance problems. Computational experiments indicate that the application of the MOG3P-MI algorithm improves accuracy and decreases computational cost with respect to other techniques.

This paper presents a cyclical methodology for the continuous improvement of e-learning courses using data mining techniques applied to education. For this purpose, a specific data mining tool has been developed, which discovers relevant relationships between data about how students use a course. Unlike others data mining approaches applied to education, which focus on the student, this method is aimed professors and how to help them improve the structure and contents of an e-learning course by making recommendations. We also use a rule discovery algorithm without parameters in order to be easily used by non-expert users in data mining. The results of experimental tests performed on an online course are also presented, demonstrating the usefulness of the proposed methodology and algorithm.

In this work we develop and compare multi-parent crossover operators based on the extraction of characteristics from the best individuals in the population (average, median, standard deviation and quantiles). These statistics evolve in parallel with the algorithm. The proposed operators are used in combination with a real-coded genetic algorithm for the evolution of polynomial functions to solve microbial growth problems. Their performance is compared to other crossover operators for real-coded genetic algorithms. Both the prediction errors made in the modelling of systems and the objectivity and speed in the identification of models show the viability of this type of models that mix base functions with evolutionary computation.