Evolution of Neural Networks in a Multi-Agent World

Size: px
Start display at page:

Download "Evolution of Neural Networks in a Multi-Agent World"

Transcription

1 ZESZYTY NAUKOWE UNIWERSYTETU JAGIELLOŃSKIEGO MMCCXLII PRACE INFORMATYCZNE Z Evolution of Neural Networks in a Multi-Agent World Marek Kisiel-Dorohinicki Magdalena Klapper-Rybicka Institute of Computer Science University of Mining and Metallurgy Kraków, Poland {doroh,mklapper}@agh.edu.pl Abstract. The concept of decentralised evolutionary computation realised as evolutionary multi- system (EMAS) is described in the paper. Also based evolutionary approach to neural network architecture optimisation is presented. Then the problem of time-series prediction and a general idea of evolutionary neural multi- predicting system is introduced. Selected design issues together with preliminary simulation results conclude the work. Keywords: decentralised evolutionary computation, evolutionary multi- systems, evolutionary neural networks, time-series prediction. 1. Introduction Artificial neural networks (ANN) is a computational technique arising from an interest in modeling basic mechanisms of biological neural structures. Probably the main advantage of using neural networks is their ability to learn from examples and generalise the aquired knowledge to new cases in such a way that any explicit problem-dependent knowledge is not needed (eg. [11]). That is why this technique is dedicated to approximation problems, like classification (pattern recognition), prediction, and control problems. Even though the use of neural networks replaces the necessity to specify the deterministic way of solving a problem, one still needs to define the network architecture suitable for that (kind of) problem. It usually involves a

2 38 lot of experiments with different network structures, which is time-consuming and does not have to result in the best solution. To avoid this situation, some mechanisms to automatically find the (nearly) best neural network architecture were proposed (eg. [18]). Evolutionary algorithms (EA) or evolutionary computation (EC) is another heuristic problem-solving approach, which is based on models of organic evolution [3]. It has been successfully used in various computationally difficult search and optimisation problems. Recently evolutionary algorithms have also been used to support search for optimal neural network architecture [5, 15]. Yet until, now the question of how to build a proper evolutionary algorithm for any given problem remains open. We have not only to decide on the representation of individuals, selection/reproduction mechanisms and genetic operators, but also have to set various parameters. One of the reasons for that is the fact, that the theory of evolutionary computation is still weak [4]. There are also many important differences between the observed mechanisms of evolution and its model employed by EA. What surely limits the possibilities of EA s to model natural evolution is the centralisation of the evolution process. What we propose is to model evolution in a decentralised environment, such as multi- system. Thus defined evolutionary multi- system (EMAS) may support searching for the correct architecture of a neural network. The resulting evolutionaryneural system may be applied to the problem of time-series prediction. 2. Evolutionary (Multi-)Agent Systems Recently, technology has gained much interest both in industry and academic research, owing to the growing need for decentralised intelligent software systems. Among lots of issues connected with multi- systems (MAS) one can also find evolutionary computation supporting systems. In most such cases an evolutionary algorithm is used by an (Fig. 1a), to aid the realisation of some of its tasks, e.g. connected with learning or reasoning [8, 12]. However, it seems that interesting results may be achieved when applying some model of evolution in MAS at a population level (Fig. 1b), i.e. among s. In this case genetic operators together with the selection/reproduction mechanisms search for a (near) optimal configuration of s in the population. Such systems form a new class of auto-adaptive multi- systems (evolutionary multi- systems), where organic-like evolution helps to accomplish population-level goals [6, 7].

3 39 Evolutionary multi- systems may also be considered a new computational method based on model of decentralised evolution. Of course, these two approaches can be used at the same time, which may be considered an experiment with meta-evolution (cf. [3]). MAS evolutionary algorithm a) evolutionary algorithm evolutionary algorithm EMAS process of evolution b) Fig. 1. Evolutionary algorithms supporting particular s in MAS (a) and evolutionary multi- system (b) 2.1. Classical evolutionary computation Various techniques of evolutionary computation were developed [4], including genetic algorithms (GA), evolution strategies (ES) and evolutionary programming (EP). Also many variants of these canonical forms of EC, as well as hybrid methods joining not only different evolutionary algorithms, but also distinct heuristics or problem-dependent mechanisms are in use. While different forms of evolutionary computation model evolution of organisms at a specific level [2] (GA gene level, ES individual level, EP species level), they all employ a similar scheme. They operate on a given number of data structures (population) and repeat the same cycle of processing (generation) consisting of a selection of parents and generation of offspring using mutation and recombination operators. The differences between specific techniques concern mainly:

4 40 what type of structures are evolved (representation of individuals), how the structures are manipulated to produce offspring (mutation and/or recombination), how the structures are evaluated (selection/reproduction mechanism). As it was already suggested, the selection of a proper evolutionary algorithm for the given problem is not a trivial problem. One of the reasons for that is the weakness of the theory of evolutionary computation: We know that they work, but we do not know why [4]. One may find several mathematical models of how specific evolutionary algorithms work. Moreover, in particular cases it is even possible to prove the convergence of the algorithm to the desired solution. But there is no coherent theory explaining how the above-described scheme leads to satisfactory results in a general case. The only explanation for us still remains the fact that mechanisms exploited by EC are based on, what we think is, natural evolution. But the model of evolution followed by most EAs (with noticeable exceptions) is much simplified and lacks many important features observed in organic evolution [4]: dynamically changing environmental conditions, many criteria in consideration, neither global knowledge nor generational synchronisation assumed, co-evolution of species, evolving genotype-fenotype mapping, etc. What is more, since the mechanisms of natural evolution are not clear to us yet, surely our models are hardly perfect. The idea of decentralised evolutionary computation, realised as evolutionary multi- system introduced above, may help to avoid some of the above-mentioned shortcomings of the model of evolution employed in classical EC The model of -based evolutionary computation A general idea of how evolutionary multi- system should work has been presented in this section. The population of s remains in a common environment. The environment may have a spatial structure and contain some resources. The space and the resources may represent some aspects of the problem to be solved or aid the process of solving. The s may sense the environment (both

5 41 space structure and resources) and act over it. The states of the s or the results of s actions should constitute a basis for the formulation of a solution (in the simplest cases the state of the or its position in the space may represent a solution or the may return a solution to the environment). The set of parameters describing the basic behaviour of the is encoded in its genotype, and is inherited from its parent(s). Besides the may possess some knowledge acquired during its life (model of the environment), which is not inherited. Both the inherited and aquired information determine the behaviour of the in the system (phenotype). Special resource(s) energetic profile, life energy play the role of a fitness function. The energy is gained and lost when the executes actions in the environment. An increase in energy is a reward for a good behaviour of the, decrease penalty for a bad behaviour. The best individuals are selected for reproduction based on the value of its life energy. Before generation of an offspring, the genetic material is (ex)changed by means of genetic operators (mutation and recombination) Advantages and disadvantages of EMAS-based computation The presented model of decentralised evolutionary computation enables the following: continuous evolution process means that evolution (when genotypes are developed) takes place in the same time space as the life of particular s (when fenotypes are developed), life energy is a more flexible way to evaluate s fitness; besides more than one criterion can be reflected with it, explicitly defined living space allows for application of advanced search mechanisms such as territorial niches, evolution process may be supported by new operations, e.g. aggregation enables a group of s to join together and act as a single and in this way modify their own environment, escape is a strong qualitative change of the s environment, which may be done by means of migration, evolution centers accelerate specialisation and support aggregation.

