Bartłomiej Śnieżyński

Comparison of strategy learning methods in Farmer–Pest problem for various complexity environments without delays

Abstract In this paper effectiveness of several agent strategy learning algorithms is compared in a new multi-agent Farmer–Pest learning environment. Learning is often utilized by multi-agent systems which can deal with complex problems by means of their decentralized approach. With a number of learning methods available, a need for their comparison arises. This is why we designed and implemented new multi-dimensional Farmer–Pest problem domain, which is suitable for benchmarking learning algorithms. This paper presents comparison results for reinforcement learning (SARSA) and supervised learning (Naive Bayes, C4.5 and Ripper). These algorithms are tested on configurations with various complexity with not delayed rewards. The results show that algorithm performances depend highly on the environment configuration and various conditions favor different learning algorithms.

Proof Searching Algorithm for the Logic of Plausible Reasoning

Logic of plausible reasoning (LPR) is a knowledge representation and inference theory which is based on human reasoning techniques. Formalism can be described as a labeled deductive system; therefore reasoning can be considered as looking for proofs of given formulas. The aim of the paper is to present a proof searching algorithm for the LPR. The algorithm would be a core element of any information system using LPR.

Learning in a multi-agent approach to a fish bank game

In this paper application of symbolic, supervised learning in a multi-agent system is presented. As an environment Fish Bank game is used. Agents represent players that manage fishing companies. Rule induction algorithm is applied to generate ship allocation rules. In this article system architecture and learning process are described and preliminary experimental results are presented. Results show that learning agent performance increases significantly when new experience is taken into account.

Converting a Naive Bayes Model into a Set of Rules

A knowledge representation based on the probability theory is currently the most popular way of handling uncertainty. However, rule based systems are still popular. Their advantage is that rules are usually more easy to interpret than probabilistic models. A conversion method would allow to exploit advantages of both techniques. In this paper an algorithm that converts Naive Bayes models into rule sets is proposed. Preliminary experimental results show that rules generated from Naive Bayes models are compact and accuracy of such rule-based classifiers are relatively high.

A Rules-to-Trees Conversion in the Inductive Database System VINLEN

Decision trees and rules are completing methods of knowledge representation. Both have advantages in some applications. Algorithms that convert trees to rules are common. In the paper an algorithm that converts rules to decision tree and its implementation in inductive database VINLEN is presented.

Intelligent Information System Based on Logic of Plausible Reasoning

The chapter presents a methodology for the application of a formalism of the Logic of Plausible Reasoning (LPR) to create knowledge about a specific problem area. In this case, the methodology has been related to the task of obtaining information about the innovative casting technologies. In the search for documents, LPR gives a much greater expressive power than the commonly used keywords. The discussion is illustrated with the results obtained using a pilot version of the original information tool. It also presents a description of intelligent information system based on the LPR and the results of tests on the functionality and performance parameters of the system.

Learning Symbolic User Models for Intrusion Detection: A Method and Initial Results

This research was supported in part by the UMCB/LUCITE #32 grant, and in part by the National Science Foundation under Grants No. IIS-0097476 and IIS-9906858.

Expert System with Web Interface Based on Logic of Plausible Reasoning

The paper presents an expert system based on Logic of Plausible Reasoning (LPR). This formalism reflects human ways of knowledge representation and reasoning. The knowledge is modeled using several kinds of formulas representing statements, hierarchies, similarities, dependencies and implications. Several types of inference patterns are defined. Knowledge uncertainty can be modeled. The paper is structured as follows. Research related to LPR is presented. Next, the formalism is introduced and a Web-based application, which was developed for this research, is described. Finally, a case study is presented – a prototype expert system which recommends a material and a technology for a casting process.

Integrating Applications Developed for Heterogeneous Platforms: Building an Environment for Criminal Analysts

In this paper we present original approach for integrating systems on an example of LINK and Mammoth – criminal analysis applications. Firstly, a problem of integration is described with short description of integrated applications. Secondly, some theoretical information about integration is depicted. Paper continues with explanation of approach chosen for LINK and Mammoth integration and presents evaluation of achieved result. Eventually some final thoughts are stated.

Training Example Generation Method for Supervised Learning Agents in Sequential Scenarios

Abstract In this paper we propose a method of training example generation from agents experience, which is suitable for sequential sce- narios. The experience consists of the agents observations and its action records. Examples generated are used by the agent to learn a classifier, which is used to make decisions about its strategy in the following problem instances. The method is tested in a Sovereign environment, which is an economics simulation created to test agent-based learning. Experimental results show that an agent using the proposed methods is able to learn and achieves better results than random and heuristic agents.