for Journals by Title or ISSN
for Articles by Keywords

Publisher: Springer-Verlag   (Total: 2187 journals)

 A  B  C  D  E  F  G  H  I  J  K  L  M  N  O  P  Q  R  S  T  U  V  W  X  Y  Z  

  First | 3 4 5 6 7 8 9 10 | Last

Eating and Weight Disorders - Studies on Anorexia, Bulimia and Obesity     Hybrid Journal   (4 followers)
EcoHealth     Hybrid Journal   (1 follower)
Ecological Research     Hybrid Journal   (7 followers)
Economic Botany     Hybrid Journal   (8 followers)
Economic Bulletin     Hybrid Journal   (3 followers)
Economic Change and Restructuring     Hybrid Journal   (1 follower)
Economic Theory     Hybrid Journal   (5 followers)
Economic Theory Bulletin     Hybrid Journal  
Economics of Governance     Hybrid Journal   (2 followers)
Ecosystems     Hybrid Journal   (18 followers)
Ecotoxicology     Hybrid Journal   (10 followers)
Education and Information Technologies     Hybrid Journal   (98 followers)
Educational Assessment, Evaluation and Accountability     Hybrid Journal   (10 followers)
Educational Psychology Review     Hybrid Journal   (13 followers)
Educational Research for Policy and Practice     Hybrid Journal   (5 followers)
Educational Studies in Mathematics     Hybrid Journal   (4 followers)
Educational Technology Research and Development     Partially Free   (86 followers)
Electrical Engineering     Hybrid Journal   (8 followers)
Electrocatalysis     Hybrid Journal  
Electronic Commerce Research     Hybrid Journal   (3 followers)
Electronic Markets     Hybrid Journal   (5 followers)
Electronic Materials Letters     Hybrid Journal   (2 followers)
Elemente der Mathematik     Hybrid Journal  
Emergency Radiology     Hybrid Journal   (4 followers)
Empirica     Hybrid Journal   (3 followers)
Empirical Economics     Hybrid Journal   (7 followers)
Empirical Software Engineering     Hybrid Journal   (4 followers)
Employee Responsibilities and Rights Journal     Hybrid Journal   (2 followers)
Endocrine     Hybrid Journal   (4 followers)
Endocrine Pathology     Hybrid Journal   (1 follower)
Energy Efficiency     Hybrid Journal   (9 followers)
Energy Systems     Hybrid Journal   (7 followers)
Engineering With Computers     Hybrid Journal   (5 followers)
Entomological Review     Hybrid Journal   (3 followers)
Environment Systems & Decisions     Hybrid Journal   (2 followers)
Environment, Development and Sustainability     Hybrid Journal   (22 followers)
Environmental and Ecological Statistics     Hybrid Journal   (5 followers)
Environmental and Resource Economics     Hybrid Journal   (16 followers)
Environmental Biology of Fishes     Hybrid Journal   (3 followers)
Environmental Chemistry Letters     Hybrid Journal   (2 followers)
Environmental Earth Sciences     Hybrid Journal   (10 followers)
Environmental Economics and Policy Studies     Hybrid Journal   (5 followers)
Environmental Evidence     Open Access  
Environmental Fluid Mechanics     Hybrid Journal   (2 followers)
Environmental Geochemistry and Health     Hybrid Journal   (2 followers)
Environmental Geology     Hybrid Journal   (10 followers)
Environmental Health and Preventive Medicine     Hybrid Journal   (2 followers)
Environmental Management     Hybrid Journal   (27 followers)
Environmental Modeling & Assessment     Hybrid Journal   (10 followers)
Environmental Monitoring and Assessment     Hybrid Journal   (9 followers)
Environmental Science and Pollution Research     Hybrid Journal   (11 followers)
Epidemiologic Perspectives & Innovations     Open Access   (1 follower)
Epileptic Disorders     Hybrid Journal   (1 follower)
EPJ A - Hadrons and Nuclei     Hybrid Journal   (1 follower)
EPJ B - Condensed Matter and Complex Systems     Hybrid Journal   (3 followers)
EPJ direct     Hybrid Journal  
EPJ E - Soft Matter and Biological Physics     Hybrid Journal   (1 follower)
ERA-Forum     Hybrid Journal   (1 follower)
Erkenntnis     Hybrid Journal   (11 followers)
Erwerbs-Obstbau     Hybrid Journal  
