Coevolutionary Computation Research Articles

Identifying Informatively Easy and Informatively Hard Concepts

In this article, we leverage ideas from the theory of coevolutionary computation to analyze interactions of students with problems. We introduce the idea of informatively easy or hard concepts. Our approach is different from more traditional analyses of problem difficulty such as item analysis in the sense that we consider Pareto dominance relationships within the multidimensional structure of student–problem performance data rather than average performance measures. This method allows us to uncover not just the problems on which students are struggling but also the variety of difficulties different students face. Our approach is to apply methods from the Dimension Extraction Coevolutionary Algorithm to analyze problem-solving logs of students generated when they use an online software tutoring suite for introductory computer programming called problets . The results of our analysis not only have implications for how to scale up and improve adaptive tutoring software but also have the promise of contributing to the identification of common misconceptions held by students and thus, eventually, to the construction of a concept inventory for introductory programming.

EvoParsons: design, implementation and preliminary evaluation of evolutionary Parsons puzzle

The automated design of a set of practice problems that co-adapts to a population of learners is a challenging problem. Fortunately, coevolutionary computation offers a rich framework to study interactions between two co-adapting populations of teachers and learners. This framework is also relevant in scenarios in which a population of students solve practice exercises that are synthesized by an evolutionary algorithm. In this study, we propose to leverage coevolutionary optimization to evolve a population of Parsons puzzles (a relatively recent new type of practice exercise for novice computer programmers). To this end, we start by experimenting with successive simulations that progressively introduce the characteristics that we anticipate finding in our target application. Using these simulations, we refine a set of guidelines that capture insights on how to successfully coevolve Parsons puzzles. These guidelines are then used to implement the proposed “EvoParsons” software, with which we conduct preliminary evaluations on real human students enrolled in an introductory Java programming course at the University of South Florida. We also propose several quantitative metrics to assess the quality of puzzles produced by EvoParsons. Both simulations and experiments establish the feasibility of evolving pedagogically relevant practice problems that cover most of the dimensions underlying the interactions between problems and students. In addition, a generation-by-generation detailed analysis of the evolving population of Parsons puzzles confirms the occurrence of incremental improvements that can be explained in pedagogical terms.

Coevolution of Mobile Malware and Anti-Malware

Mobile malware is one of today's greatest threats in computer security. Furthermore, new mobile malware is emerging daily that introduces new security risks. However, while existing security solutions generally protect mobile devices against known risks, they are vulnerable to as yet unknown risks. How anti-malware software reacts to new, unknown malicious software is generally difficult to predict. Therefore, anti-malware software is in continuous development in order to be able to detect new malware or new variants of existing malware. Similarly, as long as anti-malware software develops, malware writers also develop their malicious code by using various evasion strategies, such as obfuscation and encryption. This is the lifecycle of malicious and anti-malware software. In this paper, the use of evolutionary computation techniques are investigated, both for developing new variants of mobile malware which successfully evades anti-malware systems based on static analysis and for developing better security solutions against them automatically. A coevolutionary arms race mechanism has always been considered a potential candidate for developing a more robust system against new attacks and for system testing. To the best of the authors' knowledge, this paper is the first application of coevolutionary computation to address this problem.

A client/server-based parallel coevolutionary algorithm for parallel machines scheduling problem with penalties

A problem of n independent jobs with different ready times and due dates to be scheduled on m parallel machines which aims at minimizing the total tardiness penalty costs of all jobs is considered. The decomposition and combination characteristics of the problem are studied. Based on these two characteristics, a parallel coevolutionary algorithm (PCA) is proposed. The PCA is implemented in a client/server (C/S)-based parallel coevolutionary computation structure, in which the server is responsible for global evolution by task assignment and the client is responsible for local evolution by task ordering. According to this structure, a genetic algorithm and a virus evolutionary genetic algorithm are developed, which are executed on server and client, respectively. The simulated experiments are designed, the comparison results with adapted algorithms show that the proposed algorithm C/S-PCA has superior performances both in convergence speed and in searching optimality.

Analyzing Two-Settlement Electricity Market Equilibrium by Coevolutionary Computation Approach

Forward contracts play an important role for market power mitigation and risk hedging in electricity markets. In this paper, a two-settlement electricity market including a forward market and a spot market is formulated as a two-stage game. The linear supply function equilibrium (LSFE) model and Cournot model are used to model strategic bidding for the spot market, while the forward market is modeled by Cournot model. A coevolutionary genetic algorithm (CGA) is employed to determine the market equilibrium. The paper then examines the question whether generation companies (GenCos) would voluntarily enter forward markets due to economic inspiration and studies theoretically and numerically the factors which can affect the bidding behaviors of GenCos. The effectiveness of CGA in determining the market equilibrium is investigated and demonstrated based on two test examples. The studies show that GenCos' decisions on the participation in forward markets depend significantly on the type of competition that they have in the spot market. Moreover, GenCos' bidding behaviors in the forward market can be affected by their cost parameters and the slope of system demand function significantly.

Co-evolution of terrorist and security scenarios for water distribution systems

Identification of vulnerabilities of water distribution systems and identification of appropriate counter-measures are important components of homeland security. These are difficult and time consuming tasks. This paper provides a new approach to resolve these problems in complex infrastructure systems. It is based on the use of co-evolutionary computation for the generation of both terrorist and security scenarios. The basic concepts of co-evolutionary computation are briefly explained. The concept of co-evolutionary generation of terrorist and security scenarios is introduced in the context of a hypothetical water distribution system for a small town. A tool developed at George Mason University is used for a number of experiments that reveal a variety of emerging security patterns. The experiments show that these patterns may be helpful in effectively protecting the network.

