NEURAL NETWORK CIRCUIT AND A SYSTEM COMPRISING ONE OR A PLURALITY OF NEURAL NETWORK CIRCUITS

NEURAL NETWORK CIRCUIT AND A SYSTEM COMPRISING ONE OR A PLURALITY OF NEURAL NETWORK CIRCUITS

An RCE/RBF neural network circuit (1) that comprises an input layer (10), an hidden layer (11) and an output layer (12), the hidden layer having a plurality of neurons (110) for processing an input vector (IN) with a variable number of components (C0, C1, ..) and variable component depth that is inp...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chavade, Jacques, Málnási-Csizmadia, Örs, Bailly, Bruno, Petrova, Boriana
Format: Patent
Sprache:eng ; fre ; ger
Schlagworte:
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
ELECTRIC DIGITAL DATA PROCESSING
PHYSICS
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:An RCE/RBF neural network circuit (1) that comprises an input layer (10), an hidden layer (11) and an output layer (12), the hidden layer having a plurality of neurons (110) for processing an input vector (IN) with a variable number of components (C0, C1, ..) and variable component depth that is input into the input layer (11) for comparison against reference vectors (REF) stored in said neurons,wherein each neuron (110) is associated with one memory block (3), each memory block storing at least one said reference vector (REF);* wherein each neuron (110) comprises a distance calculator (5) for calculating a distance between the input vector (IN) and the said reference vector (REF);* wherein each memory block (3) is shared by a plurality of neurons (110).