
Helmut A. MayerUniversity of Salzburg · Department of Artificial Intelligence and Human Interfaces
Helmut A. Mayer
Dipl.-Ing. Dr.
About
52
Publications
7,928
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
311
Citations
Introduction
Having an engineering background I am very keen on basic research without losing sight of potential applications. Naturally, the field of Computational Intelligence offers vast opportunities to utilize the results of basic research for various applications. The areas I am currently working in are: Evolutionary Computation, Artificial Neural Networks, Evolutionary Robotics, Computational Neuroethology, ANN Game Playing, Industrial Applications of Computational Intelligence.
Additional affiliations
May 1988 - May 1993
Education
October 1980 - July 1987
Publications
Publications (52)
We report on experiments with robotic neurocontrollers with intrinsic noise evolved for a peg pushing task. The specific controller of the simulated robot is a feed-forward network with noisy weights, i.e, the weight values are perturbed by additive, normal noise. The neurocontrollers are evolved in a noise-free environment, and the best-performing...
One of the most prominent research goals in the field of mobile autonomous robots is to create robots that are able to adapt to new environments, i.e., the robots should be able to learn during their "lifetime" possibly without (or a minimum) of human intervention. When employing artificial neural networks (ANNs) to control the robot, reinforcement...
Digital organisms (DOs) model the basic structure and development of natural organisms to create robust, scalable, and adaptive solutions to problems from different fields. The applicability of DOs has been investigated mainly on a few synthetic problems like pattern creation, but on a very limited number of real world problems, e.g., the creation...
In this tutorial, we present a compact and holistic discussion of Deep Learning with a focus on Convolutional Neural Networks (CNNs) and supervised regression. While there are numerous books and articles on the individual topics we cover, comprehensive and detailed tutorials that address deep learning from a foundational yet rigorous and accessible...
In this tutorial, we present a compact and holistic discussion of Deep Learning with a focus on Convolutional Neural Networks (CNNs) and supervised regression. While there are numerous books and articles on the individual topics we cover, comprehensive and detailed tutorials that address Deep Learning from a foundational yet rigorous and accessible...
Human Body Dimensions Estimation (HBDE) is a task that an intelligent agent can perform to attempt to determine human body information from images (2D) or point clouds or meshes (3D). More specifically, if we define the HBDE problem as inferring human body measurements from images, then HBDE is a difficult, inverse, multi-task regression problem th...
Human Body Dimensions Estimation (HBDE) is a task that an intelligent agent can perform to attempt to determine human body information from images (2D) or point clouds or meshes (3D). More specifically, if we define the HBDE problem as inferring human body measurements from images, then HBDE is a difficult, inverse, multi-task regression problem th...
Human shape estimation has become increasingly important both theoretically and practically, for instance, in 3D mesh estimation, distance garment production and computational forensics, to mention just a few examples. As a further specialization, \emph{Human Body Dimensions Estimation} (HBDE) focuses on estimating human body measurements like shou...
In this paper we present CALVIS, a method to calculate $\textbf{C}$hest, w$\textbf{A}$ist and pe$\textbf{LVIS}$ circumference from 3D human body meshes. Our motivation is to use this data as ground truth for training convolutional neural networks (CNN). Previous work had used the large scale CAESAR dataset or determined these anthropometrical measu...
In this paper we present CALVIS, a method to calculate Chest, wAist and peLVIS circumference from 3D human body meshes. Our motivation is to use this data as ground truth for training convolutional neural networks (CNN). Previous work had used the large scale CAESAR dataset or determined these anthropometrical measurements manually from a person or...
Sleep has been proposed to indicate preserved residual brain functioning in patients suffering from disorders of consciousness (DOC) after awakening from coma. However, a reliable characterization of sleep patterns in this clinical population continues to be challenging given severely altered brain oscillations, frequent and extended artifacts in c...
Demographic data for patients.
The analyzed patient sample 12 UWS and 11 MCS subjects. Abbreviations: M = male, F = female, TBI = Traumatic Brain Injury, CVA-Cerebrovascular Accident, SSPE = Subacute Sclerosing Panencephalitis, SD- = lower severe disability (3 points on Extended Glasgow Outcome Scale), eMCS = emergence from MCS; CRC-R = Coma Recove...
We use neuroevolution to construct nonlinear transformation functions for feature construction that map points in the original feature space to augmented pattern vectors and improve the performance of generic classifiers. Our research demonstrates that we can apply evolutionary algorithms to both adapt the weights of a fully connected standard mult...
Predicting the class membership of a set of patterns represented by points in a multi-dimensional space critically depends on their specific distribution. To improve the classification performance, pattern vectors may be transformed. There is a range of linear methods for feature construction, but these are often limited in their performance. Nonli...
This paper presents a generic machine learning based approach to devise performance assessment functions for any kind of optimization problem. The need of a performance assessment process taking into account robustness of the solutions is stressed and a general methodology for devising a function to estimate such a performance on any given engineer...