Esophagus     Hybrid Journal  
Estuaries and Coasts     Hybrid Journal   (1 follower)
Ethical Theory and Moral Practice     Hybrid Journal   (7 followers)
Ethics and Information Technology     Hybrid Journal   (95 followers)
Ethik in der Medizin     Hybrid Journal  
Euphytica     Hybrid Journal   (7 followers)
Eurasian Soil Science     Hybrid Journal   (2 followers)
EURO Journal of Transportation and Logistics     Hybrid Journal   (4 followers)
EURO Journal on Computational Optimization     Hybrid Journal  
Europaisches Journal fur Minderheitenfragen     Hybrid Journal  
European Actuarial Journal     Hybrid Journal   (2 followers)
European Archives of Oto-Rhino-Laryngology     Hybrid Journal   (3 followers)
European Archives of Paediatric Dentistry     Hybrid Journal   (1 follower)
European Archives of Psychiatry and Clinical Neuroscience     Hybrid Journal   (2 followers)
European Biophysics Journal     Hybrid Journal   (4 followers)
European Child & Adolescent Psychiatry     Hybrid Journal   (4 followers)
European Clinics in Obstetrics and Gynaecology     Hybrid Journal   (3 followers)
European Food Research and Technology     Hybrid Journal   (8 followers)
European Journal for Education Law and Policy     Hybrid Journal   (5 followers)
European Journal for Philosophy of Science     Partially Free   (4 followers)
European Journal of Ageing     Hybrid Journal   (7 followers)
European Journal of Applied Physiology     Hybrid Journal   (5 followers)
European Journal of Clinical Microbiology & Infectious Diseases     Hybrid Journal   (7 followers)
European Journal of Clinical Pharmacology     Hybrid Journal   (9 followers)
European Journal of Drug Metabolism and Pharmacokinetics     Hybrid Journal   (6 followers)
European Journal of Epidemiology     Hybrid Journal   (16 followers)
European Journal of Forest Research     Hybrid Journal   (3 followers)
European Journal of Health Economics     Hybrid Journal   (10 followers)
European Journal of Law and Economics     Hybrid Journal   (108 followers)
European Journal of Nuclear Medicine and Molecular Imaging     Hybrid Journal   (5 followers)
European Journal of Nutrition     Hybrid Journal   (16 followers)
European Journal of Orthopaedic Surgery & Traumatology     Hybrid Journal   (4 followers)
European Journal of Pediatrics     Hybrid Journal   (7 followers)
European Journal of Plant Pathology     Hybrid Journal   (2 followers)
European Journal of Plastic Surgery     Hybrid Journal   (2 followers)
European Journal of Population/Revue europĂ©enne de DĂ©mographie     Hybrid Journal   (3 followers)
European Journal of Psychology of Education     Hybrid Journal   (6 followers)
European Journal of Trauma and Emergency Surgery     Hybrid Journal   (8 followers)
European Journal of Wildlife Research     Hybrid Journal   (5 followers)
European Journal of Wood and Wood Products     Hybrid Journal   (3 followers)

  First | 3 4 5 6 7 8 9 10 | Last

Evolutionary Intelligence    Follow    
  Hybrid Journal Hybrid journal (It can contain Open Access articles)
     ISSN (Print) 1864-5917 - ISSN (Online) 1864-5909
     Published by Springer-Verlag Homepage  [2187 journals]   [SJR: 0.42]   [H-I: 9]
  • Introduction to the special issue on evolutionary intelligence in games
    • PubDate: 2014-01-24
  • Creating autonomous agents for playing Super Mario Bros game by means of
           evolutionary finite state machines
