Jae C. Oh

Modern Approaches in Applied Intelligence

The two volume set LNAI 6703 and LNAI 6704 constitutes the thoroughly refereed conference proceedings of the 24th International Conference on Industrial Engineering and Other Applications of Applied Intelligend Systems, IEA/AIE 2011, held in Syracuse, NY, USA, in June/July 2011. The total of 92 papers selected for the proceedings were carefully reviewed and selected from 206 submissions. The papers cover a wide number of topics including feature extraction, discretization, clustering, classification, diagnosis, data refinement, neural networks, genetic algorithms, learning classifier systems, Bayesian and probabilistic methods, image processing, robotics, navigation, optimization, scheduling, routing, game theory and agents, cognition, emotion, and beliefs

Thermal Faults Modeling Using a RC Model with an Application to Web Farms

Todays CPUs consume a significant amount of power and generate a high amount of heat, requiring an active cooling system to support reliable operations. In case of cooling system failures, these CPUs can reduce clock speed to prevent damage due to overheating. Unfortunately, when these CPUs are used in a real-time system, a clock control based on frequency-throttling can cause missed deadlines. In this paper, we first develop and validate a system-wide thermal model that can account for various thermal fault types such as failure of a CPU fan, faults in the case fan and air-conditioning malfunctions. Then we validate the thermal model through experimentation and measurements in AMD Linux boxes. Our soft real-time power-aware load-distribution algorithm for data centers incorporates a thermal model to minimize the number of missed deadlines that can be caused by thermal faults. We implemented the algorithm in a webserver farm simulator to test the efficacy of thermal-aware load-balancing. Our results show that the new algorithm helps keep CPU temperatures within the desired thermal envelope, even in the presence of thermal faults. When thermal faults occur, our algorithm improves the QoS, at the expense of higher energy consumption.

Image learning classifier system using genetic algorithms

The authors examine aspects of machine learning by classifier systems that use genetic algorithms. In particular, adaptive image learning and classification are considered. Standard classifier systems are not well suited for seeking out multiple goals as is necessary in image learning and classification problems. To improve the performance of standard classifier systems for the image learning task, several modifications are suggested. The modifications result in a far better performance for classifier system on the ImageLearn domain.<<ETX>>

A Game Theoretic Framework for Community Detection

The mainstream approach for community detection focuses on the optimization of a metric that measures the quality of a partition over a given network. Optimizing a global metric is akin to community assignment by a centralized decision maker. In liu of global optimization, we treat each node as a player in a hedonic game and focus on their ability to form fair and stable community structures. Application on real-world networks and a well-known benchmark demonstrates that our approach produces better results than modularity optimization.

A model for recursive propagations of reputations in social networks

We model the emergence and propagation of reputations in social networks with a novel distributed algorithm. In social networks, reputations of agents (nodes) are emerged and propagated through interactions among the agents and through intrinsic and extrinsic consensus (voting) among neighbors influenced by the network topology. Our algorithm considers the degree information of nodes and of their neighbors to combine consensus in order to model how reputations travel within the network. In our algorithm, each node updates reputations on its neighbors by considering past interactions, computing the velocity of the interactions to measure how frequent the interactions have been occurring recently, and adjusting the feedback values according to the velocity of the interaction. The algorithm also captures the phenomena of accuracy of reputations decaying over time if interactions have not occurred recently. We present two contributions through experiments: (1) We show that an agents reputation value is influenced by the position of the agent in the network and the neighboring topology; (2) We also show that our algorithm can compute more accurate reputations than existing algorithms especially when the topological information matters. The experiments are conducted in random social networks and Autonomous Systems Networks to find malicious nodes.

An Approach to Dominant Resource Fairness in Distributed Environment

We study the multi-type resource allocation problem in distributed computing environment. Current approaches that guarantee the conditions of Dominant Resource Fairness DRF are centralized algorithms. However, as P2P cloud systems gain more popularity, distributed algorithms that satisfy conditions of DRF are in demand. So we propose a distributed algorithm that mostly satisfies DRF conditions. According to our simulation results, our distributed dominant resource fairness algorithm outperforms a naive distributed extension of DRF.

Developing Concepts in Applied Intelligence

The series Studies in Computational Intelligence (SCI) publishes new developments and advances in the various areas of computational intelligence quickly and with a high quality. The intent is to cover the theory, applications, and design methods of computational intelligence, as embedded in the fields of engineering, computer science, physics and life science, as well as the methodologies behind them. The series contains monographs, lecture notes and edited volumes in computational intelligence spanning the areas of neural networks, connectionist systems, genetic algorithms, evolutionary computation, artificial intelligence, cellular automata, self-organizing systems, soft computing, fuzzy systems and hybrid intelligent systems. Critical to both contributors and readers are the short publication time and world-wide distribution this permits a rapid and broad dissemination of research results.The field of Artificial Intelligence developed important concepts for simulating human intelligence. Its sister field, Applied Intelligence, has focused on techniques for developing intelligent systems for solving real life problems in all disciplines including science, social science, art, engineering, and finance. The objective of the International Conference on Industrial, Engineering & Other Applications of Applied Intelligent Systems (IEA-AIE) is to promote and disseminate research in Applied Intelligence. It seeks quality papers on a wide range of topics in applied intelligence that are employed in developing intelligent systems for solving real life problems in all disciplines. Every year this conference brings together scientists, engineers and practitioners, who work on designing and developing applications that use intelligent techniques or work on intelligent techniques and apply them to application domains. The book is comprised of seventeen chapters providing up-to-date and state-of-the-art research on the applications of applied Intelligence techniques.

Emergence of Cooperative Internet Server Sharing Among Internet Search Agents Caught in the n-Person Prisoner's Dilemma Game

Information on the Internet can be collected by autonomous agents that send out queries to the servers that may have the information sought. From a single agent’s perspective, sending out as many queries as possible maximizes the chances of finding the information sought. However, if every agent does the same, the servers will be overloaded. The first major contribution of this paper is proving mathematically that the agents situated in such environments play the n-Person Prisoner’s Dilemma Game. The second is mathematically deriving the notion of effectiveness of cooperation among the agents in such environments and then presenting the optimal interval for the number of information sites for a given number of information-seeking agents. When the optimal interval is satisfied, cooperation among agents is effective, meaning that resources (e.g., servers) are optimally shared. Experimental results suggest that agents can better share available servers through the kinship-based cooperation without explicitly knowing about the entire environment. This paper also identifies difficulties of promoting cooperation in such environments and presents possible solutions. The long-term goal of this research is to elucidate the understanding of massively distributed multiagent environments such as the Internet and to identify valuable design principles of software agents in similar environments.

RTES demo system2004

The RTES Demo System 2004 is a prototype for reliable, fault-adaptive infrastructure applicable to commodity-based dedicated application computer farms, such as the Level 2/3 trigger for the proposed BTeV high energy physics project. This paper describes the prototype, and its demonstration at the 11th IEEE Real Time and Embedded Technology Applications Symposium, RTAS 2005.

Improved learning in genetic rule-based classifier systems

Many learning algorithms tend to converge into local minima that often represent partial solutions. Schemes are presented that greatly minimize the risk of converging to a partial solution and maximize the rule discovery process for rule-based learning. For the experiments, a generic algorithm rule-based learning system called a classifier system has been used. The new strategies are supported by presenting accelerations and completion of learning in higher order letter image classification problems.<<ETX>>