acetylcholine, 131 
active learning, 295 
active set method, 938 
active vision, 858 
actor-critic, 1073 
adaptation, 678, 837 
adaptive algorithms, 381 
adaptive back-propagation, 323 
adaptive behavior, 38 
adaptive conlxol, 1010 
adaptive learning, 896 
adaptive metric, 409 
adaptive routing, 945 
ADDEMUP, 535 
additive clustering (ADCLUS), 3 
advantage updating, 1059 
adverserial learning, 896 
alermess, 931 
alphanumeric fields, 778 
amodal completion, 816 
analog circuits, 671 
analog VLSI, 699, 720, 938 
animal cognition, 61 
anomaly detection, 528, 924 
anterior thalamus, 152 
applications, 1031 
approximation rate, 183 
architectural limitations, 612 
area MT, 68 
ARMAX processes, 204 
articulated matching, 795 
articulated objects, 795 
asset allocation, 952 
assimilation, 844 
associative memory, 131 
associative networks, 1080 
asymptotic stability, 225, 372 
asymptotic theory, 176, 295 
asynchronous data, 395 
attention, 633, 802 
attention dynamics, 38 
atlxactor dynamics, 989 
attractor networks, 253 
audition, 699 
auditory modelling, 729 
auditory perception, 110 
auditory scene analysis, 699 
auditory sixearning, 52 
auditory system, 124 
auditory template hypothesis, 110 
autoassociator, 402, 924 
automatic relevance determination, 514 
autonomous robot, 989, 1031 
backgrotmd constancy, 844 
backpropagation through time (BPTr), 577, 743 
backpropagation training, 159, 225, 619 
bagging, 535, 542 
barn owl, 124 
basis functions, 10 
Baum-Welch, 472 
Bayes classifier, 542 
Bayes risk, 232 
Bayesian analysis, 514 
Bayesian-Kullback, ddd 
Bayesian learning, 402, 598 
Bayesian methods, 351 
Bayesian networks, 500 
Bayesian penalty terms, 542 
Bayesian regression, 1066 
belief network, 528, 661 
Benary Cross, 844 
beta sheet prediction, 917 
bias learning, 169 
bifurcation (saddle-node bifurcation), 38 
binary lxee, 507 
binding, 809 
bipolar cell, 159 
birdsong learning, 110 
bit stream neural networks, 267 
blind separation, 757 
blurring, 45 
boolean functions, 260 
boosting, 479 
bootstrap, 882 
brain lesions, 10 
brain stem, 89 
C4.5, 24, 654 
capacity, 556 
CART, 507 
CART analysis, 966 
center-surround opponent receptive field, 159, 678 
cerebellum, 89, 138 
channel selection, 910 
character recognition, 423 
check reader, 938 
classification, 197 
classification tree, 507 
classifiers, 176 
clustering, 3, 416 
CMAC, 1038 
CMOS, 685 
coarticulation, 486 
cochlea, 910 
cochlear implants, 910 
cognitive modeling, 3, 10 
Cohen-Grossberg model, 337 
coherence, 931 
color constancy, 844 
color segmentation, 903 
combinatorial optimization, 626 
combining classifiers, 535 
combining estimators, 190 
committee, 882 
compact representations, 1045 
transistors, 671 
competence acquisition, 1031 
competition, 837 
competitive networks, 82 
complexity, 246, 549 
complexity regularization, 183 
compliant control, 1003 
comprehensibility, 24, 654 
computational complexity, 211 
computational learning theory, 288 
computational power, 612 
computer vision, 875 
condition monitoring, 924 
confidence levels, 260 
conjugate gradient, 633 
coincidence detection, 124 
conjugate prior, 542 
connectionist reinforcement algorithm, 889 
constrained optimization, 938 
context-dependency, 750 
context dependent learning, 570 
context free grammars, 31 
continuous activations, 197 
continuous-function learning, 896 
continuous-time model, 1073 
continuous variables, 500 
continuous wavelet transform, 692 
contrast-filling-in, 844 
control, 973 
control, direct, 1052 
control, online, 1052 
convergence, 423 
convergent learning, 274 
cooperation, 1017 
correlations, 917, 931 
correspondence, 795 
cortical reorganization, 82, 131 
cost functions, 423 
covariance function, 514 