6 42 Of course, since EMAS may be regarded as a step towards a more complete model of evolution (surely more complicated one), it has two serious disadvantages: it is computationally complex (much more than the generic EA), there is no theory (like for the generic evolutionary algorithm) entirely explaining the basics of its work. Yet, as the work on theory of EC and MAS goes on, in the future it would be possible to present also theoretical foundations of the new ideas. 3. Evolutionary Neural Networks Neural networks are able to model very complex functions and non-linear structures with large numbers of variables, which are troublesome for traditional computational methods. Yet, the application of a neural network requires defining its topology (number of layers, number and type of neurons in each layer, connection structure) and initial state (weights), as well as choosing suitable training method and its parameters. The problem is meaningful, especially that too low complexity of a network may cause that it would not be able to project the function underlying the problem. On the other hand, too high complexity of a network may result in overfitting to the data. Also (time-)complexity of the learning algorithm strongly depends on the complexity of the network, especially on the number of connections Optimisation of NN architecture How to select the most appropriate architecture of a neural network for a particular application? Unfortunately, this question cannot be answered arbitrarily. Of course, some general rules of designing a network may be used. However, tuning the network precisely to a particular problem requires testing of many cases. There are several methods of optimisation of a neural network topology. The constructive methods start from a small network with a limited number of connections. Then successively add new units, training the network after each addition. In opposite, pruning algorithms start from a large full-connected network and decrease the number of units. In fact, manual optimisation of a neural network takes a lot of effort. And that is why, automatic realisations of

7 43 the above-mentioned algorithms have been proposed [5, 15]. The evolutionary neural networks (ENN), in which the search for a desirable neural architecture is made by the evolutionary algorithm is an alternative [13, 18] Genotypic representation of NN The most important question related to the design of an evolutionary algorithm for NN optimisation is the decision on the genotypic representation of a neural network. When designing an NN genotypic representation, the correctness of the resulting network architecture must be assured (e.g. there must exist some connections between two succeeding layers). Another problem is that one network topology may be represented by different chromosomes, which makes evolution process less efficient (the so-called permutation problem). If the evolution of architectures is separated from weights training, the fitness evaluation of the NN is always noisy. Two kinds of representation may be distinguished in view of the amount of: information about network topology coded into a chromosome, direct encoding (strong specification scheme) a chromosome contains full information about network nodes and connections [16], indirect encoding (weak specification scheme) developmental rules for generating a network architecture are encoded in a chromosome [10, 14]. Direct encoding requires longer chromosomes, which may slow down the process of evolution. Indirect encoding suffers from the effect of noisy fitness evaluation [17]. 4. EMAS for time-series prediction Prediction (or forecasting) is the generation of information about the possible future development of some process from data about its past and present behaviour [9]. The prediction may concern static data (as the outcome of a decease) as well as time-series data (as stock market or gas consumption). The latter one will be of our interest. Time-series prediction performed on the basis of only one independent variable (univariate) consists in searching for some trends in the sequence

8 44 of values of that variable. Multivariate prediction should also respect the relationships between particular series Predicting MAS In the most general case a time-series predicting system may be considered as a black box with some input sequences and predictions of successive values of (some of) these sequences as an output. Some intelligent mechanism inside that box should be able to discover hidden regularities and relationships in and between the input sequences. On the assumption that the characteristics of the signal(s) may change in time, this mechanism should also be able to dynamically adapt to these changes ignoring different kinds of distortion and noise a 5 1 a a a 2 1 a 1 0 Fig. 2. Multi- predicting system When the signal to be predicted is much complicated (e.g. when different trends change over time or temporal variations in relationships between particular sequences are present0 an idea of a (multi-) predicting system may be introduced [7]. The predicting MAS (see Fig. 2) may be viewed as the box (as above) with a group of intelligent s inside. The subsequent elements of the input sequence(s) are supplied to their environment, where they become available for all s. Each itself may analyse the incoming data and give predictions of (a subset of) the next-to-come elements of input. Specialisation in function or time of particular s allow for obtaining better results by cooperation or competition in the common environment. On the basis of predictions of all s, a prediction of the whole system may be generated.

9 Evolution in predicting MAS The configuration of s in such-defined system (kind of specialisation or method of cooperation) is often difficult to specify. What is more, when dynamic changes of the characteristics of the signal are possible, the configuration of the s should reflect these changes, automatically adapting to the new characteristics. The mechanisms similar to the ones found in evolutionary computation may help to transform the whole population of s (by means of mutation and/or recombination) so that it best fits the current profile of the input signal (with proper selection/reproduction). In fact, each simply possesses some vector of parameters, which describes its behaviour in the system. This vector plays the role of an s genotype, and as such may be modified by genetic operators when inherited by its offspring. The evaluation of the s is based on the quality of prediction by means of the gained/lost life energy. Then the selection is made as good s are more likely to reproduce new s, while bad s die. This evolutionary development of predicting MAS meets the general idea of evolutionary multi- system (EMAS) introduced previously Prediction with neural networks Fig. 3. Predicting neural network

10 46 For each an artificial neural network may be used as a basic mechanism to trace signal regularities in a system for time-series prediction. Many types of neural networks may be applied for prediction problems: multilayer perceptron, radial basic function network, generalised regression neural network. The choice of an architecture is greatly determined by the particular problem. The scheme of a neural network used for the time-series prediction is presented in Fig. 3. Usually the next value of a series is predicted based on a fixed number of previous ones (an input register). The number of input neurons corresponds to the number of values the prediction is based on. A network may be supervisory trained, using the comparison between the predicted and real values. Then the input register is shifted left and the next value of input is appended. 5. Design issues The behaviour of an in the environment is determined by interaction of its energetic profile and prediction profile [7]. The main part of the prediction profile consists of a three-layer fullconnected MLP network (multilayer perceptron). The fixed number of previous values of the series the network needs to predict the next one is held in a sequence-parallel register (see section 4.3). The network is trained with the back-propagation algorithm, comparing the last-predicted value with the real value of a series taken from the environment. The adjustment of the network is described by the quality of prediction, which is estimated basing on the accuracy of the fixed number of last predictions. The only resource possessed by the in the energetic profile is energy. The gains energy for accurate prediction. At the same time each action results in a decrease in the s energy. On the other hand, to perform an action the has to reach an appropriate energetic level. Some probability parameters prevent the from performing the same action several times one by one. The system is based on the event-driven simulation, i.e. the acts in response to the events registered by itself or other s, or incoming data. Subsequent elements of the input series supplied to the environment form the main stages of activity of the system. Within each stage the following scheme is realised: 1. succeeding value of the input series is delivered to the environment, 2. each acts according to the pattern:

