Neural architecture for echo suppression during sound source localization based on spiking neural cell models [Elektronische Ressource] / von Thomas Peter Zahn

Description

PNeuralBenediktArcGutachitecturebChicagoJuniEcThoJ.SuppressionvduringoSoundhenSourceDr.-Ing.LoUniv.-Prof.calizationfbasedersitonrSpikingDipl.-Ing.NeuralZahnCellamMoindelsh/VDissertationter:zurhaelErlangungIdesrer.llinoisLMUademiscohenark,Grades1.DezemDoktoringenieurI(Dr.-Ing.)2003anonderThomasFeterakult?tgebf?rrenInformatik5.und1965AutomatisierungReicderbacTogtl.echhniscUniv.-Prof.henHorst-MicUnGross,vUersit?tlmenauIlmenauDr.vnat.orgelegtGrothe,amM?nc1.OktobUniv.-Prof.erThomas2003PvUniverteidigtyameDr.henakfor.ha3dels.SummaryaThiscessingthesislainavtrinsicestigatestimecessing.fbiologicaldistributedbacokgrounddevofMAproellpsyctoho-acousticalandprecedencemixedinformationossibilitenablingphonumansprinciplestoItsuppress(NBL),fmohoailableesofduringusagethebiologicallyloItscalizationhangesofesoundtssources.implemenItwpro1videsproatteccessorshnicallysystems.

Sujets

• Gesundheit

Informations

• Publié par: technische_universitat_ilmenau
• Publié le: 01 janvier 2003
• Nombre de lectures: 8
• Langue: English
• Poids de l'ouvrage: 19 Mo

Extrait

P
Neural
Benedikt
Arc
Gutac
hitecture
b
Chicago
Juni
Ec
T
ho
J.
Suppression
v
during
o
Sound
hen
Source
Dr.-Ing.
Lo
Univ.-Prof.
calization
f
based
ersit
on
r
Spiking
Dipl.-Ing.
Neural
Zahn
Cell
am
Mo
in
dels
h/V
Dissertation
ter:
zur
hael
Erlangung
I
des
rer.
llinois
LMU
ademisc
o
hen
ark,
Grades
1.Dezem
Doktoringenieur
I
(Dr.-Ing.)
2003
an
on
der
Thomas
F
eter
akult?t
geb
f?r
ren
Informatik
5.
und
1965
Automatisierung
Reic
der
bac
T
ogtl.
ec
h
hnisc
Univ.-Prof.
hen
Horst-Mic
Un
Gross,
v
U
ersit?t
lmenau
Ilmenau
Dr.
v
nat.
orgelegt
Grothe,
am
M?nc
1.Oktob
Univ.-Prof.
er
Thomas
2003
P
v
Univ
erteidigt
y
am
e
Dr.
hen
ak
for.the
under
The
rst
be
a
h
the
of
the
auditory
of
be
a
h
vior
yp
e
of
neural
of
hh
a
s
eI
n
pro
as
w
ell
as
to
ell
DF
olleg,
be
to
use
tial
and
of
and
n
i
the
for
enables
sound
pro
f
pattern
lo
on
calization
en
and
terface
ec
s
hoic
p
conditions,
the
r
precedence
eady
sim
to
of
b
cessing.
e
eural
used
team
b
the
y
distributed
tec
of
hnical
designed
systems
m
d
the
uring
dieren
man-mac
of
hine
design
in
biologically
teractions.
bac
The
the
mo
o
del
een
i
First
s
olleg
based
on
up
dev
on
cessors
o
systems.
wn
as
electro-ph
extend
ysiological
rho
exp
(NBL),
e
dev
rimen
mo
ts
dels.
in
ailable
the
MA
mongolian
of
gerbil.
trinsic
feasible
usage

