This directory is created automatically and some papers may be mislabeled. Only document within the CiteSeer database are listed. The directory is intended to provide entry points for browsing the database and is not intended to be authoritative. Papers may not appear in all relevant categories. For example, papers in a sub-category may not appear in higher level categories.
1142.8 A Tutorial on Support Vector Machines for Pattern Recognition - Burges (1998)(Correct)
The tutorial starts with an overview of the concepts of VC dimension and structural risk minimization. We then describe linear Support Vector Machines (SVMs) for separable and non-separable data, wo... / leave the best performing neural networks severely handicapped and br machine.Thus for example a neural network with fixed architecture with
1081.1 Complements to 'Pattern Recognition and Neural Networks' - Ripley (1996)(Correct)
Introduction Page 4: The book by Przytula & Prasanna (1993) discusses in detail the parallel implementation of neural networks. Page 16: Langley (1996) provides a book-length introduction to one viewp... / to Pattern Recognition and Neural Networks' by B.D. Ripley br the parallel implementation of neural networks. Page Langley
828.9 An information-maximization approach to blind separation and blind.. - Bell, Sejnowski (1995)(Correct)
We derive a new self-organising learning algorithm which maximises
the information transferred in a network of non-linear units. The algorithm
does not assume any knowledge of the input distributions,... / learning rules for neural networks has been pioneered by Linsker br objective functions applied to neural networks with non-linear units. The
449.2 Neural Network-Based Face Detection - Rowley, Baluja, Kanade (1996)(Correct)
We present a neural network-based upright frontal face detection system. A retinally connected neural network examines small windows of an image, and decides whether each window contains a face. The s... / Rowley Baluja and Kanade Neural Network-Based Face Detection PAMI br PAMI January Neural Network-Based Face Detection Henry
449.2 A Measurement-based Admission Control Algorithm for Integrated.. - Jamin, Danzig, Shenker, Zhang (1995)(Correct)
Many designs for integrated services networks offer a
bounded delay packet delivery service to support real-time applications.
To provide bounded delay service, networks must use admission
control t... / references Hir CLG a neural network is used for dynamic bandwidth br Capacity Control by Distributed Neural Network IEEE Journal of Selected
445.7 An Evaluation of Statistical Approaches to Text Categorization - Yang (1998)(Correct)
This paper focuses on a comparative evaluation of a wide-range of text categorization
methods, including previously published results on the Reuters corpus
and new results of additional experiments... / observation kNN LLSF and a neural network method had the best br learning algorithms neural networks and on-line learning
365.9 Factorial Hidden Markov Models - Zoubin Ghahramani, Michael I. Jordan (1997)(Correct)
Hidden Markov models (HMMs) have proven to be one of the most widely used tools
for learning probabilistic models of time series data. In an HMM, information about the past
is conveyed through a sin... / mixture of Gaussians or even a neural network. In the present paper br X t as K separate neural networks one for each setting of S
359.9 GTM: The Generative Topographic Mapping - Bishop, al. (1998)(Correct)
Latent variable models represent the probability density of data in a space of several dimensions
in terms of a smaller number of latent, or hidden, variables. A familiar example is factor analysis
wh... / for example of a feed-forward neural network in which case W would br by EPSRC grant GR K Neural Networks for Visualization of
330.4 Solving Multiclass Learning Problems via Error-Correcting Output Codes - Dietterich, al. (1995)(Correct)
Multiclass learning problems involve finding a definition for an unknown function f(x)
whose range is a discrete set containing k ? 2 values (i.e., k "classes"). The definition is
acquired by studyin... / binary values. Most artificial neural network algorithms such as the br for class i. With artificial neural networks these n functions can be
302.1 Hierarchically classifying documents using very few words - Koller, Sahami (1997)(Correct)
The proliferation of topic hierarchies for text documents has resulted in a need for tools that automatically classify new documents within such hierarchies. Existing classification schemes which igno... / this difficulty by using neural network technology to automatically
294.8 Minimizing Conflicts: A Heuristic Repair Method for.. - Minton, Johnston, Philips, Laird (1992)(Correct)
This paper describes a simple heuristic approach to solving large-scale constraint satisfaction and
scheduling problems. In this approach one starts with an inconsistent assignment for a set of variab... / by a surprisingly effective neural network developed by Adorf and br review of Adorf and Johnston's neural network and then describe our
291.4 Sparse coding with an overcomplete basis set: A strategy employed by.. - Olshausen, Field (1998)(Correct)
The spatial receptive fields of simple cells in mammalian striate cortex have
been reasonably well described physiologically and can be characterized as being
localized, oriented, and bandpass, compa... / by a strictly feedforward neural network in which case the functions br as shown in Figure b. A neural network implementation of this
285.7 Natural Gradient Works Efficiently in Learning - Amari (1998)(Correct)
When a parameter space has a certain underlying structure, the ordinary gradient of
a function does not represent its steepest direction but the natural gradient does. Information
geometry is used for... / that parameter spaces of neural networks have the Riemannian br by a processor like a neural network which has a set of adjustable
255.3 The sample complexity of pattern classification with neural networks: .. - Bartlett (1997)(Correct)
Sample complexity results from computational learning theory, when applied to
neural network learning for pattern classification problems, suggest that for good generalization
performance the number o... / of pattern classification with neural networks the size of the weights is br theory when applied to neural network learning for pattern
239.9 Feature Subset Selection Using A Genetic Algorithm - Yang, Honavar (1998)(Correct)
Practical pattern classification and knowledge discovery problems require selection of a subset of attributes or features (from a much larger set) to represent the patterns to be classified. This pape... / in the automated design of neural networks for pattern classification br tree induction algorithm or a neural network learning algorithm The
217.3 Face Recognition by Elastic Bunch Graph Matching - Wiskott, Fellous, Krüger, von der.. (1996)(Correct)
We present a system for recognizing human faces from single images out of a large database with one image per person. The task is difficult because of image variance in terms of position, size, expres... / process whereas Neural Network models tend to stress br the images. In contrast to many neural network systems no extensive
202.0 When Networks Disagree: Ensemble Methods for Hybrid Neural Networks - Perrone, Cooper (1993)(Correct)
This paper presents a general theoretical framework for ensemble methods of constructing significantly improved regression estimates. Given a population of regression estimators, we construct a hybrid... / Ensemble Methods for Hybrid Neural Networks Michael P. Perrone and br estimates whereas other neural network algorithms are hindered by
195.2 Recursive Distributed Representations - Pollack (1990)(Correct)
A long-standing difficulty for connectionist modeling has been how to represent
variable-sized recursive data structures, such as trees and lists, in fixed-width patterns.