11 47 (a) gets actual set of values from the environment, (b) estimates the quality of its own prediction, (c) trains the network with back propagation algorithm, (d) decides which action to perform mutation, migration, aggregation, reproduction or death, (e) predicts the next value, (f) submits the predicted value together with the quality of prediction to the environment, 3. the environment generates output prediction, based on the values submitted by all s; it chooses the best (or the most frequent) value of s answers, or a value computed as their weighted average. 6. Results of simulation experiments 1 0,8 0,6 0,4 0,2 0 Fig. 4. Avarage prediction probability in a typical simulation Preliminary results of simulation experiments are presented in Fig. 4 and Fig.5. The average prediction probability of s is shown in the first plot. One may notice that after several steps of evolution the average probability of prediction of individual s is about 80% and from that point the answers of the whole system are about 100% correct. The second plot shows the percent all of deaths zero at the begining of the simulation, then a significant rise due to elimination of the worst individuals, and stabilisation at a level of 10% when predictions are satisfactory (meaning a low exchange rate of the population).

12 48 7. Conclusions The idea of EMAS an extension of classical evolutionary algorithms. Evolving neural networks in a multi- environment an example of how an system allows for simultanous development of s phenotypes and evolution of the whole population of s. Predicting neural networks in evolutionary multi- environment a new approach to the problem of time-series prediction. Preliminary experimental studies new ideas at work. 0,4 0,35 0,3 0,25 0,2 0,15 0,1 0,05 0 Fig. 5. Percent of deaths in a typical simulation 8. References [1] Proceedings of Second International Conference on Multi-Agent Systems, AAAI Press, [2] Angeline P.J., Genetic Programming s Continued Evolution, in: Angeline P.J., and Kinnear K.E., Jr. (eds.), Advances in Genetic Programming, MIT Press, 1996, pp

13 49 [3] Bäck T., Evolutionary Algorithms in Theory and Practice, Oxford University Press, [4] Bäck T., Hammel U., Schwefel H.P., Evolutionary Computation: Comments on the History and Current State, IEEE Transactions on Evolutionary Computation, (1), 1997, pp [5] Burgess N., A Constructive Algorithm that Converges for Realvalued Input Patterns, International Journal of Neural Systems, 1994, pp [6] Cetnarowicz K., Evolution in Multi-Agent World = Genetic Algorithms + Aggregation + Escape, in: Position Papers of MAAMAW 96, Technical Report 96-1, Vrije Universiteit Brussel, Artificial Intelligence Laboratory. [7] Cetnarowicz K., Kisiel-Dorohinicki M., Nawarecki E., The Application of Evolution Process in Multi-Agent World (MAW) to the Prediction System, in: Proc. of ICMAS 96, [1]. [8] Denzinger J., Fuchs M., Experiments in Learning Prototypical Situations for Variants of the Pursuit Game, in: Proc. of ICMAS 96 [1]. [9] Kasabov N.K., Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering, The MIT Press, [10] Kitano H., Designing Neural Network Using Genetic Algorithm with Graph Generation System, Complex Systems, 1990, pp [11] Krose B., van der Smagt P., An Introduction to Neural Networks, University of Amsterdam, [12] Liu J., Qin H., Adaptation and Learning in Animated Creatures, in: Proc. of Autonomous Agents 97, ACM Press, [13] Mitchell M., An Introduction to Genetic Algorithms, MIT Press, [14] Nolfi S., Parisi D., The Handbook of Brain Theory and Neural Networks, MIT Press, 1994, Chapter: Genotypes for neural networks. [15] Reed R., Pruning Algorithms a Survey, IEEE Trans. on Neural Networks, 1993, pp [16] Whitley L.D., and Schaffer J.D. (eds.), IEEE Computer Society Press, [17] Yao X., Evolutionary Artificial Neural Networks, International Journal of Neural Systems, 1993, pp

14 50 [18] Yao X., Liu Y., in: Proceedings of the Fifth Annual Conference on Evolutionary Programming, MIT Press, Received March 18, 1999 Department of Computer Science Jagiellonian University

Genetic Algorithm: introduction

Genetic Algorithm: introduction 1 Genetic Algorithm: introduction 2 The Metaphor EVOLUTION Individual Fitness Environment PROBLEM SOLVING Candidate Solution Quality Problem 3 The Ingredients t reproduction t + 1 selection mutation recombination

More information

ARTIFICIAL NEURAL NETWORK PART I HANIEH BORHANAZAD

ARTIFICIAL NEURAL NETWORK PART I HANIEH BORHANAZAD ARTIFICIAL NEURAL NETWORK PART I HANIEH BORHANAZAD WHAT IS A NEURAL NETWORK? The simplest definition of a neural network, more properly referred to as an 'artificial' neural network (ANN), is provided

More information

NEUROEVOLUTION. Contents. Evolutionary Computation. Neuroevolution. Types of neuro-evolution algorithms

NEUROEVOLUTION. Contents. Evolutionary Computation. Neuroevolution. Types of neuro-evolution algorithms Contents Evolutionary Computation overview NEUROEVOLUTION Presenter: Vlad Chiriacescu Neuroevolution overview Issues in standard Evolutionary Computation NEAT method Complexification in competitive coevolution

More information

Overview of ECNN Combinations

Overview of ECNN Combinations 1 Overview of ECNN Combinations Evolutionary Computation ECNN Neural Networks by Paulo Cortez and Miguel Rocha pcortez@dsi.uminho.pt mrocha@di.uminho.pt (Presentation available at: http://www.dsi.uminho.pt/

More information

Artificial Intelligence (AI) Common AI Methods. Training. Signals to Perceptrons. Artificial Neural Networks (ANN) Artificial Intelligence

Artificial Intelligence (AI) Common AI Methods. Training. Signals to Perceptrons. Artificial Neural Networks (ANN) Artificial Intelligence Artificial Intelligence (AI) Artificial Intelligence AI is an attempt to reproduce intelligent reasoning using machines * * H. M. Cartwright, Applications of Artificial Intelligence in Chemistry, 1993,

More information

Neural Systems and Artificial Life Group, Institute of Psychology, National Research Council, Rome. Evolving Modular Architectures for Neural Networks

Neural Systems and Artificial Life Group, Institute of Psychology, National Research Council, Rome. Evolving Modular Architectures for Neural Networks Neural Systems and Artificial Life Group, Institute of Psychology, National Research Council, Rome Evolving Modular Architectures for Neural Networks Andrea Di Ferdinando, Raffaele Calabretta and Domenico

More information

Application of Artificial Neural Networks in Evaluation and Identification of Electrical Loss in Transformers According to the Energy Consumption