CRAWL model, 61 
cross-correlation, 68 
cross-validation, 176, 183, 190, 218, 882 
cumulants, 437, 757 
curse of dimensionality, 10, 409, 1045 
data association, 591 
data classification, 416 
data compression, 661 
data sorting circuit, 685 
DAX stock index, 952 
decentralized control, 1017 
decision trees, extraction of 24, 45 
decorrelation, 736 
delta-bar-delta, 563 
density estimation, 465, 528, 661 
depth perception, 816 
deterministic annealing, 591,626 
development, 96 
dichotomics, 556 
digital flitering, 204 
dimensionality reduction, 10, 330 
discrete event dynamic systems, 1017 
discretized models, 372 
discriminant analysis, 409 
discriminant training, 388 
discriminitive learning, 591 
disinhibition, 816 
Donders' law, 117 
dorsal pathway, 802 
drowsiness, 931 
dual constraint model, 96 
dual space method, 938 
dynamic parameter adaptation, 225 
dynamic programming, 952, 1017, 1038, 1045, 1059, 1073 
dynamic wave model, 549 
dynamical systems approach, 989 
dynamics, 253 
early stopping, 176, 218, 365, 959 
EBNN, 640 
edge detection, 159 
EEG, 145, 931 
elastic net algorithm, 330 
elevator control, 1017 
EM algorithm, 3, 351,444,465, 472, 486, 528, 542, 647, 1003 
embedding problem, 989 
EMMA, 851 
energy, 31 
energy bands labelling, 910 
energy minimization, 917 
ensemble, 479, 535, 598, 882 
ensemble learning, 190, 351 
entropy, 458, 1080 
equilibrium points, 337 
equivariant property, 757 
error attenuation, 563 
estimation rate, 183 
eutropic loss, 316 
evoked response, 145 
evolution of complexity, 549 
evolutionary programming, 38 
exact identification, 288 
expectation-maximization, 395, 402 
experiment design, 1066 
exploration, 945 
exponentiated gradient algorithm, 309 
extra outputs, 959 
Facial Action Coding System 
FACS, 823 
facial expression recognition, 823 
factor analysis, 465 
factorization, 437 
family discovery, 402 
family relations task, 563 
fan-in/fan-out of nodes, 563 
feature discovery, 3 
feature extraction, 1024 
feature measurement, 823 
feature selection, 45, 232 
feature space models, 330 
feedback, 809 
feedforward networks, 145, 176, 197, 218, 246, 323, 931 
finite automata, 211 
finite state automata, 612 
focus of attention, 802 
focus of expansion, 720 
Fokker- Planck equation, 103 
forward-backward algorithm, 743 
Fourier transform, 260 
free energy, 351,661 
function approximation, 1038, 1045 
fusion of information, 802 
gabor projections, 778 
gain fields, 10 
gamma memory, 785 
gamma MLP, 785 
gamma filter, 785 
Gaussian, 556 
Gaussian mixtures, 395, 542 
Gaussian processes, 514 
generalization, 169, 176, 218, 267, 423, 458, 
521,598, 959, 1038 
generalization error, 344 
genetic algorithms, 430, 535 
genetic programming, 430 
gesture recognition, 858, 903 
Gibbs sampling, 472, 500, 514 
global convergence, 372 
GLVQ, 423 
gradient descent, 302, 309, 316 
gradient dynamics, 274 
Gram-Charlier expansion, 757 
grammatical induction, 612 
graph partitioning, 626 
face detection, 875 Hamilton, Jacobi, Bellman elevation, 1059 
Hamiltonian dynamics, 274 
hand recognition, 903 
handprint recognition, 778 
handwriting recognition, 743, 771,736, 764 
hard competition, 239 
hardware, 1031 
hardware implantation, 699 
harmonic functions, 996 
harmony, 31 
head direction cell, 61,152 
Hebbian leaming, 124, 131 
Helmholtz free energy, 591 
Helm
holtz machine, 444, 661 
hemineglect, 10 
hidden Markov models, 472, 485,493, 750, 1003 
hidden state, 858 
hierarchical architectures, 493 
hierarchical mixtures of experts, 351,584 
hierarchical network, 570 
hierarchical priors, 598 
higher order statistics, 437 
hillclimbing, 430 
hippocampus, 61, 152 
HIB equation, 1073 
Horwitz-matrix, 337 
human reading, 10 
