Mario Koeppen

Computational Intelligence in Wireless Sensor Networks

Wireless sensor networks (WSNs) are networks of distributed autonomous devices that can sense or monitor physical or environmental conditions cooperatively. WSNs face many challenges, mainly caused by communication failures, storage and computational constraints and limited power supply. Paradigms of Computational Intelligence (CI) have been successfully used in recent years to address various challenges such as optimal deployment, data aggregation and fusion, energy aware routing, task scheduling, security, and localization. CI provides adaptive mechanisms that exhibit intelligent behaviour in complex and dynamic environments like WSNs. CI brings about flexibility, autonomous behaviour, and robustness against topology changes, communication failures and scenario changes. However, WSN developers can make use of potential CI algorithms to overcome the challenges in Wireless Sensor Network. The seminar includes some of the WSN challenges and their solutions using CI paradigms.

Special Issue on Computational Intelligence in Computer Vision and Image Processing [Guest Editorial]

The articles in this special issue focus on the use of computer intelligence programming in computer vision and image processing applications.

Partial user-supplied information and user modeling for improving QoS

Abstract The incorporation of user-supplied information has become mandatory for the improvement of QoS in network systems. There is the question about accommodation of new users of a service, given that information about former users of a service is available. In the present work, we followed two approaches to derive information about new users in the network design and control processes, where both are based on prototype generation for the answers of former users to a QoS related questionnaire. In the first approach, attempts were made to map user attributes to prototypes. The second approach used a mapping from partial answers to a prototype. As a result, the first approach appeared to be infeasible, while the second showed good results. In the resulting trade-off between number of prototypes and classification accuracy, it is possible, for example, with 8 prototypes for around 1000 users to predict the answers of new users by using only 30% of the answers of former users, while reducing accuracy by only 13% at the same time.

Towards a Kansei WordNet by Color Design SNS Evaluation

Linguistic word nets usually focus on the semantic content and content relations of nouns while not taking into account to what degree those nouns can also reflect visual impressions in a user specific manner. Recently Color Design SNS like COLOURlovers or Adobe Color CC have gained popularity and the huge corpus of available data allows for gaining new insights into the way user associate color designs with impression words. We present the results of an experimental study for finding Kansei aspects of impression words and related user models by physiological closeness of color designs. The closeness relation between color designs is based on intensity impression matching, where a suitable way of reflecting intensity impression of contrasting colors was selected based on subjective impressions evaluation experiments. The first results of this novel combination of physiology and subjective impression can give raise to further investigations into such a direction.

User Modeling for Improving QoS Using Partial User-Supplied Information

For the improvement of QoS, incorporation of usersupplied information in the network design and control process has become mandatory. A problem arises for the handling of new users, when information about existing users is already available. In the work presented in this paper, we were following two approaches to derive information about new users, both based on prototype generation for existing user’s answer pattern to a QoS related questionnaire. In the first approach, it was tried to classify user attributes to a fitting prototype. In the second approach, a mapping from partial answers to a prototype was used. As result, the first approach appeared to be infeasible, while the second one gave good results. In the resulting trade-off between the number of clusters and classification accuracy, this way it is possible, for example, with 8 prototypes for around 1000 users to issue a new user answer patterns by using only 30% of the questions, while reducing accuracy by only 13%.

A survey on water supply system’s technology for the development of enhanced measuring strategies empowered by our gestalt image characterisation

In order to observe and gather viewpoints from various technological achievements undertaken in the frameworks of water distribution systems, a state-of-the-art is researched and presented here in form of methodological highlights related to these fields. Furthermore, to be established and guarantee the thorough efficiency of water supply systems, a method relying on pattern recognition approaches like the coocurrence matrix, edge-detection filtering and region growing algorithms is devised by means of which, an optimized model has been achieved at Sistema de Aguas de la Ciudad de Mexicopsilas (SACM) Hydraulic Systemspsila Verification Subdirection to overcome the peak traffic conditions of the rain season being currently monitored via a robust supervisory control and data acquisition (SCADA) system.

Building a Simulation Model for Distributed Human-Based Evolutionary Computation

Evolutionary computation (EC) is called “human-based EC” especially when its all main operators, which are selection, crossover, and mutation, are executed by humans. One type of human-based EC is distributed human-based EC, in which humans independently manage their solution candidates and share them by direct communication between the humans. It is expected that the EC solves problems in human organizations. However, it is not easy to conduct real experiments to investigate the effect of human behaviors on the performance of the EC because such experiments needs many cooperative people. In the paper, we, therefore, first model human behaviors and then build a simulation model including the model. The model of human behaviors focuses on physical movement and free will to decide a time of interactions with others. Furthermore, we attempt to understand the EC though simulations using the built simulation model.

The Price of Unfairness

Here, measures for unfairness in resource allocation problems are studied. The extremity of such measures should match with an intuitive concept of an unfair allocation. To achieve a computational model for unfairness, various fairness models are extended to cover a related meaning of unfairness as well. In addition to the lexicographic maxmax relation, a model based on exponential utilities is introduced. Furthermore, a new mean, the symmetric Lehmer mean is identified as being able to favour allocations restricted to a subset of users only. Case examples of network flow control problems show feasibility of the proposed unfairness measures, as well as specific differences. Especially the symmetric Lehmer mean appears capable to handle unfairness in a much more nuanced way, while lexicographic maxmax appears as the computationally most convenient and also most intuitive measure. In all cases it can be shown that unfair allocations are not always efficient and that there is a price of unfairness as well.

Non-swarm intelligence search algorithm based on the foraging behaviors of fruit flies

In the present study, we examine the foraging behavior of the fruit fly (Drosophila), which unlike social insects, independently and autonomously forages. Moreover, we propose a new population-based, non-swarm intelligence search algorithm that mimics the foraging behaviors of fruit flies. We find that the time that a fruit fly stays in a food source region follows a power-law. A power-law distribution satisfies the scale-free property, so the observed behaviors of fruit flies can be called scale-free behaviors, which are modeled by the proposed algorithm. We examine the effect of scale-free behaviors on search performance by applying the proposed algorithm to 28 test functions with two real parameters. The results reveal that the scale-free property is related to a balance between exploration and exploitation and affects the search performance. Moreover, particular parameter values for fixing the scale-free property yield a better search performance for most of the test functions considered herein. The present paper does not provide a competent search algorithm for solving static optimization problems, but rather introduces a novel temporal property of individual behaviors, i.e., a scale-free property, which is expected to provide a new way to think about search algorithms.

Evolving Fair Linear Regression for the Representation of Human-Drawn Regression Lines

Here we study a generalization of linear regression to the case of maximal elements of a general fairness relation. The regression then is based on balancing the distances to the data points. The studied relations are lexicographic minimum, maxmin fairness, proportional fairness, and majorities, all in a complementary version to represent minimality. A new combination of proportional fairness and majority is introduced as well. Experiments are performed on human subjects solving the visual task to draw a line fitting to given data points, and by use of evolutionary computation (here by Differential Evolution) the weights of a fair linear regression are adjusted to the human-provided results. The fact that this gives a more precise approximation than (weighted) linear regression hints on the inclusion of the balance among the distances to the given data points in the human decision making process.