The majority of work on artificial neural networks (ANNs) playing the game of Go focus on network architectures and training regimes to improve the quality of the neural player. A less investigated problem is the board representation conveying the information on the current state of the game to the network. Common approaches suggest a straight-forw...
We investigate evolutionary approaches to generate well- performing strategies for the iterated prisoner's dilemma (IPD) with different history lengths in static and cultural environments. The length of the history determines the number of the most recent moves of both players taken into account for the current move decision. The static environment...
Contemplating the development of the field of evolutionary computation (EC), where most basic concepts are borrowed from nature, it is remarkable that multi-chromosomal representations present in all complex organisms have rarely been studied in the artificial domain. Evidently, the addition of such an additional layer of genetic code must prove to...
The EMMA Modular Control Concept (EMCC) is a Java framework for controlling EMMA2, an autonomous,
mobile, learning and vision enabled robot, designed by the RoboLab. With the help of a highly
modular framework, the implementation of robot controllers serving a wide range of purposes becomes a
simple and user-friendly task that can be can be accompl...
A problem in computer numeric control (CNC) manufacturing is the minimization of the distance a drilling tool has to move in order to drill holes at given locations. After introducing the problem being an instance of the traveling salesman problem (TSP) we describe the Route Optimizer RO3 based on an evolutionary algorithm (EA). In experiments with...
We present experiments (co)evolving Go players based on artificial neural networks (ANNs) for a 5×5 board. ANN structure and weights are encoded in multi-chromosomal genotypes. In evolutionary scenarios, a population of generalized multi-layer perceptrons (GMLPs) has to compete with a single Go program from a set of three players of different quali...
We present experiments evolving computer Go players based on artificial neural networks (ANNs) for a 5×5 board, where ANN structure and weights are encoded in multi-chromosomal genotypes. In evolutionary scenarios a population of generalized multi-layer perceptrons (GMLPs) has to compete with a single computer Go program from a set of three artific...
Biological neural networks having been shaped in billions of years of evolution are the source of numerous, complex capabilities of living organisms. Especially, the form of intelligence attributed to humans has inspired computer scientists to model the evolution of neural networks with the ultimate goal to create computer systems with cognitive ab...
We present the application of the generic framework evAlloc for the solution of allocation and scheduling problems (ASPs) to a real-world problem. The solution engine integrated in the framework is based on an evolutionary algorithm (EA). The general design of the Java framework allows for application to all ASPs, whose problem data description can...
Poster presented at the conference
We present experiments investigating the use of multichromosomal representations in evolutionary algorithms. Specifically, the conventional representation of parameters on a single chromosome is compared to a genotype encoding with multiple chromosomes on a set of test functions. In this context we present chromosome shuffling, a genetic operator r...
After a brief survey of work dealing with dynamic neurocontrollers changing their internal structure during the "lifetime" of a mobile autonomous robot, we present experiments employing a standard sensor-motor neurocontroller with self-adapting weights. The change of behavior of the robot is linked to inputs from the environment that cause the emis...
We investigate key components of a dynamic neurocontroller changing its internal structure enabling "lifetime" learning of a mobile autonomous robot. The behavioral change of the robot is linked to inputs from the environment that cause the emission of artificial neuromodulators (ANMs) in the robot's neurocontroller. In its simplest form an outside...
In this work we present experiments evolving the membership functions of a fuzzy controller for the inverted pendulum problem. Specifically, the conventional representation of membership functions on a single chromosome is compared to a genotype encoding with multiple chromosomes in an evolutionary algorithm. In this context we present chromosome s...
In this paper we investigate the use of neuron-specific activation
functions (AFs) within generalized multilayer perceptrons (GMLP). We
utilize the netGEN system not only to evolve the structure of an
artificial neural network (ANN), but also to search for a set of AF
templates which are assigned to specific neurons by evolution. This may
be seen a...
The most common (or even only) choice of activation functions (AFs) for multi{layer perceptrons (MLPs) widely used in research, engineering and business is the logistic function. Among the reasons for this popularity are its boundedness in the unit interval, the function's and its derivative's fast computability, and a number of amenable mathematic...
In this paper we present the current stage of development of EMMA, a small autonomous robot inspired by the Khepera robot widely distributed in the scientific community. EMMA features a modern microcontroller as the central processing unit (CPU) being very efficient in processing sensor and motor signals and a black--white camera as the main sensor...
In this work we present first experiments with a lossy data compression method which is based on the genetic mechanism of frame shifting occuring during the translation of genetic code into proteins. The basic principle of the presented method is the differential usage of common genetic code which is induced by promoters and terminators marking the...
The most common (or even only) choice of activation functions for multi-layer perceptrons (MLPs) widely used in research, engineering and business is the logistic function. Among the reasons for this popularity are its boundedness in the unit interval, the function's and its derivative's fast computability, and a number of amenable mathematical pro...