This paper presents a connec... / machinery provided by neural networks. J. B. Pollack . br for example work on neural network and other learning machines
194.2 Evolution of Homing Navigation in a Real Mobile Robot - Floreano, Mondada (1996)(Correct)
In this paper we describe the evolution of a discrete-time recurrent neural network to control a real mobile robot. In all our experiments the evolutionary procedure is carried out entirely on the phy... / of a discrete-time recurrent neural network to control a real mobile br Robots Genetic Algorithms Neural Networks. I. Introduction A
191.3 Error-Correcting Output Coding Corrects Bias and Variance - Kong, Dietterich (1995)(Correct)
Previous research has shown that a technique
called error-correcting output coding
(ECOC) can dramatically improve the
classification accuracy of supervised learning
algorithms that learn to classify ... / accuracy of decision tree and neural network classifiers when compared br of different random seeds in neural network learning or different training
182.8 Map Learning and High-Speed Navigation in RHINO - Thrun, Bücken, Burgard, Fox.. (1998)(Correct)
This chapter surveys basic methods for learning maps and high speed autonomous navigation for indoor
mobile robots. The methods have been developed in our lab over the past few years, and most of them... / algorithms such as artificial neural networks interpret the robot's br and the cameras. Artificial neural networks are used to interpret sonar
182.6 The wake-sleep algorithm for unsupervised neural networks - Hinton, Dayan, Frey, Neal (1995)(Correct)
An unsupervised learning algorithm for a multilayer network of stochastic
neurons is described. Bottom-up "recognition" connections convert the input
into representations in successive hidden layers a... / algorithm for unsupervised neural networks Geoffrey E Hinton br algorithms for multilayer neural networks face two problems They
162.3 A Guide to the Literature on Learning Probabilistic Networks From Data - Buntine (1996)(Correct)
This literature review discusses different methods
under the general rubric of learning Bayesian networks
from data, and includes some overlapping work on more general
probabilistic networks. Connecti... / drawn between the statistical neural network and uncertainty communities br and to a lesser degree in neural networks where biological views offer
159.9 Rotation Invariant Neural Network-Based Face Detection - Rowley, Baluja, Kanade (1998)(Correct)
In this paper, we present a neural network-based face detection system. Unlike similar systems which are limited to detecting upright, frontal faces, this system detects faces at any degree of rotatio... / Rotation Invariant Neural Network-Based Face Detection br In this paper we present a neural network-based face detection system.
156.5 Active Learning with Statistical Models - Cohn, Ghahramani, Jordan (1996)(Correct)
For many types of machine learning algorithms, one can compute the statistically "optimal
" way to select training data. In this paper, we review how optimal data selection
techniques have been used w... / have been used with feedforward neural networks. We then show how the same br While the techniques for neural networks are computationally expensive
156.5 Empirical Support for Winnow and Weighted-Majority Algorithms.. - Blum (1995)(Correct)
This paper describes experimental results on using Winnow and Weighted-Majority based algorithms on a real-world calendar scheduling domain. These two algorithms have been highly studied in the theore... / decision tree variants and a neural-network algorithm and found these to
153.6 Gradient Calculations for Dynamic Recurrent Neural Networks: A Survey - Pearlmutter (1995)(Correct)
We survey learning algorithms for recurrent
neural networks with hidden units, and put the various techniques
into a common framework. We discuss fixedpoint
learning algorithms, namely recurrent backp... / for Ieee Transactions On Neural Networks Gradient Calculations For br for Dynamic Recurrent Neural Networks A Survey Barak A.
150.7 Challenges in Evolving Controllers for Physical Robots - Mataric, Cliff (1996)(Correct)
This paper discusses the feasibility of applying evolutionary methods
to automatically generating controllers for physical mobile robots.
We overview the state of the art in the field, describe some o... / continuous-time recurrent neural network controllers for artificial br evolving parameters for a neural network governing leg movements in a
150.6 Neural Network Exploration Using Optimal Experiment Design - Cohn (1994)(Correct)
We consider the question "How should one act when the only goal is to learn as much as possible?" Building
on the theoretical results of Fedorov [1972] and MacKay [1992], we apply techniques from Opti... / C.B.C.L. Paper No. Neural Network Exploration Using Optimal br the query action selection of a neural network learner. We demonstrate that
148.9 Face Recognition: A Convolutional Neural Network Approach - Lawrence, Giles, Tsoi, Back (1997)(Correct)
Faces represent complex, multidimensional, meaningful visual stimuli and developing a computational model for face recognition is difficult [43]. We present a hybrid neural network solution which comp... / IEEE Transactions on Neural Networks Special Issue on Neural br Networks Special Issue on Neural Networks and Pattern Recognition
145.6 Knowledge-Based Artificial Neural Networks - Geoffrey Towell (1994)(Correct)
Hybrid learning methods use theoretical knowledge of a domain and a set of classified
examples to develop a method for accurately classifying examples not seen during
training. The challenge of hybrid... / Knowledge-Based Artificial Neural Networks Geoffrey G. Towell br Knowledge-Based Artificial Neural Networks Current address is
144.6 Monte Carlo Implementation of Gaussian Process Models for Bayesian.. - Neal (1997)(Correct)
Gaussian processes are a natural way of defining prior distributions over functions
of one or more input variables. In a simple nonparametric regression problem, where
such a function gives the mean... / regression models based on neural networks converge to Gaussian br applications to which neural networks are typically applied
140.4 Independent Component Analysis Using an Extended Infomax Algorithm.. - Lee (1997)(Correct)
An extension of the infomax algorithm of Bell and Sejnowski (1995) is presented that is able to blindly separate mixed signals with sub- and super-Gaussian source distributions. This was achieved by u... / studied by many researchers in neural networks and statistical signal br in a single-layer feedforward neural network. The algorithm is effective in
137.1 Learning Maps for Indoor Mobile Robot Navigation - Thrun, Bücken (1998)(Correct)
Autonomous robots must be able to learn and maintain models of their environments.
Research on mobile robot navigation has produced two major paradigms
for mapping indoor environments: grid-based and ... / exploration mobile robots neural networks occupancy grids path br are learned using artificial neural networks and Bayesian integration.
136.2 Similarity Metric Learning for a Variable-Kernel Classifier - Lowe (1995)(Correct)
Nearest-neighbour interpolation algorithms have many useful properties
for applications to learning, but they often exhibit poor generalization.
In this paper, it is shown that much better generalizat... / growing interest in the neural network community. In part this is br missing from the most popular neural network learning methods yet they are
136.0 Learning and development in neural networks: The importance of.. - Elman (1993)(Correct)
This document was created with FrameMaker 4.0.4
Elman Learning and development in neural networks
-2- unknown Learning and development in neural networks:
The importance of starting small
Jeffrey L... / Learning and development in neural networks The importance of starting br are obtained with artificial neural network models of learning. There are
133.3 Generating Accurate and Diverse Members of a Neural-Network Ensemble - Opitz, al. (1996)(Correct)
Neural-network ensembles have been shown to be very accurate
classification techniques. Previous work has shown that an effective
ensemble should consist of networks that are not only highly
correct, ... / and Diverse Members of a Neural-Network Ensemble David W. Opitz br Abstract Neural-network ensembles have been shown to
131.9 On Comparing Classifiers: Pitfalls to Avoid and a Recommended Approach - Salzberg (1997)(Correct)
An important component of many data mining projects is finding a good classification algorithm,
a process that requires very careful thought about experimental design. If not done very carefully, co... / experimental papers on neural network learning algorithms and found br by Flexer of experimental neural network papers only out of
130.8 Automatic Creation of an Autonomous Agent: Genetic Evolution of a.. - Floreano, Mondada (1994)(Correct)
The paper describes the results of the evolutionary development of a real, neural-network driven mobile robot. The evolutionary approach to the development of neural controllers for autonomous agents ... / Agent Genetic Evolution of a Neural-Network Driven Robot Dario Floreano br development of a real neural-network driven mobile robot. The
125.9 GENET: A Connectionist Architecture for Solving Constraint.. - Davenport, Tsang, Wang, Zhu (1994)(Correct)
New approaches to solving constraint satisfaction problems using iterative improvement techniques have been found to be successful on certain, very large problems such as the million queens. However, ... / paper we present genet a neural-network architecture for solving br Architecture The genet neural network architecture is similiar to
121.6 Face Recognition Under Varying Pose - Beymer (1993)(Correct)
Researchers in computer vision and pattern recognition have worked on automatic techniques for recognizing
human faces for the last 20 years. While some systems, especially template-based ones, have b... / of hidden layer nodes in neural networks Vincent Waite and br typical matching metric. The neural network approaches construct a
120.9 Analog Computation Via Neural Networks - Siegelmann, Sontag (1994)(Correct)
We pursue a particular approach to analog computation, based on dynamical systems of the
type used in neural networks research.
