Processing

Please wait...

Settings

Settings

Goto Application

1. WO2020117989 - EXECUTION OF TRAINED NEURAL NETWORKS USING A DATABASE SYSTEM

Publication Number WO/2020/117989
Publication Date 11.06.2020
International Application No. PCT/US2019/064550
International Filing Date 04.12.2019
IPC
G06K 9/00 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
9Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
G06N 3/04 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
04Architecture, e.g. interconnection topology
G06N 3/08 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
08Learning methods
CPC
G06F 16/24578
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
FELECTRIC DIGITAL DATA PROCESSING
16Information retrieval; Database structures therefor; File system structures therefor
20of structured data, e.g. relational data
24Querying
245Query processing
2457with adaptation to user needs
24578using ranking
G06F 16/2458
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
FELECTRIC DIGITAL DATA PROCESSING
16Information retrieval; Database structures therefor; File system structures therefor
20of structured data, e.g. relational data
24Querying
245Query processing
2458Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
G06N 3/0427
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
04Architectures, e.g. interconnection topology
0427in combination with an expert system
G06N 3/0481
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
04Architectures, e.g. interconnection topology
0481Non-linear activation functions, e.g. sigmoids, thresholds
G06N 3/063
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
06Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
063using electronic means
G06N 3/08
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
08Learning methods
Applicants
  • SHAPE SECURITY, INC. [US]/[US]
Inventors
  • ZHANG, Bei
  • SHAH, Samir
  • MILLER, Kenton
Agents
  • LEINBERG, Gunnar G.
Priority Data
16/211,13805.12.2018US
Publication Language English (EN)
Filing Language English (EN)
Designated States
Title
(EN) EXECUTION OF TRAINED NEURAL NETWORKS USING A DATABASE SYSTEM
(FR) EXÉCUTION DE RÉSEAUX NEURONAUX ENTRAÎNÉS UTILISANT UN SYSTÈME DE BASE DE DONNÉES
Abstract
(EN)
In an embodiment, a computer-implemented method for efficient execution of a trained neural network using a database system, the trained neural network comprising a plurality of layers and programmed at each of the layers to execute an affine transformation of an activation function and an input value, comprises: for a particular layer of the trained neural network, dividing the affine transformation into a plurality of transformation pieces; executing each of the transformation pieces to result in computed pieces and writing the computed pieces to a first database table; using one or more database queries, combining the computed pieces and applying the activation function to generate a set of output data; writing the output data to one of a plurality of different second database tables that respectively correspond to the layers; repeating the dividing, executing, combining, applying and writing for all layers of the trained neural network.
(FR)
Dans un mode de réalisation, l'invention concerne un procédé mis en œuvre par ordinateur pour l'exécution efficace d'un réseau neuronal entraîné utilisant un système de base de données, le réseau neuronal entraîné comprenant une pluralité de couches et étant programmé au niveau de chacune des couches pour exécuter une transformation affine d'une fonction d'activation et d'une valeur d'entrée, lequel procédé comprend : pour une couche particulière du réseau neuronal entraîné, la division de la transformation affine en une pluralité de morceaux de transformation ; l'exécution de chacun des morceaux de transformation pour obtenir des morceaux calculés et l'écriture des morceaux calculés dans une première table de base de données ; à l'aide d'une ou de plusieurs interrogations de base de données, la combinaison des morceaux calculés et l'application de la fonction d'activation pour générer un ensemble de données de sortie ; l'écriture des données de sortie dans une table parmi une pluralité de secondes tables de base de données différentes qui correspondent respectivement aux couches ; la répétition de la division, de l'exécution, de la combinaison, de l'application et de l'écriture pour toutes les couches du réseau neuronal entraîné.
Also published as
Latest bibliographic data on file with the International Bureau