ell
hnically
biologically
time
cessing
in
Its
gerbils
to
obtained
hanges
results
time
rev
estigates
o
and
a
ts
sp
ha
ecial
implemen
tec
mixed
a
realized
vides
v
vior
and
of
o
sp
of
ecic
out
cells
the
of
design
pro
apply
dorsal
cessing
n
dier
ucleus
b
It
spik
sources.
v
lateral
ding
lemniscus
amplitude)
(DNLL)
(frequency
-
Base
a
h
distinct
e
region
ed
in
purp
the
of
sound
teraction
brainstem.
library
The
a
explored
sim
p
t
ersisten
b
t
mo
inhibition
and
eect
the
of
prop
these
cells.
cells
this
seems
as
to
biologists
accoun
o
t
mo
for
information
the
the
base
dynamics
e
raphical
ec
vides
ho
of
suppression
parametric
at
is
higher
y
auditory
of
cen
as
ters.
macroscopic
The
sim
yp
T
elop
asic
ed
the
mo
ase
del
e
pro
repared
v
b
ed
ecial-
capable
circuitry
to
w
duplicate
y
t
f
calization
G
lo
1
the
v
during
ossibilit
and
fully
suggests,
pro
t
By
hat
la
a
cessor
strong
t
a
in
nd
ssible
timely
ecic
precise
order
h
principles
yp
information
erp
tec
olarization
pro
is
classical
the
pro
basic
the
mec
temp
hanism
pattern
b
3
ehind
They
this
digital
cell
b
b
of
eha
b
vior.
d),
The
phase
dev
Neural
elop
Library
ed
whic
neural
enabling
arc
b
hitecture
en
mo
elop
dels
for
the
ulti
inner
ose
ear
deling
as
Spik
w
ho-acoustical
ell
Mo
as
This
v
extends
e
commercially
m
v
a
dynamic
jor
ulation
n
vironmen
uclei
of
o
TLAB/SIMULINK
f
y
t
t
he
dels
audi-
neurons
tory
synapses
brainstem
ulating
ew
in
their
dynamic
connectivit
erties
y
neural
and
The
in
p
trinsic
library
dynamics.
engineers
It
w
represen
as
ts
to
a
their
n
analog
t
plausible
es
dels
ho
neural
information
pro
suppress
without
mo
need
deling
detailed
describ
skills.
ed
g
as
in
Spik
psyc
e
access
In
structural
teraction
kground
Mo
biological
dels
c
(SIM).
and
SIM
capable
use
displa
the
the
precise
course
spatio-temp
microscopic
oral
parameters
in-
w
teraction
as
to
as
single
during
spik
ulations
e
thereafter.
ev
w
a
b
duce
elemen
for
of
co
N
ding
B
and
Library
pro
v
cessing
b
of
p
neural
for
information.
tation
Their
y
basic
p
elemen
ized
tro
analog-digital
del
.
mo
silicon
implemen
tations
tegrate-and-Fire
ere
Neurons
b
and
the
Hebbian
o
synapses,
the
whic
G
h
raduiertenk
ha
GRK
v
64
e
pro
b
ed
een
p
extended
y
b
f
y
parallel
sp
line
ecially
cessing
designed
sounds.
dynamic
using
transfer
automated
functions.
y
The
pro
mo
under
del
elopmen
is
in
capable
Graduiertenk
plausible
it
will
thesis
detect
o
tials
ell
dierences
sp
as
pro
small
in
a
This
s
the
10
w
s
biological
and
pro
emplo
to
ys
hnical
the
These
principles
cessors
o
from
f
v
coincidence
Neumann
detection
cessors
and
y
precise
Summary
umans
spatio
o
oral
cal
e
inhibition
instead
for
sequen
a
binary
uditory
alues.
pro
will
cessing.
the
It
co
consists
principle
exclusiv
y
ely
dimensions
of
space
oten
neigh-
ts
o
of
o
a
time
s
,
p
ecically
l
elemen
(spatial
of
of
to
ring
cell
programming
for
wn,
of
time
to
In
are
ts
ts
en
new
in
this
dev
of
eal
source
ec
eect,
theen
of
to
ex-
and
of
It
dy-
and
to
em-
of
and
the
the
mo
del
the
of
ated
and
ec
gap
be
t
mo
deling
of
vironmen
study
duces
arising
of
from
the
the
pro
attempt
conditions.
to
of
in
tains
v
up
estigate
y
the
using
biological
areas
principles
summary
of
b
sound
plo
pro
p
cessing
whereas
and
ysiological
mak
p
e
the
them
neuro-
a
as
v
results
ailable
and
to
tation.
tec
future
hnical
con
s
een
ystems
displa
in
sets
teracting
for
with
the
h
consists
umans
describ
under
vides
real
The
w
1
orld
the
conditions.
ho-acoustical
F
,
urthermore,
of
v
orld
e
signals,
reasons
highly
to
paradigm
use
motiv
spik
oral
e
calization
in
status
teraction
pro
mo
ok
dels
jects
are
This
giv
eld
en
t
and
computational
t
midbrain.
heir
the
no
ed
v
c
applications.
hin
hnical
set
haracteristics
conguration
are
ulation
discussed.
Chapter
Chapter
t
2
rst
in
in
tro
results
duces
wn
t
erimen
he
describ
biological<