Our systems have a fixed structure, invariant in time, corresponding to ... / Analog Computation Via Neural Networks Hava T. Siegelmann br systems of the type used in neural networks research. Our systems have a
119.9 A Model of Saliency-based Visual Attention for Rapid Scene Analysis - Itti (1998)(Correct)
A visual attention system, inspired by the behavior and the neuronal architecture of the early primate visual system, is presented. Multiscale image features are combined into a single topographical s... / saliency map. A dynamical neural network then selects attended br map modeled as a dynamical neural network. One difficulty in combining
119.9 Support Vector Machine Reference Manual - Saunders, Stitson, Weston, Bottou.. (1998)(Correct)
this document will describe these programs. To find out more about SVMs, see the bibliography. We will not describe how SVMs work here. The first program we will describe is the paragen program, as it... / Basis Function Two Layer Neural Network paragen Infinite br Vap Pi . Two layer neural network tanh b x Delta y
119.9 Support Vector Machine - Reference Manual - Saunders, Stitson, Weston, Bottou.. (1998)(Correct)
this document will describe these programs. To find out more about SVMs, see the bibliography. We will not describe how SVMs work here. The first program we will describe is the paragen program, as it... / Basis Function Two Layer Neural Network paragen Infinite br Vap Pi . Two layer neural network tanh b x Delta y
118.5 Neural Network Synthesis Using Cellular Encoding And The Genetic.. - Frédéric Gruau (1994)(Correct)
Artificial neural networks used to be considered only as a machine that learns using small
modifications of internal parameters. Now this is changing. Such learning method do not allow
to generate big... / Specialty Computer Science Neural Network Synthesis Using Cellular br to be interested in Genetic Neural Networks. I believe that this subject
116.0 Non linear neurons in the low noise limit: a factorial code maximizes .. - Nadal (1994)(Correct)
We investigate the consequences of maximizing information transfer in a simple neural network (one input layer, one output layer), focussing on the case of non linear transfer functions. We assume tha... / transfer in a simple neural network one input layer one output br a nonlinear neural algorithm. Neural Networks - .
115.9 Combining Simulated Annealing with Local Search Heuristics - Martin, Otto (1996)(Correct)
We introduce a meta-heuristic to combine simulated annealing with local search methods
for CO problems. This new class of Markov chains leads to significantly more powerful
optimization methods than e... / genetic algorithms and neural network approaches The present br In J. Denker editor Neural Networks for Computing . AIP
115.9 An Experimental and Theoretical Comparison of Model Selection Methods - Kearns, Mansour, Ng, Ron (1995)(Correct)
this paper is to provide such a comparison, and more importantly, to describe the general conclusions to which it has led. Relying on evidence that is divided between controlled experimental results a... / number of hidden nodes in a neural network determining the right amount br to choose for instance a neural network with a number of nodes
106.3 Networks of Spiking Neurons: The Third Generation of Neural Network.. - Maass (1997)(Correct)
The computational power of formal models for networks of spiking neurons
is compared with that of other neural network models based on McCulloch
Pitts neurons (i.e. threshold gates), respectively sigm... / The Third Generation of Neural Network Models Wolfgang Maass br is compared with that of other neural network models based on McCulloch
101.4 Discriminant Analysis by Gaussian Mixtures - Hastie, Tibshirani (1996)(Correct)
Fisher-Rao linear discriminant analysis (LDA) is a valuable tool for
multigroup classification. LDA is equivalent to maximum likelihood classification
assuming Gaussian distributions for each class. I... / and classification trees. Neural network classifiers have become a br pursuit regression and neural networks. They call the resulting
101.4 Exploiting Tractable Substructures in Intractable Networks - Saul, Jordan (1995)(Correct)
We develop a refined mean field approximation for inference and
learning in probabilistic neural networks. Our mean field theory,
unlike most, does not assume that the units behave as independent
degr... / and learning in probabilistic neural networks. Our mean field theory br parameters in a probabilistic neural network may be viewed as a problem in
101.2 Population Based Incremental Learning: A Method for Integrating.. - Baluja (1994)(Correct)
Genetic algorithms (GAs) are biologically motivated adaptive systems which have been used,
with varying degrees of success, for function optimization. In this study, an abstraction of the
basic geneti... / algorithms and artificial neural networks are two recently developed br effort in the field of neural networks and competitive learning has
101.0 Combining estimates in regression and classification - LeBlanc, Tibshirani (1993)(Correct)
We consider the problem of how to combine a collection of general
regression fit vectors in order to obtain a better predictive model. The
individual fits may be from subset linear regression, ridge r... / something more complex like a neural network. We develop a general br regression or neural network model. Or the collection of c
98.7 Packet Routing in Dynamically Changing Networks: A Reinforcement.. - Boyan, Littman (1994)(Correct)
This paper describes the Q-routing algorithm for packet routing,
in which a reinforcement learning module is embedded into each
node of a switching network. Only local communication is used
by each no... / approximating Q x with a neural network as in e.g. which br . J. Boyan. Modular neural networks for learning
97.1 Incorporating Information From Syllable-length Time Scales Into.. - Wu (1998)(Correct)
Incorporating the concept of the syllable into speech recognition may improve
recognition accuracy through the integration of information over syllable-length time
spans. Evidence from psychoacoustics... / . . Probability Estimation Neural Network br from an onset- detection neural network at frames covered by
96.9 Competitive Environments Evolve Better Solutions for Complex Tasks - Angeline, Pollack (1993)(Correct)
In the typical genetic algorithm experiment, the fitness function is constructed to be independent of the contents of the population to provide a consistent objective measure. Such objectivity entails... / Tesauro describes a neural network that learns to play br G. Neurogammon a neural network backgammon program.IJCNN
93.8 Mixtures of Controllers for Jump Linear and Non-linear Plants - Cacciatore (1994)(Correct)
We describe an extension to the Mixture of Experts architecture for
modelling and controlling dynamical systems which exhibit multiple
modes of behavior. This extension is based on a Markov process
mo... / the piecewise solutions in a neural network context. This architecture has br Sutton R.S. and Werbos P.J. Neural Networks for Control MIT Press
92.7 On Convergence Properties of the EM Algorithm for Gaussian Mixtures - Xu, Jordan (1996)(Correct)
We build up the mathematical connection between the "Expectation-Maximization" (EM) algorithm and gradient-based approaches for maximum likelihood learning of finite Gaussian mixtures. We show that th... / The recent emphasis in the neural network literature on probabilistic br be the methods of choice in the neural network literature. However we also
91.4 Estimating Dependency Structure as a Hidden Variable - Meila, Jordan, Morris (1998)(Correct)
This paper introduces a probability model, the mixture of trees that can account for sparse, dynamically changing dependence relationships. We present a family of efficient algorithms based on the EM ... / used in the statistics and neural network literature. Of relevance to br and who used multilayer neural networks and knowledge-based neural
91.4 Assessing Relevance Determination Methods Using DELVE - Neal (1998)(Correct)
Empirically assessing the predictive performance of learning methods is
an essential component of research in machine learning. The DELVE environment
was developed to support such assessments. It pr... / To appear in Generalization in Neural Networks and Machine Learning C. M. br to assess the performance of neural network methods when the inputs
89.8 Efficient Approximations for the Marginal Likelihood of Bayesian.. - Chickering, Heckerman (1996)(Correct)
We discuss Bayesian methods for model averaging and model selection among Bayesiannetwork
models with hidden variables. In particular, we examine large-sample approximations
for the marginal likelihoo... / learning for probabilistic neural-network models. In this paper we br respectively in feed-forward neural networks. Roughly speaking in using
89.7 Theory Refinement on Bayesian Networks - Buntine (1991)(Correct)
Theory refinement is the task of updating a domain theory in the light of new cases, to be done automatically or with some expert assistance. The problem of theory refinement under uncertainty is revi... / domain is transcribed into a neural network to initialize the network br to some feed-forward neural networks. Information and Control.