    • Abstract: Abstract This paper shows the design and improvement of an autonomous agent based in using evolutionary methods to improve behavioural models (finite state machines), which are part of the individuals to evolve. This leads to the obtention of a so-called bot that follows the Gameplay track rules of the international Mario AI Championship and is able to autonomously complete different scenarios on a simulator of Super Mario Bros. game. Mono- and multi-seed approaches (evaluation in one play or multiple plays respectively) have been analysed, in order to compare respectively the performance of an approach focused in solving a specific scenario, and another more general, devoted to obtain an agent which can play successfully in different scenarios. The analysis considers the machine resources consumption, which turns in a bottleneck in some experiments. However, the methods yield agents which can finish several stages of different difficulty levels, and playing much better than an expert human player, since they can deal with very difficult situations (several enemies surrounding Mario, for instance) in real time. According to the results and considering the competition’s restrictions (time limitations) and objectives (complete scenarios up to difficulty level 3), these agents have enough performance to participate in this competition track.
      PubDate: 2014-01-24
  • On evolutionary subspace clustering with symbiosis
    • Abstract: Abstract Subspace clustering identifies the attribute support for each cluster as well as identifying the location and number of clusters. In the most general case, attributes associated with each cluster could be unique. A multi-objective evolutionary method is proposed to identify the unique attribute support of each cluster while detecting its data instances. The proposed algorithm, symbiotic evolutionary subspace clustering (S-ESC) borrows from ‘symbiosis’ in the sense that each clustering solution is defined in terms of a host (single member of the host population) and a number of coevolved cluster centroids (or symbionts in an independent symbiont population). Symbionts define clusters and therefore attribute subspaces, whereas hosts define sets of clusters to constitute a non-degenerate solution. The symbiotic representation of S-ESC is the key to making it scalable to high-dimensional datasets, while an integrated subsampling process makes it scalable to tasks with a large number of data items. Benchmarking is performed against a test suite of 59 subspace clustering tasks with four well known comparator algorithms from both the full-dimensional and subspace clustering literature: EM, MINECLUS, PROCLUS, STATPC. Performance of the S-ESC algorithm was found to be robust across a wide cross-section of properties with a common parameterization utilized throughout. This was not the case for the comparator algorithms. Specifically, performance could be sensitive to the particular data distribution or parameter sweeps might be necessary to provide comparable performance. An additional evaluation is performed against a non-symbiotic GA, with S-ESC still returning superior clustering solutions.
      PubDate: 2014-01-12
  • A fast anomaly detection system using probabilistic artificial immune
           algorithm capable of learning new attacks
    • Abstract: Abstract In this paper, we propose anomaly based intrusion detection algorithms in computer networks using artificial immune systems, capable of learning new attacks. Unique characteristics and observations specific to computer networks are considered in developing faster algorithms while achieving high performance. Although these characteristics play a key role in the proposed algorithms, we believe they have been neglected in the previous related works. We evaluate the proposed algorithms on a number of well-known intrusion detection datasets, as well as two new real datasets extracted from the data networks for intrusion detection. We analyze the detection performance and learning capabilities of the proposed algorithms, in addition to performance criteria such as false alarm rate, detection rate, and response time. The experimental results demonstrate that the proposed algorithms exhibit fast response time, low false alarm rate, and high detection rate. They can also learn new attack patterns, and identify them the next time they are introduced to the network.