Application of Artificial Neural Networks in Evaluation and Identification of Electrical Loss in Transformers According to the Energy Consumption Application of Artificial Neural Networks in Evaluation and Identification of Electrical Loss in Transformers According to the Energy Consumption ANDRÉ NUNES DE SOUZA, JOSÉ ALFREDO C. ULSON, IVAN NUNES

More information

Evolutionary computation

Evolutionary computation Evolutionary computation Andrea Roli andrea.roli@unibo.it Dept. of Computer Science and Engineering (DISI) Campus of Cesena Alma Mater Studiorum Università di Bologna Outline 1 Basic principles 2 Genetic

More information

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others)

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others) Machine Learning Neural Networks (slides from Domingos, Pardo, others) For this week, Reading Chapter 4: Neural Networks (Mitchell, 1997) See Canvas For subsequent weeks: Scaling Learning Algorithms toward

More information

AN INTRODUCTION TO NEURAL NETWORKS. Scott Kuindersma November 12, 2009

AN INTRODUCTION TO NEURAL NETWORKS. Scott Kuindersma November 12, 2009 AN INTRODUCTION TO NEURAL NETWORKS Scott Kuindersma November 12, 2009 SUPERVISED LEARNING We are given some training data: We must learn a function If y is discrete, we call it classification If it is

More information

Choosing Variables with a Genetic Algorithm for Econometric models based on Neural Networks learning and adaptation.

Choosing Variables with a Genetic Algorithm for Econometric models based on Neural Networks learning and adaptation. Choosing Variables with a Genetic Algorithm for Econometric models based on Neural Networks learning and adaptation. Daniel Ramírez A., Israel Truijillo E. LINDA LAB, Computer Department, UNAM Facultad

More information

Data Mining Part 5. Prediction

Data Mining Part 5. Prediction Data Mining Part 5. Prediction 5.5. Spring 2010 Instructor: Dr. Masoud Yaghini Outline How the Brain Works Artificial Neural Networks Simple Computing Elements Feed-Forward Networks Perceptrons (Single-layer,

More information

Introduction to Artificial Neural Networks

Introduction to Artificial Neural Networks Facultés Universitaires Notre-Dame de la Paix 27 March 2007 Outline 1 Introduction 2 Fundamentals Biological neuron Artificial neuron Artificial Neural Network Outline 3 Single-layer ANN Perceptron Adaline

More information

An Evolution Strategy for the Induction of Fuzzy Finite-state Automata

An Evolution Strategy for the Induction of Fuzzy Finite-state Automata Journal of Mathematics and Statistics 2 (2): 386-390, 2006 ISSN 1549-3644 Science Publications, 2006 An Evolution Strategy for the Induction of Fuzzy Finite-state Automata 1,2 Mozhiwen and 1 Wanmin 1 College

More information

Evolutionary computation

Evolutionary computation Evolutionary computation Andrea Roli andrea.roli@unibo.it DEIS Alma Mater Studiorum Università di Bologna Evolutionary computation p. 1 Evolutionary Computation Evolutionary computation p. 2 Evolutionary

More information

Towards Synthesizing Artificial Neural Networks that Exhibit Cooperative Intelligent Behavior: Some Open Issues in Artificial Life Michael G.

Towards Synthesizing Artificial Neural Networks that Exhibit Cooperative Intelligent Behavior: Some Open Issues in Artificial Life Michael G. Towards Synthesizing Artificial Neural Networks that Exhibit Cooperative Intelligent Behavior: Some Open Issues in Artificial Life Michael G. Dyer Computer Science Department, UCLA Overview Introduction

More information

Lecture 9 Evolutionary Computation: Genetic algorithms

Lecture 9 Evolutionary Computation: Genetic algorithms Lecture 9 Evolutionary Computation: Genetic algorithms Introduction, or can evolution be intelligent? Simulation of natural evolution Genetic algorithms Case study: maintenance scheduling with genetic

More information

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others)

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others) Machine Learning Neural Networks (slides from Domingos, Pardo, others) Human Brain Neurons Input-Output Transformation Input Spikes Output Spike Spike (= a brief pulse) (Excitatory Post-Synaptic Potential)

More information

COMPARING PERFORMANCE OF NEURAL NETWORKS RECOGNIZING MACHINE GENERATED CHARACTERS

COMPARING PERFORMANCE OF NEURAL NETWORKS RECOGNIZING MACHINE GENERATED CHARACTERS Proceedings of the First Southern Symposium on Computing The University of Southern Mississippi, December 4-5, 1998 COMPARING PERFORMANCE OF NEURAL NETWORKS RECOGNIZING MACHINE GENERATED CHARACTERS SEAN

More information

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others)

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others) Machine Learning Neural Networks (slides from Domingos, Pardo, others) For this week, Reading Chapter 4: Neural Networks (Mitchell, 1997) See Canvas For subsequent weeks: Scaling Learning Algorithms toward

More information

Balancing and Control of a Freely-Swinging Pendulum Using a Model-Free Reinforcement Learning Algorithm

Balancing and Control of a Freely-Swinging Pendulum Using a Model-Free Reinforcement Learning Algorithm Balancing and Control of a Freely-Swinging Pendulum Using a Model-Free Reinforcement Learning Algorithm Michail G. Lagoudakis Department of Computer Science Duke University Durham, NC 2778 mgl@cs.duke.edu

More information

A FUZZY NEURAL NETWORK MODEL FOR FORECASTING STOCK PRICE

A FUZZY NEURAL NETWORK MODEL FOR FORECASTING STOCK PRICE A FUZZY NEURAL NETWORK MODEL FOR FORECASTING STOCK PRICE Li Sheng Institute of intelligent information engineering Zheiang University Hangzhou, 3007, P. R. China ABSTRACT In this paper, a neural network-driven

More information

Chapter 1 Introduction

Chapter 1 Introduction Chapter 1 Introduction 1.1 Introduction to Chapter This chapter starts by describing the problems addressed by the project. The aims and objectives of the research are outlined and novel ideas discovered

More information

Evolutionary Computation. DEIS-Cesena Alma Mater Studiorum Università di Bologna Cesena (Italia)

Evolutionary Computation. DEIS-Cesena Alma Mater Studiorum Università di Bologna Cesena (Italia) Evolutionary Computation DEIS-Cesena Alma Mater Studiorum Università di Bologna Cesena (Italia) andrea.roli@unibo.it Evolutionary Computation Inspiring principle: theory of natural selection Species face

More information

Neural Networks. Nicholas Ruozzi University of Texas at Dallas

Neural Networks. Nicholas Ruozzi University of Texas at Dallas Neural Networks Nicholas Ruozzi University of Texas at Dallas Handwritten Digit Recognition Given a collection of handwritten digits and their corresponding labels, we d like to be able to correctly classify

More information

Lecture 7 Artificial neural networks: Supervised learning

Lecture 7 Artificial neural networks: Supervised learning Lecture 7 Artificial neural networks: Supervised learning Introduction, or how the brain works The neuron as a simple computing element The perceptron Multilayer neural networks Accelerated learning in