89.6 A Practical Bayesian Framework for Backprop Networks - MacKay (1992)(Correct)
A quantitative and practical Bayesian framework is described for learning of mappings in feedforward networks. The framework makes possible: (1) objective comparisons between solutions using alternati... / running over the pairs. A neural network architecture A is invented br such as the choice of neural network architecture and of the
88.8 A General Result on the Stabilization of Linear Systems Using Bounded .. - Sussmann, Sontag, Yang (1994)(Correct)
We present two constructions of controllers that globally stabilize linear systems subject to control saturation. We allow essentially arbitrary saturation functions. The only conditions imposed on th... / constructions is in terms of a neural-network type one-hidden layer br In the language of neural networks one wants control laws that
88.6 Keeping Neural Networks Simple by Minimizing the Description Length.. - Hinton (1993)(Correct)
Supervised neural networks generalize well if
there is much less information in the weights
than there is in the output vectors of the training
cases. So during learning, it is important
to keep the w... / Keeping Neural Networks Simple by Minimizing the br Canada Abstract Supervised neural networks generalize well if there is
86.9 Evolving Mobile Robots in Simulated and Real Environments - Miglino, Lund, Nolfi (1996)(Correct)
The problem of the validity of simulation is particularly relevant for
methodologies that use machine learning techniques to develop control
systems for autonomous robots, like, for instance, the Arti... / controlled by an artificial neural network that should explore an open br experiment in which dynamical neural networks were evolved in simulation to
84.5 A Minimum Description Length Framework for Unsupervised Learning - Zemel (1993)(Correct)
A fundamental problem in learning and reasoning about a set of information is finding the right
representation. The primary goal of an unsupervised learning procedure is to optimize the quality
of a s... / for training self-supervised neural networks where the goal is to br . A Neural network approach
84.4 Improving Generalization with Active Learning - Cohn, Atlas, al. (1992)(Correct)
Active learning differs from passive "learning from examples" in that the learning algorithm assumes
at least some control over what part of the input domain it receives information about. In some sit... / approximately implemented by a neural network. In selective sampling a br vs. Active Learning Most neural network generalization problems are
84.0 Extraction of Rules from Discrete-Time Recurrent Neural Networks - Omlin, Giles (1996)(Correct)
The extraction of symbolic knowledge from trained neural networks and the direct encoding of (partial) knowledge into networks prior to training are important issues. They allow the exchange of inform... / from Discrete-Time Recurrent Neural Networks Christian W. Omlin C. br symbolic knowledge from trained neural networks and the direct encoding of
83.9 Coevolving High-Level Representations - Angeline (1994)(Correct)
Several evolutionary simulations allow for a dynamic resizing of the
genotype. This is an important alternative to constraining the genotype's
maximum size and complexity. In this paper, we add an add... / state automata FSA or a neural network as the phenotype. The content br previous work to evolve modular neural networks for the control of artificial
81.0 A Simple Weight Decay Can Improve Generalization - Krogh (1992)(Correct)
It has been observed in numerical simulations that a weight decay can improve
generalization in a feed-forward neural network. This paper explains
why. It is proven that a weight decay has two effects... / in a feed-forward neural network. This paper explains why. It br the generalization ability of a neural network or any other learning
80.8 Connectionist Theory Refinement: Genetically Searching the Space of.. - Opitz, al. (1997)(Correct)
An algorithm that learns from a set of examples should ideally be able to exploit the
available resources of (a) abundant computing power and (b) domain-specific knowledge to
improve its ability to ge... / knowledge to select a neural network's topology and initial br to refine the topology of the neural networks they produce thereby
80.4 Convergence results for the EM approach to mixtures of experts.. - Jordan, Xu (1993)(Correct)
The Expectation-Maximization (EM) algorithm is an iterative approach to maximum likelihood parameter estimation. Jordan and Jacobs (1993) recently proposed an EM algorithm for the mixture of experts a... / Introduction Although neural networks are capable in principle of br the language of fully-connected neural networks. Achieving better scaling
79.9 Relational Learning Techniques for Natural Language Information.. - Califf (1998)(Correct)
The recent growth of online information available in the form of natural language documents creates a greater need for computing systems with the ability to process those documents to simplify access ... / significant research applying neural-network methods to language processing
79.9 Is Machine Colour Constancy Good Enough? - Brian Funt (1998)(Correct)
This paper presents a negative result: current machine colour
constancy algorithms are not good enough for colour-based object
recognition. This result has surprised us since we have previously used... / D gamut-constraint and neural network. These algorithms all either br centroid in the D case. Neural Network Colour Constancy Previously
78.2 Comparative Experiments on Disambiguating Word Senses: An.. - Mooney (1996)(Correct)
This paper describes an experimental comparison
of seven different learning algorithms on the
problem of learning to disambiguate the meaning
of a word from context. The algorithms tested
include stat... / tested include statistical neural-network decision-tree rule-based br as context. The statistical and neural-network methods perform the best on
76.5 Skin-Color Modeling and Adaptation - Yang, Lu, Waibel (1997)(Correct)
This report studies a statistical skin-color model and its adaptation. By quantitative analysis
and goodness-of-fit test, we reveal that (1) skin-color differences among people can be
reduced by inten... / image invariants and neural networks These methods are br a x image by using a neural network to detect faces. Color is
76.5 Fast Exact Multiplication by the Hessian - Pearlmutter (1994)(Correct)
Just storing the Hessian H (the matrix of second derivatives @
2
E=@w i @w j of the error E
with respect to each pair of weights) of a large neural network is difficult. Since a common
use of a lar... / pair of weights of a large neural network is difficult. Since a common br information from large neural networks is an important problem
75.3 Evolutionary Algorithms for Neural Network Design and Training - Branke (1995)(Correct)
Neural networks and genetic algorithms are two relatively young research areas
that were subject to a steadily growing interest during the past years. Both models
are inspired by nature, but whereas n... / Evolutionary Algorithms for Neural Network Design and Training Jurgen br January Abstract Neural networks and genetic algorithms are
72.7 The Thisl Broadcast News Retrieval System - Dave Abberley (1999)(Correct)
This paper described the THISL spoken document retrieval
system for British and North American Broadcast
News. The system is based on the ABBOT large
vocabulary speech recognizer, using a recurrent ne... / estimates produced by the neural network acoustic model by pruning all
72.7 Search in a Small World - Walsh (1999)(Correct)
In a graph with a "small world" topology, nodes are highly clustered yet the path length between them is small. Such a topology can make search problems very difficult since local decisions quickly pr... / a biological graph the neural network of the nematode worm C.
