Search

Abstract: Genetic-based machine learning (GBML) systems, which employ evolutionary algorithms (EAs) as search mechanisms, evolve rule-based classification models to represent target concepts. Compared to Michigan-style GBML, Pittsburgh-style GBML is expected to achieve more compact solutions. It has been shown that standard recombination operators in EAs do not assure an effective evolutionary search to solve sophisticated problems that contain strong interactions between features. On the other hand, when dealing with real-world classification tasks, irrelevant features not only complicate the problem but also incur unnecessary matchings in GBML systems, which increase the computational cost a lot. To handle the two problems mentioned above in an integrated manner, a new Pittsburgh-style GBML system is proposed. In the proposed method, classifiers are generated and recombined at two levels. At the high level, classifiers are recombined by rule-wise uniform crossover operators since each classifier consists of a variable-size rule set. At the low level, single rules contained in classifiers are reproduced via sampling Bayesian networks that characterize the global statistical information extracted from promising rules found so far. Furthermore, according to the statistical information in the rule population, an embedded approach is presented to detect and remove redundant features incrementally following the evolution of rule population. Results of empirical evaluation show that the proposed method outperforms the original Pittsburgh-style GBML system in terms of classification accuracy while reducing the computational cost. Furthermore, the proposed method is also competitive to other non-evolutionary, highly used machine learning methods. With respect to the performance of feature reduction, the proposed embedded approach is able to deliver solutions with higher classification accuracy when removing the same number of features as other feature reduction techniques do. [Copyright &y& Elsevier]

Abstract: Anti-germ performance test is critical in the production of detergents. However, actual biochemical tests are often costly and time-consuming. In this paper, we present an Elman neural network-based model to predict detergents’ anti-germ performance and ingredient levels, respectively. The model made it much faster and cost effective than doing actual biochemical tests. We also present preprocessing methods that can reduce data conflicts while keeping the monotonicity on limited experimental data. The model can find out the relationship between ingredient levels and the corresponding anti-germ performance, which is not widely used in solving biochemical problems. The results of experiments are generated on the base of two detergent products for two types of bacteria, and appear reasonable according to natural rules. The prediction results show a high accuracy and fitting with the monotonicity rule mostly. [Copyright &y& Elsevier]

Abstract: Features play a fundamental role in sentiment classification. How to effectively select different types of features to improve sentiment classification performance is the primary topic of this paper. Ngram features are commonly employed in text classification tasks; in this paper, sentiment-words, substrings, substring-groups, and key-substring-groups, which have never been considered in sentiment classification area before, are also extracted as features. The extracted features are then compared and analyzed. To demonstrate generality, we use two authoritative Chinese data sets in different domains to conduct our experiments. Our statistical analysis of the experimental results indicate the following: (1) different types of features possess different discriminative capabilities in Chinese sentiment classification; (2) character bigram features perform the best among the Ngram features; (3) substring-group features have greater potential to improve the performance of sentiment classification by combining substrings of different lengths; (4) sentiment words or phrases extracted from existing sentiment lexicons are not effective for sentiment classification; (5) effective features are usually at varying lengths rather than fixed lengths. [Copyright &y& Elsevier]

Abstract: The stock market is considered as a high complex and dynamic system with noisy, non-stationary and chaotic data series. So it is widely acknowledged that stock price series modeling and forecasting is a challenging work. A significant amount of work has been done in this field, and in them, soft computing techniques have showed good performance. Generally most of these works can be divided into two categories. One is to predict the future trend or price; another is to construct decision support system which can give certain buy/sell signals. In this paper, we propose a new intelligent trading system based on oscillation box prediction by combining stock box theory and support vector machine algorithm. The box theory believes a successful stock buying/selling generally occurs when the price effectively breaks out the original oscillation box into another new box. In the system, two SVM estimators are first utilized to make forecasts of the upper bound and lower bound of the price oscillation box. Then a trading strategy based on the two bound forecasts is constructed to make trading decisions. In the experiment, we test the system on different stock movement patterns, i.e. bull, bear and fluctuant market, and investigate the training of the system and the choice of the time span of the price box. The experiments on 442 S&P500 components show a promising performance is achieved and the system dramatically outperforms buy-and-hold strategy. [Copyright &y& Elsevier]