      PubDate: 2013-12-25
  • Swarm intelligence based algorithms: a critical analysis
    • Abstract: Abstract Many optimization algorithms have been developed by drawing inspiration from swarm intelligence (SI). These SI-based algorithms can have some advantages over traditional algorithms. In this paper, we carry out a critical analysis of these SI-based algorithms by analyzing their ways to mimic evolutionary operators. We also analyze the ways of achieving exploration and exploitation in algorithms by using mutation, crossover and selection. In addition, we also look at algorithms using dynamic systems, self-organization and Markov chain framework. Finally, we provide some discussions and topics for further research.
      PubDate: 2013-12-17
  • Overview of Harmony Search algorithm and its applications in Civil
    • Abstract: Abstract Harmony Search (HS), a meta-heuristic algorithm, conceptualizes a musical process of searching for a perfect state of harmony (optimal solution). It allows a random search without initial values and removes the necessity for information of derivatives. Since the HS algorithm was first developed and published in 2001, it has been applied to various research areas and the world wide attention on it has rapidly increased. In this paper, applications of HS algorithm in Civil Engineering (CE) are to be overviewed. Articles in CE areas including water resources, structural, geotechnical, environmental, and traffic engineering are to be reviewed thoroughly. As a results, variety of application results show that HS can be effectively used as a tool for optimization problems in CE.
      PubDate: 2013-12-12
  • New solver and optimal anticipation strategies design based on
           evolutionary computation for the game of MasterMind
    • Abstract: Abstract This paper presents and compares several evolutionary solutions for the well-known MasterMind game, a classic board game invented in the 1970s. First, we propose a novel evolutionary approach (which we call nested hierarchical evolutionary search) to solve the MasterMind game, comparing the obtained results with that of existing algorithms. Second, we show how to design novel game anticipation strategies for the MasterMind game, by applying genetic programming. In this case we compare the performance of the new obtained strategies with that of the classical ones, obtaining advantages in all the cases tested.
      PubDate: 2013-12-11
  • Reconstructing biological gene regulatory networks: where optimization
           meets big data
    • Abstract: Abstract The importance of ‘big data’ in biology is increasing as vast quantities of data are being produced from high-throughput experiments. Techniques such as DNA microarrays are providing a genome-wide picture of gene expression levels, allowing us to investigate the structure and interactions of gene networks in biological systems. Inference of gene regulatory network (GRN) is an underdetermined problem suited to Metaheuristic algorithms which can operate on limited information. Thus GRN inference offers a platform for investigations into data intensive sciences and large scale optimization problems. Here we examine the link between data intensive research and optimization problems for the reverse engineering of GRNs. Briefly, we detail the benefit of the data deluge and the study of ALife for modelling GRNs as well as their reconstruction. We discuss how metaheuristics can solve big data problems and the inference of GRNs offer real world problems for both areas of research. We overview some current reconstruction algorithms and investigate some modelling and computational limits of the inference processes and suggest some areas for development. Furthermore we identify links and synergies between optimization and big data, e.g., dynamic, uncertain and large scale optimization problems, and discuss the potential benefit of multi- and many-objective optimization. We stress the importance of data integration techniques in order to maximize the data available, particularly for the case of inferring GRNs from microarray data. Such multi-disciplinary research is vital as biology is rapidly becoming a quantitative, data intensive science.
      PubDate: 2013-11-22
  • Special issue on advances in Learning Classifier Systems
    • PubDate: 2013-11-08
  • Self organizing classifiers: first steps in structured evolutionary
           machine learning
    • Abstract: Abstract Learning classifier systems (LCSs) are evolutionary machine learning algorithms, flexible enough to be applied to reinforcement, supervised and unsupervised learning problems with good performance. Recently, self organizing classifiers were proposed which are similar to LCSs but have the advantage that in its structured population no balance between niching and fitness pressure is necessary. However, more tests and analysis are required to verify its benefits. Here, a variation of the first algorithm is proposed which uses a parameterless self organizing map (SOM). This algorithm is applied in challenging problems such as big, noisy as well as dynamically changing continuous input-action mazes (growing and compressing mazes are included) with good performance. Moreover, a genetic operator is proposed which utilizes the topological information of the SOM’s population structure, improving the results. Thus, the first steps in structured evolutionary machine learning are shown, nonetheless, the problems faced are more difficult than the state-of-art continuous input-action multi-step ones.