72.4 Holographic Reduced Representations - Plate (1995)(Correct)
Associative memories are conventionally used to represent data with very simple structure: sets of pairs of vectors. This paper describes a method for representing more complex compositional structure... / Most work on neural-network style associative memories br IEEE Transactions on Neural Networks - complex
72.3 Blind Source Separation of Real World Signals - Lee, Bell (1997)(Correct)
We present a method to separate and deconvolve sources
which have been recorded in real environments. The use of
noncausal FIR filters allows us to deal with nonminimum
mixing systems. The learning ru... / by researchers in the field of neural networks and br have proposed a simple infomax neural network algorithm where they maximize
70.6 Biological Metaphors and the Design of Modular Artificial Neural.. - Boers, Kuiper (1992)(Correct)
In this thesis, a method is proposed with which good modular artificial neural network structures can be found automatically using a computer program. A number of biological metaphors are incorporated... / design of modular artificial neural networks Master's thesis of Egbert br and Function of Modular Neural Networks'done at the department of
69.5 Sequencing Run-Time Reconfigured Hardware with Software - Wirthlin (1996)(Correct)
Run-Time Reconfigured systems offer additional
hardware resources to systems based on reconfigurable
FPGAs. These systems, however, are often difficult
to build and must tolerate substantial reconfigu... / FPGAs Using RTR two neural network systems were developed to br Density enhancement of a neural network using FPGAs and run-time
69.5 Evolutionary Design of Neural Architectures - A Preliminary Taxonomy.. - Balakrishnan, Honavar (1995)(Correct)
This report briefly motivates current research on evolutionary design of neural
architectures (EDNA) and presents a short overview of major research issues in
this area. It also includes a preliminary... / Introduction Artificial neural networks Rumelhart McClelland br superficial to biological neural networks brains Despite much
69.5 High-Performance Job-Shop Scheduling With A Time-Delay TD(lambda).. - Zhang, Dietterich (1995)(Correct)
Job-shop scheduling is an important task for manufacturing industries.
We are interested in the particular task of scheduling payload
processing for NASA's space shuttle program. This paper summarizes... / how to extend the time-delay neural network TDNN architecture to apply br The tests also show that both neural network approaches significantly
69.1 Prediction Risk and Architecture Selection for Neural Networks - Moody (1994)(Correct)
We describe two important sets of tools for neural network modeling:
prediction risk estimation and network architecture selection. Prediction risk is defined
as the expected performance of an estim... / in From Statistics to Neural Networks Theory and Pattern br and Architecture Selection for Neural Networks John Moody Oregon
69.1 Toward an Evolvable Model of Development for Autonomous Agent.. - Dellaert, Beer (1994)(Correct)
We are interested in the synthesis of autonomous agents using evolutionary techniques. Most work in this area utilizes a direct mapping from genotypic space to phenotypic space. In order to address so... / coding for evolving neural networks Kitano and Gruau and br encoding to develop artificial neural networks. Harp Samad and Guha
68.5 Making Use of Population Information in Evolutionary Artificial.. - Yao, Liu (1998)(Correct)
This paper is concerned with the simultaneous evolution
of artificial neural network (ANN) architectures and
weights. The current practice in evolving ANNs is to choose
the best ANN in the last genera... / in Evolutionary Artificial Neural Networks Xin Yao SMIEEE and Yong br evolution of artificial neural network ANN architectures and
68.5 Using Decision Tree Confidence Factors for Multiagent Control - Stone, Veloso (1998)(Correct)
Although Decision Trees are widely used for classification tasks, they are typically not used for agent control. This paper presents a novel technique for agent control in a complex multiagent domain ... / an example of such a domain a Neural Network NN was used to learn a br Soccer Server clients used a Neural Network NN to learn a low-level
68.5 Probabilistic Planning in the Graphplan Framework - Blum, Langford (1998)(Correct)
The Graphplan planner compiles a STRIPS planning
problem into a compact graph structure in which information
can be easily stored and propagated to aid
in the search for a plan. Empirical results have... / one might try to train a neural network to map states to values or to
68.5 PUBLIC: A Decision Tree Classifier that Integrates Building and.. - Rastogi (1998)(Correct)
Classification is an important problem in
data mining. Given a database of records,
each with a class label, a classifier generates
a concise and meaningful description
for each class that can be used... / classification CKS neural networks Rip genetic br for this. First compared to a neural network or a bayesian classifier a
68.0 Multitask Learning - Caruana (1997)(Correct)
Multitask Learning is an approach to inductive transfer that improves generalization by using the
domain information contained in the training signals of related tasks as an inductive bias. It does ... / the same domain. An artificial neural network or a decision tree or a . br Y. S.Learning from Hints in Neural Networks Journal of Complexity
68.0 Stability Analysis Of Adaptive Blind Source Separation - Amari, Chen, Cichocki (1997)(Correct)
Recently a number of adaptive learning algorithms have been proposed for blind source
separation. Although the underlying principles and approaches are different, most of them
have very similar forms.... / by a feed-forward linear neural network with connection weight matrix br by a feedback fully recurrent neural network described as y t
66.6 Integrating Grid-Based and Topological Maps for Mobile Robot.. - Thrun, Bücken (1996)(Correct)
Research on mobile robot navigation has produced two major
paradigms for mapping indoor environments: grid-based
and topological. While grid-based methods produce accurate
metric maps, their complexit... / are learned using artificial neural networks and Bayesian integration. br interpreted by an artificial neural network and mapped into probabilities
66.6 Blind Separation Of Convolved Sources Based On Information.. - Torkkola (1996)(Correct)
Blind separation of independent sources from their convolutive mixtures is a problem in many real world multi-sensor applications. In this paper we present a solution to this problem based on the info... / IEEE workshop on Neural Networks for Signal Processing Kyoto br some nonlinear functions neural network-like algorithms
64.6 An Efficient Gradient-Based Algorithm for On-Line Training of.. - Williams, Peng (1990)(Correct)
A novel variant of a familiar recurrent network learning algorithm is described. This algorithm is capable of shaping the behavior of an arbitrary recurrent network as it runs, and it is specifically ... / Introduction Artificial neural networks having feedback connections br number of investigators in the neural network field and their origins can
63.8 Incremental Evolution of Complex General Behavior - Gomez (1997)(Correct)
Several researchers have demonstrated how complex action sequences can be learned through neuro-evolution
(i.e. evolving neural networks with genetic algorithms). However, complex general behavior suc... / neuro-evolution i.e. evolving neural networks with genetic algorithms br the training of Artificial Neural Networks with Genetic Algorithms has
63.7 Neural Network Ensembles, Cross Validation, and Active Learning - Krogh, al. (1995)(Correct)
Learning of continuous valued functions using neural network ensembles
(committees) can give improved accuracy, reliable estimation
of the generalization error, and active learning. The ambiguity
is ... / Press Cambridge MA . Neural Network Ensembles Cross Validation br valued functions using neural network ensembles committees can
62.8 Natural Language Grammatical Inference with Recurrent Neural Networks - Lawrence, Giles, Fong (1998)(Correct)
This paper examines the inductive inference of a complex grammar with neural networks -- specifically, the task considered is that of training a network to classify natural language sentences as gramm... / Inference with Recurrent Neural Networks Steve Lawrence C. Lee br of a complex grammar with neural networks -specifically the task
62.8 Constructive Neural Network Learning Algorithms for Pattern.. - Parekh, Yang, Honavar (1998)(Correct)
Constructive learning algorithms offer an attractive approach for the incremental construction of near-minimal neural network architectures for pattern classification. They help overcome the need for ... / Constructive Neural Network Learning Algorithms for br construction of near-minimal neural network architectures for pattern
62.8 Support Vector Machines for Classification and Regression - Gunn (1998)(Correct)
this
report the term SVM will refer to both classification and regression methods, and the
terms Support Vector Classification (SVC) and Support Vector Regression (SVR)