A Study on Modeling Error Estimation for Mobile Robot Based on Coevolutionary Computation and Image Processing

Evolutionary Computation (EC) is used to minimize modeling errors between robotic movement in computer simulation and trajectories of an actual robot. Generally, this task is important and so difficult. This paper proposes the method to minimize the modeling error between robotic movements and simulation results using coevolutionary computations with image processing technique. In the proposed method, a video camera on the ceiling captures robot movement, actual robot trajectories are detected from captured images by image processing, and modeling errors are estimated. Results of the experiments using an actual robot confirmed the effectiveness of our proposal and showed that modeling errors are reduced effectively. The sections that follow detail the problems overcome and our projected work.

A cooperative constructive method for neural networks for pattern recognition

In this paper, we propose a new constructive method, based on cooperative coevolution, for designing automatically the structure of a neural network for classification. Our approach is based on a modular construction of the neural network by means of a cooperative evolutionary process. This process benefits from the advantages of coevolutionary computation as well as the advantages of constructive methods. The proposed methodology can be easily extended to work with almost any kind of classifier. The evaluation of each module that constitutes the network is made using a multiobjective method. So, each new module can be evaluated in a comprehensive way, considering different aspects, such as performance, complexity, or degree of cooperation with the previous modules of the network. In this way, the method has the advantage of considering not only the performance of the networks, but also other features. The method is tested on 40 classification problems from the UCI machine learning repository with very good performance. The method is thoroughly compared with two other constructive methods, cascade correlation and GMDH networks, and other classification methods, namely, SVM, C4.5, and k nearest-neighbours, and an ensemble of neural networks constructed using four different methods.

Analyzing Oligopolistic Electricity Market Using Coevolutionary Computation

This paper presents a new unified framework of electricity market analysis based on coevolutionary computation (CCEM) for both the one-shot and the repeated games of oligopolistic electricity markets. The standard Cournot model and the new Pareto improvement model are used. The linear and constant elasticity demand functions are considered. Case study shows that CCEM is highly efficient and can handle the nonlinear market models that are difficult to be handled by conventional methods. The framework presented in this paper can help to overcome the difficulties of demand function specification encountered by the Cournot models. CCEM is found to be an effective and powerful approach for electricity market analysis.

Genetic Recurrent Fuzzy System by Coevolutionary Computation With Divide-and-Conquer Technique

A genetic recurrent fuzzy system which automates the design of recurrent fuzzy networks by a coevolutionary genetic algorithm with divide-and-conquer technique (CGA-DC) is proposed in this paper. To solve temporal problems, the recurrent fuzzy network constructed from a series of recurrent fuzzy if-then rules is adopted. In the CGA-DC, based on the structure of a recurrent fuzzy network, the design problem is divided into the design of individual subrules, including spatial and temporal, and that of the whole network. Then, three populations are created, among which two are created for spatial and temporal subrules searches, and the other for the whole network search. Evolution of the three populations are performed independently and concurrently to achieve a good design performance. To demonstrate the performance of CGA-DC, temporal problems on dynamic plant control and chaotic system processing are simulated. In this way, the efficacy and efficiency of CGA-DC can be evaluated as compared with other genetic-algorithm-based design approaches.

Multi-objective cooperative coevolution of artificial neural networks (multi-objective cooperative networks)

In this paper we present a cooperative coevolutive model for the evolution of neural network topology and weights, called MOBNET. MOBNET evolves subcomponents that must be combined in order to form a network, instead of whole networks. The problem of assigning credit to the subcomponents is approached as a multi-objective optimization task. The subcomponents in a cooperative coevolutive model must fulfill different criteria to be useful, these criteria usually conflict with each other. The problem of evaluating the fitness on an individual based on many criteria that must be optimized together can be approached as a multi-criteria optimization problems, so the methods from multi-objective optimization offer the most natural way to solve the problem. In this work we show how using several objectives for every subcomponent and evaluating its fitness as a multi-objective optimization problem, the performance of the model is highly competitive. MOBNET is compared with several standard methods of classification and with other neural network models in solving four real-world problems, and it shows the best overall performance of all classification methods applied. It also produces smaller networks when compared to other models. The basic idea underlying MOBNET is extensible to a more general model of coevolutionary computation, as none of its features are exclusive of neural networks design. There are many applications of cooperative coevolution that could benefit from the multi-objective optimization approach proposed in this paper.

Fuzzy CoCo: a cooperative-coevolutionary approach to fuzzy modeling

Co-evolutionary algorithms have received increased attention in the past few years within the domain of evolutionary computation. In this paper, we combine the search power of co-evolutionary computation with the expressive power of fuzzy systems, and introduce a novel algorithm, called Fuzzy CoCo (fuzzy cooperative coevolution). We demonstrate the efficacy of Fuzzy CoCo by applying it to a hard, real-world problem - breast cancer diagnosis, obtaining the best results to date while expending less computational effort than previous processes. Analyzing our results, we derive guidelines for setting the algorithm parameters given a (hard) problem to solve. We hope Fuzzy CoCo proves to be a powerful tool in the fuzzy modeler toolkit.