      PubDate: 2013-10-26
  • Methods for approximating value functions for the Dominion card game
    • Abstract: Abstract Artificial neural networks have been successfully used to approximate value functions for tasks involving decision making. In domains where decisions require a shift in judgment as the overall state changes, it is hypothesized here that methods utilizing multiple artificial neural networks are likely to provide a benefit as an approximation of a value function over those that employ a single network. The card game Dominion was chosen as the domain to examine this. This paper compares artificial neural networks generated by multiple machine learning methods successfully applied to other games (such as in TD-Gammon) to a genetic algorithm method for generating two neural networks for different phases of the game along with evolving the transition point. The results demonstrate a greater success ratio with the genetic algorithm applied to two neural networks. This suggests that future work examining more complex neural network configurations and richer evolutionary exploration could apply to Dominion as well as other domains necessitating shifts in strategy.
      PubDate: 2013-10-24
  • Adaptive artificial datasets through learning classifier systems for
           classification tasks
    • Abstract: Abstract In producing an artificial dataset, humans usually play a major role in creating and controlling the problem domain. In particular, humans set up and tune the problem’s difficulty. If humans can set up the difficulty levels appropriately, then learning systems can solve classification tasks successfully. This paper introduces an autonomous classification problem generation approach. The problem’s difficulty is adapted based on the classification agent’s performance within the defined attributes. An automated problem generator has been created to evolve simulated datasets whilst the classification agent, in this case a learning classifier system (LCS), attempts to learn the evolving datasets. The idea here is to tune the problem’s difficulty autonomously such that the problem’s characteristics may be determined effectively. Furthermore, this framework can empirically test the learning bounds of the classification agent whilst lowering human involvement. Initially, tabu search was integrated in the problem generator to discover the best combination of domain features in order to adjust the problem’s difficulty. In order to overcome stagnation in local optimum, a Pittsburgh-style LCSs, A-PLUS, was adapted for the first time to the problem generator. In this way, the effect of the problem’s characteristics, e.g. noise, which alter the classification agent’s performance, becomes human readable. Experiments confirm that the problem generator was able to tune the problem’s difficulty either to make the problem ‘harder’ or ‘easier’ so that it can either ‘increase’ or ‘decrease’ the classification agent’s performance.
      PubDate: 2013-10-22
  • Performance analysis of rough set ensemble of learning classifier systems
           with differential evolution based rule discovery
    • Abstract: Abstract Data mining, and specifically supervised data classification, is a key application area for Learning Classifier Systems (LCS). Scaling to larger classification problems, especially to higher dimensional problems, is a key challenge. Ensemble based approaches can be applied to LCS to address scalability issues. To this end a rough set based ensemble of LCS is proposed, which relies on a pre-processed feature partitioning step to train multiple LCS on feature subspaces. Each base classifier in the ensemble is a Michigan style supervised LCS. The traditional genetic algorithm based rule evolution is replaced by a differential evolution based rule discovery, to improve generalisation capabilities of LCS. A voting mechanism is then used to generate output for test instances. This paper describes the proposed ensemble algorithm in detail, and compares its performance with different versions of base LCS on a number of benchmark classification tasks. Analysis of computational time and model accuracy show the relative merits of the ensemble algorithm and base classifiers on the tested data sets. The rough set based ensemble learning approach and differential evolution based rule searching out-perform the base LCS on classification accuracy over the data sets considered. Results also show that small ensemble size is sufficient to obtain good performance.