More information

A Hybrid Method of CART and Artificial Neural Network for Short-term term Load Forecasting in Power Systems

A Hybrid Method of CART and Artificial Neural Network for Short-term term Load Forecasting in Power Systems A Hybrid Method of CART and Artificial Neural Network for Short-term term Load Forecasting in Power Systems Hiroyuki Mori Dept. of Electrical & Electronics Engineering Meiji University Tama-ku, Kawasaki

More information

A Feature Based Neural Network Model for Weather Forecasting

A Feature Based Neural Network Model for Weather Forecasting World Academy of Science, Engineering and Technology 4 2 A Feature Based Neural Network Model for Weather Forecasting Paras, Sanjay Mathur, Avinash Kumar, and Mahesh Chandra Abstract Weather forecasting

More information

MLPR: Logistic Regression and Neural Networks

MLPR: Logistic Regression and Neural Networks MLPR: Logistic Regression and Neural Networks Machine Learning and Pattern Recognition Amos Storkey Amos Storkey MLPR: Logistic Regression and Neural Networks 1/28 Outline 1 Logistic Regression 2 Multi-layer

More information

Functional Preprocessing for Multilayer Perceptrons

Functional Preprocessing for Multilayer Perceptrons Functional Preprocessing for Multilayer Perceptrons Fabrice Rossi and Brieuc Conan-Guez Projet AxIS, INRIA, Domaine de Voluceau, Rocquencourt, B.P. 105 78153 Le Chesnay Cedex, France CEREMADE, UMR CNRS

More information

Artificial Neural Networks Examination, March 2004

Artificial Neural Networks Examination, March 2004 Artificial Neural Networks Examination, March 2004 Instructions There are SIXTY questions (worth up to 60 marks). The exam mark (maximum 60) will be added to the mark obtained in the laborations (maximum

More information

Outline. MLPR: Logistic Regression and Neural Networks Machine Learning and Pattern Recognition. Which is the correct model? Recap.

Outline. MLPR: Logistic Regression and Neural Networks Machine Learning and Pattern Recognition. Which is the correct model? Recap. Outline MLPR: and Neural Networks Machine Learning and Pattern Recognition 2 Amos Storkey Amos Storkey MLPR: and Neural Networks /28 Recap Amos Storkey MLPR: and Neural Networks 2/28 Which is the correct

More information

Evolutionary Computation

Evolutionary Computation Evolutionary Computation - Computational procedures patterned after biological evolution. - Search procedure that probabilistically applies search operators to set of points in the search space. - Lamarck

More information

Keywords- Source coding, Huffman encoding, Artificial neural network, Multilayer perceptron, Backpropagation algorithm

Keywords- Source coding, Huffman encoding, Artificial neural network, Multilayer perceptron, Backpropagation algorithm Volume 4, Issue 5, May 2014 ISSN: 2277 128X International Journal of Advanced Research in Computer Science and Software Engineering Research Paper Available online at: www.ijarcsse.com Huffman Encoding

More information

ARTIFICIAL NEURAL NETWORKS گروه مطالعاتي 17 بهار 92

ARTIFICIAL NEURAL NETWORKS گروه مطالعاتي 17 بهار 92 ARTIFICIAL NEURAL NETWORKS گروه مطالعاتي 17 بهار 92 BIOLOGICAL INSPIRATIONS Some numbers The human brain contains about 10 billion nerve cells (neurons) Each neuron is connected to the others through 10000

More information

CSE 417T: Introduction to Machine Learning. Final Review. Henry Chai 12/4/18

CSE 417T: Introduction to Machine Learning. Final Review. Henry Chai 12/4/18 CSE 417T: Introduction to Machine Learning Final Review Henry Chai 12/4/18 Overfitting Overfitting is fitting the training data more than is warranted Fitting noise rather than signal 2 Estimating! "#$

More information

Evolution of Populations. Chapter 17

Evolution of Populations. Chapter 17 Evolution of Populations Chapter 17 17.1 Genes and Variation i. Introduction: Remember from previous units. Genes- Units of Heredity Variation- Genetic differences among individuals in a population. New

More information

A Simple Haploid-Diploid Evolutionary Algorithm

A Simple Haploid-Diploid Evolutionary Algorithm A Simple Haploid-Diploid Evolutionary Algorithm Larry Bull Computer Science Research Centre University of the West of England, Bristol, UK larry.bull@uwe.ac.uk Abstract It has recently been suggested that

More information

Agent-Based Simulation Model of Sexual Selection Mechanism

Agent-Based Simulation Model of Sexual Selection Mechanism Agent-Based Simulation Model of Sexual Selection Mechanism Rafał Dreżewski AGH University of Science and Technology, Department of Computer Science, Kraków, Poland drezew@agh.edu.pl Abstract. Agent-based

More information

ARTIFICIAL INTELLIGENCE. Artificial Neural Networks

ARTIFICIAL INTELLIGENCE. Artificial Neural Networks INFOB2KI 2017-2018 Utrecht University The Netherlands ARTIFICIAL INTELLIGENCE Artificial Neural Networks Lecturer: Silja Renooij These slides are part of the INFOB2KI Course Notes available from www.cs.uu.nl/docs/vakken/b2ki/schema.html

More information

SimAnt Simulation Using NEAT

SimAnt Simulation Using NEAT SimAnt Simulation Using NEAT Erin Gluck and Evelyn Wightman December 15, 2015 Abstract In nature, ants in a colony often work together to complete a common task. We want to see if we can simulate this

More information

Evolutionary Design I

Evolutionary Design I Evolutionary Design I Jason Noble jasonn@comp.leeds.ac.uk Biosystems group, School of Computing Evolutionary Design I p.1/29 This lecture Harnessing evolution in a computer program How to construct a genetic

More information

EE04 804(B) Soft Computing Ver. 1.2 Class 2. Neural Networks - I Feb 23, Sasidharan Sreedharan

EE04 804(B) Soft Computing Ver. 1.2 Class 2. Neural Networks - I Feb 23, Sasidharan Sreedharan EE04 804(B) Soft Computing Ver. 1.2 Class 2. Neural Networks - I Feb 23, 2012 Sasidharan Sreedharan www.sasidharan.webs.com 3/1/2012 1 Syllabus Artificial Intelligence Systems- Neural Networks, fuzzy logic,

More information

Lecture 4: Feed Forward Neural Networks

Lecture 4: Feed Forward Neural Networks Lecture 4: Feed Forward Neural Networks Dr. Roman V Belavkin Middlesex University BIS4435 Biological neurons and the brain A Model of A Single Neuron Neurons as data-driven models Neural Networks Training

More information

CSC 4510 Machine Learning

CSC 4510 Machine Learning 10: Gene(c Algorithms CSC 4510 Machine Learning Dr. Mary Angela Papalaskari Department of CompuBng Sciences Villanova University Course website: www.csc.villanova.edu/~map/4510/ Slides of this presenta(on