The steady increase of the amount of information gathered by remote sensing systems makes the development of highly efficient algorithms to reduce the data volume even more important. Ideally, however, such methods should achieve this goal without or only slight loss of information contained in the data. Generally, feature selection algorithms offe...
In this paper we compare recently developed and highly effective sequential feature selection algorithms with approaches based on evolutionary algorithms enabling parallel feature subset selection. We introduce the oscillating search method, employ permutation encoding offering some advantages over the more traditional bitmap encoding for the evolu...
The prediction of future values of a time series generated by a chaotic dynamical system is an extremely challenging task. Amongst several non-linear models employed for the prediction of chaotic time series artificial neural networks (ANNs) have gained major attention in the past decade. One widely recognized aspect of ANN design in order to achie...
While artificial neural networks (ANNs) have proven their almost
universal applicability in a broad variety of problem domains, many
scientists (theorists and practitioners as well) are still worried by
the opacity of neural problem solvers. As well a network may perform, it
is often desirable, if not necessary, to know at least some general
concep...
Among the most important design issues to be addressed to optimize the generalization abilities of trained artificial neural networks (ANNs) are the specific architecture and the composition of the training data set (TDS). Recent work has focused on investigating each of these prerequisites separately. However, some researchers have pointed out the...
We introduce evolutionary resampling and combine (erc) - a genetic algorithm based selection scheme for training examples for a multilayer perceptron (MLP) classifier. The erc method is compared to various adaptive resample and combine techniques arc-fs, arc-lh and arc-x4. To diminish the dependencies on the size of the training data set (TDS) and...
Supervised training of a neural classifier and its performance not only relies on the artificial neural network (ANN) type, architecture and the training method, but also on the size and composition of the training data set (TDS). For the parallel generation of TDSs for a multi-layer perceptron (MLP) classifier we introduce evolutionary resampling...
For supervised training of an artificial neural network (ANN) the size and composition of the training data set (TDS) is one of the most important prerequisites for an ANN with good generalization abilities. The exisiting methods to construct training data sets (TDSs) vary from random sampling to incrementally increasing the number of samples in th...
In this article we present work on chromosome structures for genetic algorithms (GAs) based on biological principles. Mainly, the influence of noncoding segments on GA behavior and performance is investigated. We compare representations with noncoding sequences at predefined, fixed locations with "junk" code induced by the use of promoter/terminato...
At the workshop we would like to present an overview of our work on chromosome structures for genetic algorithms based on biological principles. The use of promoter/terminator sequences (the initials pt and genetic algorithms (GAs) are blended to ptGA) defining start and end of a gene not only enables evolution of parameter values, but also allows...
While most efforts in artificial neural network (ANN) research
have been put into the investigation of network types, network
topologies, various types of neurons and training algorithms, work on
training data sets (TDSs) for ANNs has been comparably small. There are
some approximations for the size of ANN TDSs, but little is known about
the qualit...
A main criterion for the accuracy of solutions of Artificial Neural Networks (ANNs) for classification tasks is the architecture. In order to find problem-adapted topologies of ANNs, we adopted the evolutionary approach to ANN design by employing a Genetic Algorithm (GA) to evolve ANNs which are represented using a direct encoding method.
The role...
A main criterion for the accuracy of solutions of an Artificial Neural Network (ANN) for classification tasks is the architecture. In order to find problem-adapted topologies of ANNs, we adopted the evolutionary approach to ANN design by employing a Genetic Algorithm (GA) to evolve ANNs which are represented using a direct encoding method. As ANNs...
Artificial Neural Networks (ANN) have gained increasing popularity as
an alternative to statistical methods for classification of Remote
Sensing Data. Their superiority to some of the classical statistical
methods has been shown in the literature. Therefore, ANNs are commonly used
for segmentation and classification purposes. We address the problem...
Artificial Neural Networks (ANN) have gained increasing popularity as an alternative to statistical methods for classification of Remote Sensing Data. Their superiority to some of the classical statistical methods has been shown in the literature. Therefore, ANNs are commonly used for segmentation and classification purposes. We address the problem...
Artificial neural networks (ANNs) have shown to perform satisfactorily for pattern recognition tasks. It has also been shown that ANNs are superior to some of the classical statistical methods in pattern classification, but little is known how to design the ANN. A genetic algorithm (GA) based method can be used to determine the ANN architecture for...
Artificial Neural Networks (ANN) have gained increasing popularity as an alternative to statistical methods for classification of Remote Sensing Data. Their superiority to some of the classical statistical methods has been shown in the literature. Therefore, ANNs are commonly used for segmentation and classification purposes. In this paper we addre...
Questions
Question (1)
Is there an algorithm to train a SNN for some classification benchmark problem? If so, how are the inputs/outputs encoded (as the feature vectors of classification problems are usually numbers, but not spike trains)?