will be used for specification.... / of Hadamard Traditional neural network approaches have suffered br employed by conventional neural networks. SRM minimises an upper
62.8 Analog Neural Nets with Gaussian or other Common Noise Distributions.. - Maass, Sontag (1998)(Correct)
We consider recurrent analog neural nets where the output of each gate is subject
to Gaussian noise, or any other common noise distribution that is nonzero on
a sufficiently large part of the state sp... / the special case of noisy neural networks. Before we can give the br of computations in noisy neural networks. From the conceptual point
60.8 Selecting Input Variables Using Mutual Information and Nonparametric.. - Bonnlander, Weigend (1996)(Correct)
In learning problems where a connectionist network is trained with a finite sized training
set, better generalization performance is often obtained when unneeded weights in the network
are eliminated.... / density estimation neural networks do not use a kernel for each br the parameters in a neural network are more flexible. The
60.8 Comparison Of Learning Algorithms For Handwritten Digit Recognition - LeCun, Jackel, Bottou, Brunot.. (1995)(Correct)
This paper compares the performance of several classifier algorithms
on a standard database of handwritten digits. We consider not only raw
accuracy, but also rejection, training time, recognition tim... / Fully Connected Multi-Layer Neural Network Another classifier that we br a fully connected multi-layer neural network with two layers of weights
60.2 The principal components of natural images - Hancock, Baddeley, Smith (1991)(Correct)
A neural net was used to analyse samples of natural images and text. For the natural images, components resemble derivatives of Gaussian operators, similar to those found in visual cortex and inferred... / computation. We are using a neural network technique developed by Sanger br International Conference on Neural Networks San Diego IEEE New
59.5 ADIC: An Extensible Automatic Differentiation Tool for ANSI-C - Bischof, Roh, Mauer (1997)(Correct)
In scientific computing, we often require the derivatives @f=@x of a function f expressed
as a program with respect to some input parameter(s)x, say. Automatic differentiation (AD) techniques
augmen... / vehicle simulator and neural network code. Key words. Automatic br motion control simulator and a neural network model. These applications
59.5 Boosting And Naive Bayesian Learning - Elkan (1997)(Correct)
Although so-called "naive" Bayesian classification makes the unrealistic assumption that the values of the attributes of an example are independent given the class of the example, this learning method... / equivalent to a feedforward neural network with sparse encoding of br algorithm for feedforward neural networks. The argument can be
57.9 Learning long-term dependencies in NARX recurrent neural networks - Lin, Horne, Tino, Giles (1996)(Correct)
It has recently been shown that gradient-descent learning algorithms for recurrent neural
networks can perform poorly on tasks that involve long--term dependencies, i.e. those problems
for which the d... / dependencies in NARX recurrent neural networks Tsungnan Lin br algorithms for recurrent neural networks can perform poorly on tasks
57.9 Improved Gaussian Mixture Density Estimates Using Bayesian Penalty.. - Ormoneit, Tresp (1995)(Correct)
We compare two regularization methods which can be used to improve the generalization
capabilities of Gaussian mixture density estimates. The first method consists
of defining a Bayesian prior distrib... / major attention in the neural network community. Important examples br may be implemented as a fast neural network learning rule Now
57.7 Flexible Discriminant Analysis by Optimal Scoring - Hastie, Tibshirani, Buja (1993)(Correct)
Fisher's linear discriminant analysis is a valuable tool for multigroup
classification. With a large number of predictors, one can find
a reduced number of discriminant coordinate functions that are "... / technique such as MARS or neural networks can be post-processed to br multiple logistic regression. Neural network classifiers have become a
57.7 Ill-Conditioning In Neural Network Training Problems - Saarinen, Bramley, CYBENKO (1993)(Correct)
The training problem for feedforward neural networks is nonlinear parameter estimation
that can be solved by a variety of optimization techniques. Much of the literature on neural
networks has focus... / Ill-Conditioning In Neural Network Training Problems S. br problem for feedforward neural networks is nonlinear parameter
57.1 Learning to Perceive the World as Articulated: An Approach for.. - Tani, Nolfi (1998)(Correct)
This paper describes how agents can learn an internal model of the world
structurally by focusing on the problem of behavior-based articulation. We develop
an on-line learning scheme -- the so-called ... / version is in press. in Neural Networks Special Issue . br types of approach. One is the neural network learning approach. Krose and
57.1 Feedforward Neural Networks for Nonparametric Regression - David Rios Insua, Peter Müller (1998)(Correct)
Feed forward neural networks (FFNN) with an unconstrained random number
of hidden neurons define flexible non-parametric regression models. In
M¨uller and Rios Insua (1998) we have argued that variabl... / Opaque this Feedforward neural networks for nonparametric regression br Abstract Feed forward neural networks FFNN with an unconstrained
56.7 Neural Networks and Statistical Models - Sarle (1994)(Correct)
There has been much publicity about the ability of artificial neural
networks to learn and generalize. In fact, the most commonly
used artificial neural networks, called multilayer perceptrons, are
no... / Neural Networks and Statistical Models br the ability of artificial neural networks to learn and generalize. In
55.3 Achieving High-Accuracy Text-to-Speech with Machine Learning - Bakiri, Dietterich (1997)(Correct)
In 1987, Sejnowski and Rosenberg developed their famous NETtalk system for English textto
-speech. This chapter describes a machine learning approach to text-to-speech that builds
upon and extends the... / for training feed-forward neural networks Rumelhart Hinton br and Rosenberg trained a large neural network via the backpropagation
55.3 Variational Gaussian Process Classifiers - Gibbs, MacKay (1997)(Correct)
Gaussian processes are a promising non-linear interpolation tool (Williams 1995; Williams and Rasmussen 1996), but it is not straightforward to solve classification problems with them. In this paper t... / to IEEE Transactions on Neural Networks Abstract Gaussian br might be say the weights of a neural network or the coefficients of a
55.3 Gaussian Processes for Bayesian Classification via Hybrid Monte Carlo - Barber (1997)(Correct)
The full Bayesian method for applying neural networks to a prediction problem is to set up the prior/hyperprior structure for the net and then perform the necessary integrals. However, these integrals... / Bayesian method for applying neural networks to a prediction problem is br In the Bayesian approach to neural networks a prior on the weights in the
55.3 Efficient Implementation of Gaussian Processes - Gibbs, MacKay (1997)(Correct)
Neural networks and Bayesian inference provide a useful framework within which to solve regression
problems. However their parameterization means that the Bayesian analysis of neural networks can
be d... / May Abstract Neural networks and Bayesian inference br that the Bayesian analysis of neural networks can be difficult. In this
55.0 An Overview of Strategies for Neurosymbolic Integration - Hilario (1995)(Correct)
At the crossroads of symbolic and neural
processing, researchers have been actively
investigating the synergies that might be
obtained from combining the strengths of
these two paradigms. Neurosymboli... / representational models with neural networks. This papers attempts to br symbolic capabilities using neural networks alone while hybrid
54.5 The Self-Organizing Map in Industry Analysis - Simula, Vasara, Vesanto, Helminen (1999)(Correct)
The Self-Organizing Map (SOM) is a powerful neural network method for the analysis and visualization of high-dimensional data. It maps nonlinear statistical relationships between high-dimensional meas... / Map SOM is a powerful neural network method for the analysis and br is one of the most popular neural network models. The SOM implements a
52.9 Error-Correcting Output Codes: A General Method for Improving.. - Dietterich, Bakiri (1991)(Correct)
Multiclass learning problems involve finding a definition
for an unknown function f(x) whose range is a
discrete set containing k ? 2 values (i.e., k "classes").
The definition is acquired by studying... / binary values. Most artificial neural network algorithms such as the br for class i. With artificial neural networks these n functions can be
51.8 On Group Communication in Large-Scale Distributed Systems - Babaoglu, Schiper (1994)(Correct)
s are available from the same host in the directory /pub/TR/UBLCS/ABSTRACTS in plain text
format. All local authors can be reached via e-mail at the address last-name@cs.unibo.it.
UBLCS Technical Rep... / to Build GoodTraining Sets for Neural-Network Classifiers F. Tamburini R.