More information

Neural Networks and the Back-propagation Algorithm

Neural Networks and the Back-propagation Algorithm Neural Networks and the Back-propagation Algorithm Francisco S. Melo In these notes, we provide a brief overview of the main concepts concerning neural networks and the back-propagation algorithm. We closely

More information

HIGH PERFORMANCE ADAPTIVE INTELLIGENT DIRECT TORQUE CONTROL SCHEMES FOR INDUCTION MOTOR DRIVES

HIGH PERFORMANCE ADAPTIVE INTELLIGENT DIRECT TORQUE CONTROL SCHEMES FOR INDUCTION MOTOR DRIVES HIGH PERFORMANCE ADAPTIVE INTELLIGENT DIRECT TORQUE CONTROL SCHEMES FOR INDUCTION MOTOR DRIVES M. Vasudevan and R. Arumugam Department of Electrical and Electronics Engineering, Anna University, Chennai,

More information

CS:4420 Artificial Intelligence

CS:4420 Artificial Intelligence CS:4420 Artificial Intelligence Spring 2018 Neural Networks Cesare Tinelli The University of Iowa Copyright 2004 18, Cesare Tinelli and Stuart Russell a a These notes were originally developed by Stuart

More information

STATE GENERALIZATION WITH SUPPORT VECTOR MACHINES IN REINFORCEMENT LEARNING. Ryo Goto, Toshihiro Matsui and Hiroshi Matsuo

STATE GENERALIZATION WITH SUPPORT VECTOR MACHINES IN REINFORCEMENT LEARNING. Ryo Goto, Toshihiro Matsui and Hiroshi Matsuo STATE GENERALIZATION WITH SUPPORT VECTOR MACHINES IN REINFORCEMENT LEARNING Ryo Goto, Toshihiro Matsui and Hiroshi Matsuo Department of Electrical and Computer Engineering, Nagoya Institute of Technology

More information

Evolutionary Algorithms

Evolutionary Algorithms Evolutionary Algorithms a short introduction Giuseppe Narzisi Courant Institute of Mathematical Sciences New York University 31 January 2008 Outline 1 Evolution 2 Evolutionary Computation 3 Evolutionary

More information

Neural Networks. Nethra Sambamoorthi, Ph.D. Jan CRMportals Inc., Nethra Sambamoorthi, Ph.D. Phone:

Neural Networks. Nethra Sambamoorthi, Ph.D. Jan CRMportals Inc., Nethra Sambamoorthi, Ph.D. Phone: Neural Networks Nethra Sambamoorthi, Ph.D Jan 2003 CRMportals Inc., Nethra Sambamoorthi, Ph.D Phone: 732-972-8969 Nethra@crmportals.com What? Saying it Again in Different ways Artificial neural network

More information

AP Curriculum Framework with Learning Objectives

AP Curriculum Framework with Learning Objectives Big Ideas Big Idea 1: The process of evolution drives the diversity and unity of life. AP Curriculum Framework with Learning Objectives Understanding 1.A: Change in the genetic makeup of a population over

More information

What does it take to evolve behaviorally complex organisms?

What does it take to evolve behaviorally complex organisms? 1 Neural Systems and Artificial Life Group Institute of Psychology National Research Council, Rome What does it take to evolve behaviorally complex organisms? Raffaele Calabretta 1, Andrea Di Ferdinando

More information

Unit 8: Introduction to neural networks. Perceptrons

Unit 8: Introduction to neural networks. Perceptrons Unit 8: Introduction to neural networks. Perceptrons D. Balbontín Noval F. J. Martín Mateos J. L. Ruiz Reina A. Riscos Núñez Departamento de Ciencias de la Computación e Inteligencia Artificial Universidad

More information

The Perceptron. Volker Tresp Summer 2016

The Perceptron. Volker Tresp Summer 2016 The Perceptron Volker Tresp Summer 2016 1 Elements in Learning Tasks Collection, cleaning and preprocessing of training data Definition of a class of learning models. Often defined by the free model parameters

More information

Course 395: Machine Learning - Lectures

Course 395: Machine Learning - Lectures Course 395: Machine Learning - Lectures Lecture 1-2: Concept Learning (M. Pantic) Lecture 3-4: Decision Trees & CBC Intro (M. Pantic & S. Petridis) Lecture 5-6: Evaluating Hypotheses (S. Petridis) Lecture

More information

ECE521 Lecture 7/8. Logistic Regression

ECE521 Lecture 7/8. Logistic Regression ECE521 Lecture 7/8 Logistic Regression Outline Logistic regression (Continue) A single neuron Learning neural networks Multi-class classification 2 Logistic regression The output of a logistic regression

More information

Situation. The XPS project. PSO publication pattern. Problem. Aims. Areas

Situation. The XPS project. PSO publication pattern. Problem. Aims. Areas Situation The XPS project we are looking at a paradigm in its youth, full of potential and fertile with new ideas and new perspectives Researchers in many countries are experimenting with particle swarms

More information

A Particle Swarm Optimization (PSO) Primer

A Particle Swarm Optimization (PSO) Primer A Particle Swarm Optimization (PSO) Primer With Applications Brian Birge Overview Introduction Theory Applications Computational Intelligence Summary Introduction Subset of Evolutionary Computation Genetic

More information

Neural Networks DWML, /25

Neural Networks DWML, /25 DWML, 2007 /25 Neural networks: Biological and artificial Consider humans: Neuron switching time 0.00 second Number of neurons 0 0 Connections per neuron 0 4-0 5 Scene recognition time 0. sec 00 inference

More information

Multivariate Analysis, TMVA, and Artificial Neural Networks

Multivariate Analysis, TMVA, and Artificial Neural Networks http://tmva.sourceforge.net/ Multivariate Analysis, TMVA, and Artificial Neural Networks Matt Jachowski jachowski@stanford.edu 1 Multivariate Analysis Techniques dedicated to analysis of data with multiple

More information

Engineering Part IIB: Module 4F10 Statistical Pattern Processing Lecture 6: Multi-Layer Perceptrons I

Engineering Part IIB: Module 4F10 Statistical Pattern Processing Lecture 6: Multi-Layer Perceptrons I Engineering Part IIB: Module 4F10 Statistical Pattern Processing Lecture 6: Multi-Layer Perceptrons I Phil Woodland: pcw@eng.cam.ac.uk Michaelmas 2012 Engineering Part IIB: Module 4F10 Introduction In

More information

Short Term Load Forecasting Using Multi Layer Perceptron

Short Term Load Forecasting Using Multi Layer Perceptron International OPEN ACCESS Journal Of Modern Engineering Research (IJMER) Short Term Load Forecasting Using Multi Layer Perceptron S.Hema Chandra 1, B.Tejaswini 2, B.suneetha 3, N.chandi Priya 4, P.Prathima

More information

Forecasting & Futurism

Forecasting & Futurism Article from: Forecasting & Futurism December 2013 Issue 8 A NEAT Approach to Neural Network Structure By Jeff Heaton Jeff Heaton Neural networks are a mainstay of artificial intelligence. These machine-learning