      PubDate: 2013-10-09
  • Learning complex, overlapping and niche imbalance Boolean problems using
           XCS-based classifier systems
    • Abstract: Abstract XCS is an accuracy-based learning classifier system, which has been successfully applied to learn various classification and function approximation problems. Recently, it has been reported that XCS cannot learn overlapping and niche imbalance problems using the typical experimental setup. This paper describes two approaches to learn these complex problems: firstly, tune the parameters and adjust the methods of standard XCS specifically for such problems. Secondly, apply an advanced variation of XCS. Specifically, we developed previously an XCS with code-fragment actions, named XCSCFA, which has a more flexible genetic programming like encoding and explicit state-action mapping through computed actions. This approach is examined and compared with standard XCS on six complex Boolean datasets, which include overlapping and niche imbalance problems. The results indicate that to learn overlapping and niche imbalance problems using XCS, it is beneficial to either deactivate action set subsumption or use a relatively high subsumption threshold and a small error threshold. The XCSCFA approach successfully solved the tested complex, overlapping and niche imbalance problems without parameter tuning, because of the rich alphabet, inconsistent actions and especially the redundancy provided by the code-fragment actions. The major contribution of the work presented here is overcoming the identified problem in the wide-spread XCS technique.
      PubDate: 2013-10-08
  • A multi-core parallelization strategy for statistical significance testing
           in learning classifier systems
    • Abstract: Abstract Permutation-based statistics for evaluating the significance of class prediction, predictive attributes, and patterns of association have only appeared within the learning classifier system (LCS) literature since 2012. While still not widely utilized by the LCS research community, formal evaluations of statistical confidence are imperative to large and complex real world applications such as genetic epidemiology where it is standard practice to quantify the likelihood that a seemingly meaningful statistic could have been obtained purely by chance. Learning classifier system algorithms are relatively computationally expensive on their own. The compounding requirements for generating permutation-based statistics may be a limiting factor for some researchers interested in applying LCS algorithms to real world problems. Technology has made LCS parallelization strategies more accessible and thus more popular in recent years. In the present study we examine the benefits of externally parallelizing a series of independent LCS runs such that permutation testing with cross validation becomes more feasible to complete on a single multi-core workstation. We test our python implementation of this strategy in the context of a simulated complex genetic epidemiological data mining problem. Our evaluations indicate that as long as the number of concurrent processes does not exceed the number of CPU cores, the speedup achieved is approximately linear.
      PubDate: 2013-10-08
  • A comparative study: the effect of the perturbation vector type in the
           differential evolution algorithm on the accuracy of robot pose and heading
    • Abstract: Abstract Evolutionary algorithms (EAs) belong to a group of classic optimizers these days, and can be used in many application areas. Autonomous mobile robotics is not an exception. EAs are utilized profusely for the purposes of localization and map building of unknown environment—SLAM. This paper concentrates on one particular class of EA, the so called differential evolution (DE). It addresses the problem of selecting a suitable set of parameter values for the DE algorithm applied to the task of continuous robot localization in a known environment under the presence of additive noise in sensorial data. The primary goal of this study is to find at least one type of perturbation vector from a set of known perturbation vector types, suitable to navigate a robot using 2D laser scanner (2DLS) sensorial system. The basic navigational algorithm used in this study uses a vector representation for both the data and the environment map, which is used as a reference data source for the navigation. Since the algorithm does not use a probability occupancy grid, the precision of the results is not limited by the grid resolution. The comparative study presented in this paper includes a relatively large amount of experiments in various types of environments. The results of the study suggest that the DE algorithm is a suitable tool for continuous robot localization task in an indoor environment, with or without moving objects, and under the presence of various levels of additive noise in sensorial data. Two perturbation vector types were found as the most suitable for this task on average, namely rand/1/exp and randtobest/1/bin.
