Federal gosudarstvennoe uchebnoe predpriyatie Chair of the System of Artificial Intelligence

Neyronnye seti - eto razdel iskusstvennogo intellekta, v kotorom
Basic
perspective of the creation of a universal processor with a uniform structure, capable of
pererabatÿvat raznoobraznuyu information and not trebuyushchego
obobshcheniyu nakoplennyx znaniy. Neyronnaya set obladaet chertami
prostranstvenno-vremennoy obrabotki informatsii. Stoxasticheskie i
in neuronax jivyx sushchestv. Vajneyshaya osobennost neyronnyx setey, svidetelstvuyushchaya
ob ix shirokix vozmojnostyax i ogromnom potentsiale, sostoit v parallelnosti obrabotki dannyx
pri apparatnoy realizatsii. At bolshom kolichestve mejneyronnyx svyazey it pozvolyaet
znachitelno
task.
Methods of training for the formation of resistant to fluctuating parameters
vozmojnym preobrazovanie signalov in realnom vremeni. Chrome togo, pri
set, sposobna obobshchat poluchennuyu informatsiyu i pokazÿvat xoroshie
Krome togo, architecture neyronnyx setey pozvolyaet realizovat ee s
INS. Software-hardware models and hardware realization INS. Neurocomputer. characteristic
commercial
oshibkam, voznikayushchim na nekotorÿh liniyax. Funktsii povrejdennyx
neurocomputers.
ne preterpevaet sushchestvennyx vozmushcheniy.
elements seti neveliko, a ix povtoryaemost ogromna. It opens
dlya obrabotki signalov ispolzuyutsya yavleniya, analogichnÿe proisxodyashchim
Specialized INS. Basic definitions and classification. Seti
Ne menee vajna sposobnost neyronnyx setey k obucheniyu i
obyazatelnogo nalichiya programmÿ obrabotki, dostatochna tolko postanovka
7
Machine Translated by Google

---

robots, imeyut classification tekushchego sostoyaniya i vÿrabotka resheniy o
vychislitelnoy technical, caused in the last year ogromnÿy rost interesa
Dlya klassifikatsii i raspoznavaniya obrazov set nakaplivaet v
vÿrabatÿvaet reshenie o tom, chemu doljno bÿt ravno otsenivaemoe znachenie
baza dlya vÿrabotki novyh tehnologicheskix resheniy, kasayushchixsya
geometric design of the image structure, distribution of the main
In zadachax upravleniya dynamic processes neyronnaya set
soboy nelineynuyu model etogo protsessa i identifitsiruet ego osnovnÿe
metodologicheskuyu osnovu.
Funktsii, vÿpolnyaemÿe setyami, podrazdelyayutsya na neskolko grupp:
approximation; classification and identification of images; prognozirovanie; identification
and evaluation; assotsiativnoe upravlenie. Approximiruyushchaya set plays the role of
a universal approximator
aktsentiruyutsya otlichiya obrazov drug ot druga, kotorye i sostavlyayut osnovu
vida u = f (x), gde x - vxodnoy vector, a u - realizovannaya function
predskazanie budushchego povedeniya sistemy po imeyushcheysya
signal. Vo-vtoryx, set vÿpolnyaet funktsii sledyashchey sistemy, otslejivaet
izmenyayushchiesya usloviya okrujayushchey sredy i adaptiruetsya k
peremennoy v momenty vremeni, predshestvuyushchie prognozirovaniyu, set
sverxvysokoy stepenyu integratsii (VLSI) i povsemestnogo primeneniya
staging.
traditional structures. Vajnoe znachenie, especially when managing
k neyronnym setyam i sushchestvennyy progress v ix issledovanii. Sozdana
protsesse obucheniya znaniya ob osnovnyx svoystvax etix obrazov, takix, kak
issleduemoy posledovatelnosti v tekushchiy moment vremeni.
further development of the process.
vÿpolnyaet, kak pravilo, neskolko funktsiy. Vo-pervyx, ona predstavlyaet
vospriyatiya, iskusstvennogo raspoznavaniya i obobshcheniya videoinformatsii,
upravleniya slojnymi sistemami, obrabotki rechevyx signalov i t.p. Iskusstvennÿe
neyronnye seti v prakticheskix prilojeniyax, kak pravilo, ispolzuyutsya v kachestve
podsistemy upravleniya ili vÿrabotki resheniy, peredayushchey ispolnitelnÿy signal
drugim podsistemam, imeyushchim inuyu
components (RSA), or other characteristics. When obobshchenii
funktsii neskolkix peremennyx, kotoryy realizuet nelineynuyu funktsiyu
for processing klassifikatsionnyx resheniy. In the
area of forecasting the task set is formulated as
parameters, neobxodimÿe dlya vÿrabotki sootvetstvuyushchego upravlyayushchego
neskolkix peremennyx. Mnojestvo zadach modelirovaniya, identifikatsii, obrabotki
signalov udaetsya sformulirovat v approksimatsionnoy
Ispolzovanie perechislennyx svoystv na fone razvitiya ustroystv so
posledovatelnosti ee predydushchix sostoyaniy. According to the information about the values
nim. She can also play the role of a neuroregulator, zamenyayushchego soboy
8
Machine Translated by Google