More information

Modeling High-Dimensional Discrete Data with Multi-Layer Neural Networks

Modeling High-Dimensional Discrete Data with Multi-Layer Neural Networks Modeling High-Dimensional Discrete Data with Multi-Layer Neural Networks Yoshua Bengio Dept. IRO Université de Montréal Montreal, Qc, Canada, H3C 3J7 bengioy@iro.umontreal.ca Samy Bengio IDIAP CP 592,

More information

(Feed-Forward) Neural Networks Dr. Hajira Jabeen, Prof. Jens Lehmann

(Feed-Forward) Neural Networks Dr. Hajira Jabeen, Prof. Jens Lehmann (Feed-Forward) Neural Networks 2016-12-06 Dr. Hajira Jabeen, Prof. Jens Lehmann Outline In the previous lectures we have learned about tensors and factorization methods. RESCAL is a bilinear model for

More information

Multitask Learning of Environmental Spatial Data

Multitask Learning of Environmental Spatial Data 9th International Congress on Environmental Modelling and Software Brigham Young University BYU ScholarsArchive 6th International Congress on Environmental Modelling and Software - Leipzig, Germany - July

More information

Neural Networks biological neuron artificial neuron 1

Neural Networks biological neuron artificial neuron 1 Neural Networks biological neuron artificial neuron 1 A two-layer neural network Output layer (activation represents classification) Weighted connections Hidden layer ( internal representation ) Input

More information

Holdout and Cross-Validation Methods Overfitting Avoidance

Holdout and Cross-Validation Methods Overfitting Avoidance Holdout and Cross-Validation Methods Overfitting Avoidance Decision Trees Reduce error pruning Cost-complexity pruning Neural Networks Early stopping Adjusting Regularizers via Cross-Validation Nearest

More information

A SEASONAL FUZZY TIME SERIES FORECASTING METHOD BASED ON GUSTAFSON-KESSEL FUZZY CLUSTERING *

A SEASONAL FUZZY TIME SERIES FORECASTING METHOD BASED ON GUSTAFSON-KESSEL FUZZY CLUSTERING * No.2, Vol.1, Winter 2012 2012 Published by JSES. A SEASONAL FUZZY TIME SERIES FORECASTING METHOD BASED ON GUSTAFSON-KESSEL * Faruk ALPASLAN a, Ozge CAGCAG b Abstract Fuzzy time series forecasting methods

More information

Adaptive Generalized Crowding for Genetic Algorithms

Adaptive Generalized Crowding for Genetic Algorithms Carnegie Mellon University From the SelectedWorks of Ole J Mengshoel Fall 24 Adaptive Generalized Crowding for Genetic Algorithms Ole J Mengshoel, Carnegie Mellon University Severinio Galan Antonio de

More information

Genetic Engineering and Creative Design

Genetic Engineering and Creative Design Genetic Engineering and Creative Design Background genes, genotype, phenotype, fitness Connecting genes to performance in fitness Emergent gene clusters evolved genes MIT Class 4.208 Spring 2002 Evolution

More information

Artificial Neural Networks Examination, June 2005

Artificial Neural Networks Examination, June 2005 Artificial Neural Networks Examination, June 2005 Instructions There are SIXTY questions. (The pass mark is 30 out of 60). For each question, please select a maximum of ONE of the given answers (either

More information

Big Idea 1: The process of evolution drives the diversity and unity of life.

Big Idea 1: The process of evolution drives the diversity and unity of life. Big Idea 1: The process of evolution drives the diversity and unity of life. understanding 1.A: Change in the genetic makeup of a population over time is evolution. 1.A.1: Natural selection is a major

More information

Chapter 8: Introduction to Evolutionary Computation

Chapter 8: Introduction to Evolutionary Computation Computational Intelligence: Second Edition Contents Some Theories about Evolution Evolution is an optimization process: the aim is to improve the ability of an organism to survive in dynamically changing

More information

Evolutionary Functional Link Interval Type-2 Fuzzy Neural System for Exchange Rate Prediction

Evolutionary Functional Link Interval Type-2 Fuzzy Neural System for Exchange Rate Prediction Evolutionary Functional Link Interval Type-2 Fuzzy Neural System for Exchange Rate Prediction 3. Introduction Currency exchange rate is an important element in international finance. It is one of the chaotic,

More information

Artificial Neural Network Method of Rock Mass Blastability Classification

Artificial Neural Network Method of Rock Mass Blastability Classification Artificial Neural Network Method of Rock Mass Blastability Classification Jiang Han, Xu Weiya, Xie Shouyi Research Institute of Geotechnical Engineering, Hohai University, Nanjing, Jiangshu, P.R.China

More information

Artificial Neural Network

Artificial Neural Network Artificial Neural Network Contents 2 What is ANN? Biological Neuron Structure of Neuron Types of Neuron Models of Neuron Analogy with human NN Perceptron OCR Multilayer Neural Network Back propagation

More information

Neural Networks Introduction

Neural Networks Introduction Neural Networks Introduction H.A Talebi Farzaneh Abdollahi Department of Electrical Engineering Amirkabir University of Technology Winter 2011 H. A. Talebi, Farzaneh Abdollahi Neural Networks 1/22 Biological

More information

AN ARTIFICIAL NEURAL NETWORK MODEL FOR ROAD ACCIDENT PREDICTION: A CASE STUDY OF KHULNA METROPOLITAN CITY

AN ARTIFICIAL NEURAL NETWORK MODEL FOR ROAD ACCIDENT PREDICTION: A CASE STUDY OF KHULNA METROPOLITAN CITY Proceedings of the 4 th International Conference on Civil Engineering for Sustainable Development (ICCESD 2018), 9~11 February 2018, KUET, Khulna, Bangladesh (ISBN-978-984-34-3502-6) AN ARTIFICIAL NEURAL

More information

FEEDBACK GMDH-TYPE NEURAL NETWORK AND ITS APPLICATION TO MEDICAL IMAGE ANALYSIS OF LIVER CANCER. Tadashi Kondo and Junji Ueno

FEEDBACK GMDH-TYPE NEURAL NETWORK AND ITS APPLICATION TO MEDICAL IMAGE ANALYSIS OF LIVER CANCER. Tadashi Kondo and Junji Ueno International Journal of Innovative Computing, Information and Control ICIC International c 2012 ISSN 1349-4198 Volume 8, Number 3(B), March 2012 pp. 2285 2300 FEEDBACK GMDH-TYPE NEURAL NETWORK AND ITS

More information

A new method for short-term load forecasting based on chaotic time series and neural network

A new method for short-term load forecasting based on chaotic time series and neural network A new method for short-term load forecasting based on chaotic time series and neural network Sajjad Kouhi*, Navid Taghizadegan Electrical Engineering Department, Azarbaijan Shahid Madani University, Tabriz,

More information

Topic 7: Evolution. 1. The graph below represents the populations of two different species in an ecosystem over a period of several years.