      PubDate: 2013-09-25
  • Improved thresholding based on negative selection algorithm (NSA)
    • Abstract: Abstract Thresholding is a tool of image segmentation which groups the pixels in a logical way. In this paper, a novel algorithm based on negative selection algorithm a model of artificial immune system is proposed for image thresholding. The proposed algorithm is applied on the thresholded images of lathe tool produced using maximum information entropy (MIE) and global thresholding based technique resulting in an improved image. To verify the algorithm and results, it has also been applied on some of the inbuilt MATLAB (MATrix LABoratory) images. Histogram is employed to analyze the results. Further, the results of improved algorithm are compared with the results of MIE and the global thresholding methods to check the effectiveness of the proposed method. The experimental results confirm the potential of the developed algorithm.
      PubDate: 2013-09-01
  • Accelerating IEC and EC searches with elite obtained by dimensionality
           reduction in regression spaces
    • Abstract: Abstract We propose a method for accelerating interactive evolutionary computation (IEC) and evolutionary computation (EC) searches using elite obtained in one-dimensional spaces and use benchmark functions to evaluate the proposed method. The method projects individuals onto n one-dimensional spaces corresponding to each of the n searching parameter axes, approximates each landscape using interpolation or an approximation method, finds the best coordinate from the approximated shape, obtains the elite by combining the best n found coordinates, and uses the elite for the next generation of the IEC or EC. The advantage of this method is that the elite may be easily obtained thanks to their projection onto each one-dimensional space and there is a higher possibility that the elite individual locates near the global optimum. We compare the proposal with methods for obtaining the landscape in the original search space, and show that our proposed method can significantly save computational time. Experimental evaluations of the technique with differential evolution using a simulated IEC user (Gaussian mixture model with different dimensions) and 34 benchmark functions show that the proposed method substantially accelerates IEC and EC searches.
      PubDate: 2013-05-16
  • Knowledge extraction and rule set compaction in XCS for non-Markov
           multi-step problems
    • Abstract: Abstract The application of the XCS family of classifier systems within multi-step problems has shown promise, but XCS classifier systems usually solve those problems with a large number of classifiers. If they can generate compact and readily interpretable solutions when used in multi-step problems, it will enhance their applicability and usefulness in a wide range of robot control and knowledge discovery tasks. In this paper, we describe some methods developed for XCS to compact the rule set and acquire explicit policy knowledge for multi-step problems, especially non-Markov problems. Experimental results show the ability to obtain a very compact rule set out of the final population of classifiers with little or no degradation of overall performance.
      PubDate: 2013-04-30
  • Scalable multiagent learning through indirect encoding of policy geometry
    • Abstract: Abstract Multiagent systems present many challenging, real-world problems to artificial intelligence. Because it is difficult to engineer the behaviors of multiple cooperating agents by hand, multiagent learning has become a popular approach to their design. While there are a variety of traditional approaches to multiagent learning, many suffer from increased computational costs for large teams and the problem of reinvention (that is, the inability to recognize that certain skills are shared by some or all team member). This paper presents an alternative approach to multiagent learning called multiagent HyperNEAT that represents the team as a pattern of policies rather than as a set of individual agents. The main idea is that an agent’s location within a canonical team layout (which can be physical, such as positions on a sports team, or conceptual, such as an agent’s relative speed) tends to dictate its role within that team. This paper introduces the term policy geometry to describe this relationship between role and position on the team. Interestingly, such patterns effectively represent up to an infinite number of multiagent policies that can be sampled from the policy geometry as needed to allow training very large teams or, in some cases, scaling up the size of a team without additional learning. In this paper, multiagent HyperNEAT is compared to a traditional learning method, multiagent Sarsa(λ), in a predator–prey domain, where it demonstrates its ability to train large teams.
      PubDate: 2013-01-01
School of Mathematical and Computer Sciences
Heriot-Watt University
Edinburgh, EH14 4AS, UK
Tel: +00 44 (0)131 4513762
Fax: +00 44 (0)131 4513327
About JournalTOCs
News (blog, publications)
JournalTOCs on Twitter   JournalTOCs on Facebook

JournalTOCs © 2009-2014