51.4 Automatic Text Detection and Tracking in Digital Video - Li, Doermann, Kia (1998)(Correct)
Text which either appears in a scene or is graphically added to video can provide an important
supplemental source of index information as well as clues for decoding the video's structure
and for clas... / Indexing Digital Libraries Neural Network Wavelet The support of br We use a hybrid wavelet neural network segmenter Figure to
51.4 Running Across the Reality Gap: Octopod Locomotion Evolved in a.. - Nick Jakobi (1998)(Correct)
This paper describes experiments in which neural network
control architectures were evolved in minimal simulation for an octopod
robot. The robot is around 30cm long and has 4 infra red sensors that... / describes experiments in which neural network control architectures were br the experiments was to evolve neural network control architectures that
51.0 Redundancy Reduction and Independent Component Analysis: Conditions.. - Nadal (1997)(Correct)
In the context of both sensory coding and signal processing, building factorized codes has been shown to be an efficient strategy. In a wide variety of situations, the signal to be processed is a line... / by using a simple feedforward neural network. In this paper we discuss br processing with a linear neural network model Atick A
49.2 Fast Learning VIEWNET Architectures for Recognizing 3-D Objects from.. - Bradski, Grossberg (1995)(Correct)
The recognition of 3-D objects from sequences of their 2-D views is modeled by a family of self-organizing neural architectures, called VIEWNET, that use View Information Encoded With NETworks. VIEWNE... / To appear in the Neural Networks special issue on Automatic br Words Pattern recognition neural networks ART ARTMAP -D object
49.2 An incremental approach to developing intelligent neural network.. - Meeden (1995)(Correct)
By beginning with simple reactive behaviors and gradually building up to more memorydependent
behaviors, it may be possible for connectionist systems to eventually achieve the
level of planning. This ... / to developing intelligent neural network controllers for robots Lisa br These methods are applied to a neural network controller for a simple
49.2 Generating Fuzzy Rules from Examples using Genetic Algorithms - Herrera, Lozano, VERDEGAY (1995)(Correct)
The problem of generation desirable fuzzy rules
is very important in the development of fuzzy
systems. The purpose of this paper is to
present a generation method of fuzzy control
rules by learning fr... / Descent Method or Neural Network or fuzzy br of fuzzy inference rules with a neural network. Fourth IFSA Congress
48.5 AntFarm: Towards Simulated Evolution - Robert J. Collins, David R. Jefferson (1991)(Correct)
The most easily observed ant behavior is workers foraging for food. Foraging
workers do not eat the food, but carry it back to the nest, where it is processed
and consumed by all members of the colony... / is specified by an artificial neural network ANN In addition to the br of pheromone to drop The whole neural network consists of neural units
48.2 Bayesian Methods for Adaptive Models - MacKay (1992)(Correct)
The Bayesian framework for model comparison and regularisation is demonstrated by studying
interpolation and classification problems modelled with both linear and non--linear models.
This framework qu... / data better. When applied to neural networks'the Bayesian framework br . What are neural networks and why do they need Occam's
47.9 Back Propagation is Sensitive to Initial Conditions - Kolen, Pollack (1990)(Correct)
This paper explores the effect of initial weight selection on feed-forward networks learning simple functions with the back-propagation technique. We first demonstrate, through the use of Monte Carlo ... / method of choice for many neural network projects and for good reason. br carefully circumvented by many neural network researchers e.g. through the
47.4 Simulated Annealing for Hard Satisfiability Problems - Spears (1993)(Correct)
Satisfiability (SAT) refers to the task of finding a truth assignment that makes an
arbitrary boolean expression true. This paper compares a simulated annealing algorithm
(SASAT) with GSAT (Selman et ... / Spears showed that a neural network with simulated annealing br satisfiability problems. The neural network algorithm makes no
47.0 Optimal Prefetching via Data Compression - Vitter (1991)(Correct)
Caching and prefetching are important mechanisms for speeding up access
time to data on secondary storage. Recent work in competitive online algorithms
has uncovered several promising new algorithms f... / Palmer and Zdonik who use a neural network approach to prediction PaZ
46.9 Comparative Bibliography of Ontogenic Neural Networks - Fiesler (1994)(Correct)
This document will be published in: Proceedings of the International Conference on
Artificial Neural Networks (ICANN '94).
until it is large enough to handle the problem (or to eliminate possible lo... / Bibliography of Ontogenic Neural Networks E. Fiesler IDIAP Case br of the most powerful aspects of neural networks is their ability to adapt to
46.9 Classification Using Hierarchical Mixtures Of Experts - Waterhouse, Robinson (1994)(Correct)
There has recently been widespread interest in the use of multiple
models for classification and regression in the statistics and neural networks
communities. The Hierarchical Mixture of Experts (HME)... / In Proc. IEEE Workshop on Neural Networks for Signal Processing IV pp. br in the statistics and neural networks communities. The
46.8 Constructive Neural Network Learning Algorithms for Multi-Category.. - Parekh, Yang, Honavar (1997)(Correct)
Constructive learning algorithms offer an attractive approach for incremental
construction of potentially near-minimal neural network architectures for pattern
classification tasks. These algorithms h... / Constructive Neural Network Learning Algorithms for br Neural nets. Keywords neural networks constructive learning
46.8 Adaptive On-line Learning in Changing Environments - Murata, Müller, Ziehe, Amari (1997)(Correct)
An adaptive on-line algorithm extending the learning of learning
idea is proposed and theoretically motivated. Relying only on gradient
flow information it can be applied to learning continuous
functi... / signals. Introduction Neural networks provide powerful tools to br to this data. To this end the neural network modifies its parameter w t
46.5 A Framework for Combining Symbolic and Neural Learning - Shavlik (1992)(Correct)
This article describes an approach to combining symbolic and connectionist approaches to
machine learning. A three-stage framework is presented and the research of several groups is
reviewed with resp... / of symbolic knowledge into neural networks the second addresses the br Keywords knowledge-based neural networks theory refinement use of
46.3 Global Optimization for Neural Network Training - Shang (1996)(Correct)
In this paper, we study various supervised learning methods for training feed-forward neural networks.
In general, such learning can be considered as a nonlinear global optimization problem in which t... / Global Optimization For Neural Network Training Yi Shang And br for training feed-forward neural networks. In general such learning
46.3 Context-Dependent Classes in a Hybrid Recurrent Network-HMM Speech.. - Kershaw, Robinson, Hochberg (1995)(Correct)
A method for incorporating context-dependent phone classes in
a connectionist-HMM hybrid speech recognition system is introduced.