---

setey vstrechayutsya namnogo reje, mojno skazat pro raboty Rossieva D.A. i
avtoassotsiativnogo tipa, v kotoroy vzaimozavisimosti oxvatÿvayut tolko
Interesnym predstavlyaetsya obÿedinenie razlichnyx vidov neyronnyx
predobrabotku i interpretatsiyu otvetov obÿchnogo tipa v kajdom iz takix
tipa, s pomoshchyu kotoroy set opredelyaet vzaimosvyazi razlichnyx vektorov. Daje
esli na vxod seti podaetsya vector, iskajennÿy shumom, libo
teacher. Takie kombinatsii poluchili nazvanie "hybridnye seti". First
obucheniya na dve chasti: vnachale obuchaetsya component s samoorganizatsiey, a
tehnologii v rabotax shkoly Galushkina A. I. Poetomu dannaya rabota
zaklyuchaetsya in snijenii vÿchislitelnoy slozhnosti protsessa obucheniya, a
polnyy i ochushchennyy ot shumov isxodnyy vector putem generatsii
klastery po priznakam sovpadeniya svoystv. She plays a role
ix vzaimodeystviya, priveli k sozdaniyu setey raznyx tipov. Kajdyy tip
signalam, otnesennÿm k konkretnÿm clusters, sootvetstvuyushchie im
The literature for the study of neuronal networks is multifaceted
or postprocessor. (Vprochem, eto ne izbavlyaet ot neobxodimosti vÿpolnyat
assotsiativnogo zapominayushchego ustroystva. Here mojno vydelit pamyat
vesov mejneyronnyx svyazey (t.e. obucheniya).
pro izuchenie arhitektury i algorithms raboty neyronnyx setey. Razvernutye teksty po
tehnologicheskim aspektam ispolzovaniya neyronnyx
konkretnye components of the input vector, and the memory of heteroassotsiativnogo
setey mejdu soboy, osobenno setey s samoorganizatsiey i obuchaemyx s
components). Podobnaya setevaya struktura pozvolyaet razdelit fazu
other cars. V rabote Komartsovoy sdelana ssylka na metodiki i
potom - set s uchitelem. Dopolnitelnoe dostoinstvo takogo podkhoda
lishennÿy otdelnyx fragments of dannyx, to set sposobna vosstanovit
component - this set with self-organization on the basis of competition,
funktsioniruyushchaya on a variety of input signals and gruppiruyushchaya ix v
okajetsya poleznoy dlya bolshogo chisla polzovateley.
sootvetstvuyushchego emu vyxodnogo vector.
Different possibilities of integration of neurons between soboy and organization
predobrabotchika (preprotsessora) dannyx. The second component - in the vide set,
obuchaemoy s uchitelem (naprimer, perseptronnoy), sopostavlyaet vxodnÿm
takje v luchshey interpretatsii poluchaemyx rezultatov.
seti, v svoyu ochered, closely connected with sootvetstvuyushchim metodom podbora
In zadachax assotsiatsii neyronnaya set vÿstupaet in roli
zadannÿe znacheniya. It can be related to the block of interpreter answers
presented in the list of literature in the final work. Odnako, bOlshaya chast ee
9
Machine Translated by Google

---

dannyx.