Abstract

Abstract

Abstract
Document classification has already been widely studied. In fact, some studies compared feature selection techniques or feature space transformation whereas some others compared the performance of different algorithms. Recently, following the rising interest towards the Support Vector Machine, various studies showed that SVM outperforms other classification algorithms. So should we just not bother about other classification algorithms and opt always for SVM ? We have decided to investigate this issue and compared SVM to kNN and naive Bayes on binary classification tasks. An important issue is to compare optimized versions of these algorithms, which is what we have done. Our results show all the classifiers achieved comparable performance on most problems. One surprising result is that SVM was not a clear winner, despite quite good overall performance. If a suitable preprocessing is used with kNN, this algorithm continues to achieve very good results and scales up well with the number of documents, which is not the case for SVM. As for naive Bayes, it also achieved good performance.

Abstract
Inductive Logic Programming (ILP) is concerned with construction of logic programs from examples. It shares many concerns of Machine Learning (ML), but is committed to logic. As logic can help to provide a basis for elaborating such a methodology for learning, the area of ILP has attracted a wide attention of many researchers1. This paper reviews some of the methods and techniques in ML that exploit logic.

Abstract
In recent years various methods from the field of artificial intelligence (AI) have been applied to economic problems. The subarea of multiagent systems (MAS) is particularly useful as it enables to simulate individuals or organizations and various interactions among them. In this paper we investigate a scenario with a set of agents, each belonging to a certain sector of activity (e.g. agriculture, clothing, health sector etc.). The agents produce, consume goods or services in their area of activity. Besides, our model includes also the resource of free time. The goods and resources are exchanged on a market governed by auction, which determines the prices of all goods. We discuss the problem of developing an adaptive producer that exploits reward-based learning. This facet enables the agent to exploit previous information gathered and adapt its production to the current conditions. We describe a set of experiments that show how such information can be gathered and explored in decision making. Besides, we describe a scheme that we plan to adopt in a full-fledged experiments in near future.

Abstract