humanoid hand, 889 
HVC, 110 
hybrid Monte Carlo, 514, 598 
Models, 764 
HyperNet, 980 
hyperparameters, 598 
hypothesis boosting, 654 
I1000 chip, 938 
ICA, 145 
ID2-of-3, 24 
ID3leaves, 45 
illumination constancy, 844 
image classification, 409 
image processing, 633 
image recognition, 823 
image understanding, 875 
improved current precision, 671 
incremental learning, 605, 896 
independent component analysis, 757 
index of resoluability, 183 
industrial production, 458 
information, 145 
information conservation, 437 
information matrix, 295 
information theory, 75, 281 
inhibition, 68, 837 
inhibition of retum, 802 
input cycles, 612 
input/output hidden Markov models, 493 
input representations, 45 
instinct-rules, 1031 
integrate-and-fire neurons, 75, 103, 124, 729 
intelligent sensory processing, 52 
internal representation, 736 
invariances, 640 
inverse problems, 145 
investment learning, 570 
irregular computation, 713 
job-shop scheduling, 1024 
KBANN, 535 
Kalman filter estimation' 239 
knowledge-based neural networks, 535 
Kohonen networks, 110 
kriging, 514 
latent variable, 465, 661 
lateral connections, 736 
learning, 851,875 
learning curves, 344 
learning dynamics, 302 
learning from examples, 302 
learning invariances, 640 
learning rates, 563 
learning-rate adaptation, 225 
learning representations, 640 
learning rule, 337 
learning theory, 260, 323, 365 
learning with queries, 24 
lie algebra, 117 
lifelong learning, 640 
lightness perception, 844 
likelihood bounds, 528 
limit cycling, 989 
linear models, 365 
linear networks, 190 
linear threshold element, 246 
LISSOM, 736 
Listing's law, 117 
local bias, 605 
local distance metric, 605 
local feature detection, 605 
local learning, 605 
local linear map, 903 
local minima, 309, 316 
locally weighted regression, 1066 
long/short strategy, 966 
long-term dependencies, 493, 577 
LVQ, 423 
Lyapunov function, 337 
manipulation learning, 889 
man-machine interface, 903 
MAP estimation, 388 
Markov decision problems, 952 
Markov decision process, 1052 
mask perceptron, 973 
matching loss function, 309, 316 
matrix perturbation theory, 337 
maximum entropy, 591 
maximum information, 444
maximum liklihood, 528, 591 
McCulloch-Pitts neurons, 211 
mean field theory, 472, 486, 661 
medical diagnosis, 882 
medical risk, 959 
memory-based learning, 640, 896, 945, 1066 
minimum description length, 507 
missing data, 395, 465, 647 
missing values, 395 
mixture models, 381,472, 1003 
mixture of experts, 351,584, 1080 
MLP, 556 
model-based recognition, 713 
model selection, 183 
modular networks, 239 
modular system, 903 
modulation, 131,692 
morphogenesis, 96 
motion, 68, 809 
motion control, 996 
motion detection, 706 
motion perception, 837 
motion sensor, 720 
motors, 924 
motor control, 138 
mountain car, 1038 
multifactor models, 966 
multilayer perceptron, 295, 323 
multiple models, 239, 858 
multiple time scales, 493 
multiplication, 809 
multiscale, 633 
multiscale filters, 591 
multitask learning, 959 
mutual information, 757 
mutual neighborhood value, 416 
NARX networks, 577 
natural basis functions, 591 
natural gradient, 757 
navigation, 61,152, 989 
neural hardware engineering, 671 
near-neighbor classification, 409 
nearest neighbor, 232, 640 
network averaging, 542 
network structure, 288, 563 
neural coding, 75, 281 
neural controller, 1010 
neural development, 816 
neural field theory, 82 
neural integration, 152 
neuro-fuzzy, 952 
neuromodulation, 131 
neuromorphic architectures, 720 
neuromuscular junction, 96 
neuron, 152 
neuron gains, 372 
neuron MOS transistor, 685 
neuronal coding, 211 
new learning models, 444 
noise robustness, 211 
non-Gaussian distributions, 437 