A modular approach is adopted, where single-layer networks
discriminat... / Although the recurrent neural network RNN model acoustic context br i. However training recurrent neural networks in this format would be
45.7 Generalized Support Vector Machines - Mangasarian (1998)(Correct)
By setting apart the two functions of a support vector machine: separation
of points by a nonlinear surface in the original space of patterns,
and maximizing the distance between separating planes in ... / d ffl Example . Neural Network Kernel AB ab br kernels such as the neural network kernel with a discontinuous
45.4 Towards Unrestricted Lip Reading - Meier, Stiefelhagen, Yang, Waibel (1999)(Correct)
Lip reading provides useful information in speech perception
and language understanding, especially when the auditory
speech is degraded. However, many current automatic
lip reading systems impose som... / Multiple State-Time Delayed Neural Network MS-TDNN system. We have br Another approach is to use a neural network to compute the combination
45.3 Explanation-Based Neural Network Learning for Robot Control - Mitchell, Thrun (1993)(Correct)
How can artificial neural nets generalize better from fewer examples? In order to generalize successfully, neural network learning methods typically require large training data sets. We introduce a ne... / CA Explanation-Based Neural Network Learning for Robot Control br to generalize successfully neural network learning methods typically
44.4 On the Analysis of Pattern Sequences by Self-Organizing Maps - Kangas (1994)(Correct)
This thesis is organized in three parts. In the first part, the Self-Organizing Map algorithm is introduced. The discussion focuses on the analysis of the Self-Organizing Map algorithm. It is shown th... / Map iv . Neural Networks and Analysis of Pattern br . . Time-Delay Neural Network
43.4 Cellular Encoding for Interactive Evolutionary Robotics - Gruau, Quatramaran (1996)(Correct)
Research in robotics programming is divided
in two camps. The direct hand programmming
approach uses an explicit model or a behavioral
model ( subsumption architecture). The machine
learning community... / learning community uses neural network and or genetic algorithm. We br recurrent dynamic Artificial Neural Networks ANN as a potentially
43.4 The Evolution of Emergent Computation - James Crutchfield (1995)(Correct)
on of simple components
has important advantages over explicit central control in both natural and human-constructed
information-processing systems. There are substantial costs incurred in having cent... / distributed processing of neural networks typically has some br with a counter register or a neural network with global connectivity. But
42.5 Inductive Logic Programming for Natural Language Processing - Raymond Mooney (1997)(Correct)
This paper reviews our recent work on applying inductive
logic programming to the construction of natural language processing
systems. We have developed a system, Chill, that learns a parser from a
... / of recent research on applying neural-network techniques such as simple br that learns to control a neural-network parsing architecture that
42.5 Adapting Simulated Behaviors For New Characters - Hodgins, Pollard (1997)(Correct)
This paper describes an algorithm for automatically adapting existing
simulated behaviors to new characters. Animating a new character
is difficult because a control system tuned for one character
wil... / Panne and Fiume to produce neural network-based control systems for a
42.5 Learning With Unreliable Boundary Queries - Blum, Chalasani, Goldman, Slonim (1997)(Correct)
We introduce a model for learning from examples and membership queries in situations where the boundary between positive and negative examples is somewhat ill-defined. In our model, queries near the b... / hypothesis class say a simple neural network to learn images of 's. For br IEEE Transactions on Neural Networks - . E. B.
41.9 Adaptive Sentence Boundary Disambiguation - Palmer, Hearst (1994)(Correct)
Labeling of sentence boundaries is a necessary prerequisite for many natural language processing tasks, including part-ofspeech tagging and sentence alignment. End-of-sentence punctuation marks are am... / and a feed-forward neural network. This work demonstrates the br report using a feedforward neural network to disambiguate periods
41.2 A Distributed Reinforcement Learning Scheme for Network Routing - Littman, Boyan (1993)(Correct)
In this paper we describe a self-adjusting algorithm for packet routing, in
which a reinforcement learning module is embedded into each node of a
switching network. Only local communication is used to... / d is often approximated by a neural network see e.g. this can br Learning for Robots Using Neural Networks. PhD thesis School of
40.5 A New Evolutionary System for Evolving Artificial Neural Networks - Yao (1996)(Correct)
This paper presents a new evolutionary system, i.e., EPNet, for evolving artificial neural networks (ANNs). The evolutionary algorithm used in EPNet is based on Fogel's evolutionary programming (EP) [... / Ieee Transactions On Neural Networks A New Evolutionary System br System for Evolving Artificial Neural Networks Xin Yao S.M.IEEE and
40.5 Dynamically Adding Symbolically Meaningful Nodes to Knowledge-Based.. - Opitz, Shavlik (1995)(Correct)
Traditional connectionist theory-refinement systems map the dependencies of a domainspecific
rule base into a neural network, and then refine this network using neural learning
techniques. Most of the... / Nodes to Knowledge-Based Neural Networks David W. Opitz and Jude br rule base into a neural network and then refine this network
39.9 Evolving and Breeding Robots - Lund, Miglino (1998)(Correct)
Our experiences with a range of evolutionary robotic experiments
have resulted in major changes to our set-up of artificial life experiments
and our interpretation of observed phenomena. Initially, we... / work with evolution of neural network agents to real robots. This is br agents e.g. controlled by neural networks to model
39.9 Building Intelligent Agents for Web-Based Tasks: A Theory-Refinement.. - Shavlik, Eliassi-Rad (1998)(Correct)
We present and evaluate an infrastructure with
which to rapidly and easily build intelligent software
agents for Web-based tasks. Our design is
centered around two basic functions: ScoreThisLink
and ... / This advice is mapped into neural network implementations of the two br are then compiled into neural networks thereby allowing subsequent
39.5 Fast Pruning Using Principal Components - Asriel Levin (1994)(Correct)
We present a new algorithm for eliminating excess parameters and
improving network generalization after supervised training. The
method, "Principal Components Pruning (PCP)", is based on principal
com... / can be extended to multilayer neural networks. A complete analysis of the br Consider now a feedforward neural network where each layer is of the
39.1 Exploration and Model Building in Mobile Robot Domains - Thrun (1993)(Correct)
I present first results on COLUMBUS, an autonomous mobile robot. COLUMBUS
operates in initially unknown, structured environments. Its task is to explore and model
the environment efficiently while avo... / International Conference on Neural Networks San Francisco CA March br generalized via two artificial neural networks that encode the
39.1 Multitask Learning: A Knowledge-Based Source of Inductive Bias - Caruana (1993)(Correct)
This paper suggests that it may be easier to learn
several hard tasks at one time than to learn these
same tasks separately. In effect, the information
provided by the training signal for each task
se... / of MTL applied to artificial neural networks. After this demonstration we br tennis. An artificial neural network or a decision tree trained
39.1 Incremental Evolution of Neural Network Architectures for Adaptive.. - Cliff (1993)(Correct)
This paper describes aspects of our ongoing work in evolving recurrent dynamical
artificial neural networks which act as sensory-motor controllers, generating adaptive
behaviour in artificial agents. ... / Incremental Evolution of Neural Network Architectures for Adaptive
38.2 Probabilistic Analysis of Learning in Artificial Neural Networks: The .. - Anthony (1997)(Correct)
There are a number of mathematical approaches to the study of learning and generalization in
artificial neural networks. Here we survey the `probably approximately correct' (PAC) model
of learning and... / of Learning in Artificial Neural Networks The PAC Model and its br generalization in artificial neural networks. Here we survey the probably
38.2 Computational capabilities of recurrent NARX neural networks - Siegelmann, al. (1997)(Correct)
Recently, fully connected recurrent neural networks have been proven to be computationally
rich --- at least as powerful as Turing machines. This work focuses on another network which
is popular in co... / capabilities of recurrent NARX neural networks y Hava T. Siegelmann br fully connected recurrent neural networks have been proven to be
38.2 Exploiting Global Input/Output Access Pattern Classification - Tara Madhyastha (1997)(Correct)
ly, an access pattern is a qualitative
statement describing future file accesses that can
be used to select and tune file system policies.
Applications often have qualitative, recognizable
input/outpu... / train a feedforward artificial neural network ANN to classify br patterns. We provide the neural network with examples of access
38.2 Homograph Disambiguation in Text-to-speech Synthesis - Yarowsky (1997)(Correct)
This chapter presents a statistical decision procedure for lexical
ambiguity resolution in text-to-speech synthesis. Based on decision lists,
the algorithm incorporates both local syntactic patterns a... / vector space model and neural network LTV Homograph
38.2 Extraction of crisp logical rules using constrained backpropagation.. - Duch, Adamczak, Grabczewski (1997)(Correct)
The problem of extraction of crisp logical rules from neural networks trained with backpropagation algorithm is solved
by transforming these networks into simpler networks performing logical functions... / of crisp logical rules from neural networks trained with backpropagation br Many methods to analyze trained neural networks extract logical rules and