Topic 7: Evolution. 1. The graph below represents the populations of two different species in an ecosystem over a period of several years. 1. The graph below represents the populations of two different species in an ecosystem over a period of several years. Which statement is a possible explanation for the changes shown? (1) Species A is

More information

Chapter 17: Population Genetics and Speciation

Chapter 17: Population Genetics and Speciation Chapter 17: Population Genetics and Speciation Section 1: Genetic Variation Population Genetics: Normal Distribution: a line graph showing the general trends in a set of data of which most values are near

More information

Data Mining. Preamble: Control Application. Industrial Researcher s Approach. Practitioner s Approach. Example. Example. Goal: Maintain T ~Td

Data Mining. Preamble: Control Application. Industrial Researcher s Approach. Practitioner s Approach. Example. Example. Goal: Maintain T ~Td Data Mining Andrew Kusiak 2139 Seamans Center Iowa City, Iowa 52242-1527 Preamble: Control Application Goal: Maintain T ~Td Tel: 319-335 5934 Fax: 319-335 5669 andrew-kusiak@uiowa.edu http://www.icaen.uiowa.edu/~ankusiak

More information

An Evolutionary Programming Based Algorithm for HMM training

An Evolutionary Programming Based Algorithm for HMM training An Evolutionary Programming Based Algorithm for HMM training Ewa Figielska,Wlodzimierz Kasprzak Institute of Control and Computation Engineering, Warsaw University of Technology ul. Nowowiejska 15/19,

More information

Mining Classification Knowledge

Mining Classification Knowledge Mining Classification Knowledge Remarks on NonSymbolic Methods JERZY STEFANOWSKI Institute of Computing Sciences, Poznań University of Technology COST Doctoral School, Troina 2008 Outline 1. Bayesian classification

More information

Introduction to Natural Computation. Lecture 9. Multilayer Perceptrons and Backpropagation. Peter Lewis

Introduction to Natural Computation. Lecture 9. Multilayer Perceptrons and Backpropagation. Peter Lewis Introduction to Natural Computation Lecture 9 Multilayer Perceptrons and Backpropagation Peter Lewis 1 / 25 Overview of the Lecture Why multilayer perceptrons? Some applications of multilayer perceptrons.

More information

ARTIFICIAL NEURAL NETWORKS APPROACH IN MICROWAVE FILTER TUNING

ARTIFICIAL NEURAL NETWORKS APPROACH IN MICROWAVE FILTER TUNING Progress In Electromagnetics Research M, Vol. 13, 173 188, 2010 ARTIFICIAL NEURAL NETWORKS APPROACH IN MICROWAVE FILTER TUNING J. J. Michalski TeleMobile Electronics Ltd. Pomeranian Science and Technology

More information

Enduring understanding 1.A: Change in the genetic makeup of a population over time is evolution.

Enduring understanding 1.A: Change in the genetic makeup of a population over time is evolution. The AP Biology course is designed to enable you to develop advanced inquiry and reasoning skills, such as designing a plan for collecting data, analyzing data, applying mathematical routines, and connecting

More information

Inferring a System of Differential Equations for a Gene Regulatory Network by using Genetic Programming

Inferring a System of Differential Equations for a Gene Regulatory Network by using Genetic Programming Inferring a System of Differential Equations for a Gene Regulatory Network by using Genetic Programming Erina Sakamoto School of Engineering, Dept. of Inf. Comm.Engineering, Univ. of Tokyo 7-3-1 Hongo,

More information

A. Pelliccioni (*), R. Cotroneo (*), F. Pungì (*) (*)ISPESL-DIPIA, Via Fontana Candida 1, 00040, Monteporzio Catone (RM), Italy.

A. Pelliccioni (*), R. Cotroneo (*), F. Pungì (*) (*)ISPESL-DIPIA, Via Fontana Candida 1, 00040, Monteporzio Catone (RM), Italy. Application of Neural Net Models to classify and to forecast the observed precipitation type at the ground using the Artificial Intelligence Competition data set. A. Pelliccioni (*), R. Cotroneo (*), F.

More information

Active Guidance for a Finless Rocket using Neuroevolution

Active Guidance for a Finless Rocket using Neuroevolution Active Guidance for a Finless Rocket using Neuroevolution Gomez, F.J. & Miikulainen, R. (2003). Genetic and Evolutionary Computation Gecco, 2724, 2084 2095. Introduction Sounding rockets are used for making

More information

Intelligent Modular Neural Network for Dynamic System Parameter Estimation

Intelligent Modular Neural Network for Dynamic System Parameter Estimation Intelligent Modular Neural Network for Dynamic System Parameter Estimation Andrzej Materka Technical University of Lodz, Institute of Electronics Stefanowskiego 18, 9-537 Lodz, Poland Abstract: A technique

More information

The Perceptron. Volker Tresp Summer 2014

The Perceptron. Volker Tresp Summer 2014 The Perceptron Volker Tresp Summer 2014 1 Introduction One of the first serious learning machines Most important elements in learning tasks Collection and preprocessing of training data Definition of a

More information

Non-Adaptive Evolvability. Jeff Clune

Non-Adaptive Evolvability. Jeff Clune Non-Adaptive Evolvability! Jeff Clune Assistant Professor Computer Science Evolving Artificial Intelligence Laboratory Evolution Fails to Optimize Mutation Rates (though it would improve evolvability)

More information

The Agent-Based Model and Simulation of Sexual Selection and Pair Formation Mechanisms

The Agent-Based Model and Simulation of Sexual Selection and Pair Formation Mechanisms entropy Article The Agent-Based Model and Simulation of Sexual Selection and Pair Formation Mechanisms Rafał Dreżewski ID AGH University of Science and Technology, Department of Computer Science, -59 Cracow,

More information

Subcellular Localisation of Proteins in Living Cells Using a Genetic Algorithm and an Incremental Neural Network

Subcellular Localisation of Proteins in Living Cells Using a Genetic Algorithm and an Incremental Neural Network Subcellular Localisation of Proteins in Living Cells Using a Genetic Algorithm and an Incremental Neural Network Marko Tscherepanow and Franz Kummert Applied Computer Science, Faculty of Technology, Bielefeld

More information

22c145-Fall 01: Neural Networks. Neural Networks. Readings: Chapter 19 of Russell & Norvig. Cesare Tinelli 1

22c145-Fall 01: Neural Networks. Neural Networks. Readings: Chapter 19 of Russell & Norvig. Cesare Tinelli 1 Neural Networks Readings: Chapter 19 of Russell & Norvig. Cesare Tinelli 1 Brains as Computational Devices Brains advantages with respect to digital computers: Massively parallel Fault-tolerant Reliable

More information

Computational Complexity and Genetic Algorithms

Computational Complexity and Genetic Algorithms Computational Complexity and Genetic Algorithms BART RYLANDER JAMES FOSTER School of Engineering Department of Computer Science University of Portland University of Idaho Portland, Or 97203 Moscow, Idaho

More information