non-linear, 500 
non-linear systems, 1010 
non-parametric clustering, 416 
nonlinear control, 1073 
nonlinear feature extraction, 437 
nonlinear perceplxon, 316 
nonlinear system identification, 274 
nonparametric statistics, 605 
non-rigid matching, 795 
novelty detection, 924 
nuisance parameters, 521 
OBD, 521 
OBS, 521 
object detection, 875 
object recognition, 795, 865 
occlusion, 816 
OCR, 10, 479, 938 
ocular dominance columns, 330 
oculomotor control, 117 
oculomotor system, 89 
on-line learning, 302, 309, 381,757 
onset cells, 729 
onset-offset filter, 729 
opponent processing, 816 
optic flow, 720, 823 
optical character recognition, 10 
optimal design, 295 
optimization, 330, 372, 430, 851, 1066 
Ornstein-Uhlenbeck Process, 103 
oscillators, 451 
otoliths, 89 
overfitting, 176, 190, 218, 458 
oversampling, 692 
overtraining, 176, 458 
PAC learning, 197, 204, 267, 288, 344, 654 
parallel architectures, 996 
parallel and sequential dynamics, 372 
parallel hardware processing, 720 
parameter adaptation, 225 
parameterized models, 402 
parameterized self-organizing map, 570, 903 
parametric uncertainty, 1010 
parietal cortex, 10 
particle monitoring, 980 
Parzen windows, 500 
path integration, 61 
path planning, 996 
path pruning, 584 
pattern recognition, 232, 253, 736, 875 
perceptton, 218 
performance comparison, 598 
period-doubling bifurcation, 372 
phase transition, 416 
phase locking, 124 
phoneme recognition, 785 
physiology, 837 
piecewise affine transformations, 795 
pitiform cortex, 131 
pixel classification, 232 
place cells, 61, 152 
pneumonia, 959 
point matching, 795 
Poisson model, 103 
portfolio management, 952 
positioning, 778 
Potts model, 416 
prediction, 138, 809 
predictive Q-routing, 945 
predictivity, 816 
presynaptic inhibition, 131 
principal components, 437, 465, 823 
principal component pruning, 966 
probabilistic networks, 486, 528 
probabilistic transducers, 381 
probing, 945 
product distributions, 288 
promoter recognition, 647 
protein stxucture prediction, 917 
pruning, 521 
pseudo-dimension, 204 
PSOM, 570 
psychophysics, 837 
pulse stxeam, 1031 
Q-learning, 858, 945, 952, 1017 
Q-routing, 945 
quadratic assignment, 626 
radial basis functions, 239, 591 
RAM-based nodes, 980 
rankprop, 959 
rapid learning, 570 
rat, 61,131,152 
read-once formulas, 288 
reading, 10 
real-world data, 598 
receptive fields, 10, 605 
recurrent cascade correlation, 612 
recurrent networks, 38, 89, 204, 253, 274, 395, 458, 493, 577, 612, 743, 750, 771, 837, 973, 989 
recurrent perceptrons, 204 
recursive estimation, 239 
recursive model selection, 239 
reference frames, 10 
REGENT, 535 
regression, 514 
regularization, 190, 458 
reinforcement learning, 110, 858, 945, 1017, 1024, 1038, 1045, 1052, 1073, 1080 
relaxation, 633 
relaxing network, 395 
representation quality, 45 
representation selection, 45 
resampling techniques, 882 
response function, 190 
response surface methods, 1066 
retina, 159 
ring oscillator, 685 
ridge function, 556 
robot guidance, 903 
robot learning, 570, 605 
robotic control, 1031 
robotic soccer, 896 
rotation group, 117 
rule extraction, 24 
saccades, 117, 591 
saliency, 521 
saliency map, 591,802 
saliency measure, 633 
sample complexity, 197, 204 
sample sizes, 267 
sarsa, 1038, 1052 
scheduling, 430 
second-order low-pass filters, 671 
secondary structure prediction, 917 
segmentation, 743, 778 
selective attention, 771 
self-organization, 110, 131,500 
self-organizing algorithms, 330 
self-organizing map, 736 
sensory signal processing, 713 
sequence classification, 388 
sequential associations, 52 
sequential data, 395,493 
shunting of error signals, 563 
sigma-pi nodes, 980 
sigmoid, 218 
sigmoidal networks, 197 
sigmoidal transfer function, 316 
Signal-to-Noise Ratio (SNR)-dependent 
plasticity, 159 
signal processing, 851 
silicon audition, 699 
silicon cochlea, 671,699 
silicon retina, 678, 706 
SIMD/MIMD hybrid, 713 
similarity, 3 
simple neuron, 316 
simulated annealing, 917 
single neuron, 309 
singularity, 295 
sleep, 931 
sliding window, 778 
small training data set, 570 
smooth regularizers, 458 
soft assign, 626, 795 
soft competition, 239 
soft max function, 591, 626 
somatosensory cortex, 82, 131 
song sparrow, 110 
sorting, 959 
sound segmentation, 729 
source analysis, 145 
sparse perceptrons, 654 
spatial reasoning, 61 
spatial representations, 10 
spario-temporal receptive field, 159, 678 
spectral classification, 910 
speech processing, 910 
speech recognition, 388,402, 486, 750, 785, 910 
SPERT, 619 
SPERT-II, 619 
spike coding, 75, 124, 281,699 
spike trains, 68 
spiking, 809 
spiking neurons, 211 
splice junction determination, 647 
SPMD, 713 
square loss, 316 
stability, 1010 
stability criterion, 225 
stable dynamic parameter adaptation, 225 
state of contact estimation, 1003 
statistical mechanics, 218, 302, 323,416 
statistical methods, 288 
statistical physics, 626 
statistics, 851 
steepest descent, 423 
stochastic, 430 
stochastic approximation, 1045 
stochastic computing, 267 
stochastic control, 1045 
stock selection, 966 
strong unimodality, 288 
structure learning, 500 
sub threshold, 671 
suffix trees, 381 
supervised learning, 591 
suppression, 131 
surface learning, 402 
symbolic representations, 24 
symmetric networks, 372 
symmetry breaching, 323 
synaptic competition, 96 
synaptic modification, 131 
synchronous digital hierarchy (SDH), 973 
synchronous optical network (SONET), 973 
system identification, 1010 
TANN, 1010 
target tracking, 706 
TD learning, 1024 
teams, 1017 
telecommunications, 973 
tempering, 563 
templates, 938 
template matching, 823 
temporal dependencies, 493 
temporal difference, 1073 
temporal difference learning, 1038 
temporal winner search, 685 
tensor representation, 31 
thalamocortical, 152 
theoretical analysis, 612 
threshold circuits, 211 
time-delay neural network, 1024 
time series, 472 
time-m-contact, 720 
TIMIT database, 910 
topographic map, 82, 330, 465 
tracking chip, 706 
trade-off theorem for backpropagation, 225 
training-test split, 183 
trajectory learning, 274, 451 
transfer functions, 309 
transfer in learning, 640 
transparency measures, 45 
traveling salesman problem, 626 
tree, 479 
tree growing, 584 
TREPAN, 24 
trust region, 633 
unified learning scheme, 444 
uniform convergence, 169 
universal approximation, 451 
unlabeled data, 647 
unsupervised, 444,809 
unsupervised learning, 124, 416, 661 
value function, 1073 
value function approximator, 952 
Vapnik-Chervonenkis dimension, 183 
VC dimension, 197, 204, 267, 344 
vector processor, 619 
vehicle havigation, 720 
vergence, 89 
vestibular nucleus, 89 
vestibulo-ocular reflex (VOR), 89 
video indexing, 875 
view-based approach, 865 
vigilance, 931 
vision, 633, 851,865, 875 
vision, primate, 802 
vasion chip for tracking, 706 
visual adaptation, 159 
visual attention, 771,858 
visual cortex, 68.330, 591,809, 816, 865 
visual motion chip, 706 
visual processing, 837, 858 
visual search, 591 
visual smooth pursuit, 706 
visualization, 465 
Viterbi algorithm, 743 
VLSI, 699, 1031 
VLSI model of smooth pursuit, 706 
volume conservation, 437 
weak inversion, 671 
weight decay, 458 
weight sharing, 917 
weight size, 246 
"where" pathway, 802 
White's illusion, 844 
winner search hardware, 685 
winner take all circuit, 685 
word recognition, 10 
word sense resolution, 647 
worst-case loss bounds, 309 
Ying-Yang machine, 444 
