Processing

Please wait...

Settings

Settings

Goto Application

1. WO2016182671 - FIXED POINT NEURAL NETWORK BASED ON FLOATING POINT NEURAL NETWORK QUANTIZATION

Publication Number WO/2016/182671
Publication Date 17.11.2016
International Application No. PCT/US2016/027589
International Filing Date 14.04.2016
Chapter 2 Demand Filed 06.03.2017
IPC
G06N 3/10 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
10Simulation on general purpose computers
G06N 3/06 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR 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
CPC
G06K 9/4628
GPHYSICS
06COMPUTING; CALCULATING; 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
36Image preprocessing, i.e. processing the image information without deciding about the identity of the image
46Extraction of features or characteristics of the image
4604Detecting partial patterns, e.g. edges or contours, or configurations, e.g. loops, corners, strokes, intersections
4609by matching or filtering
4619Biologically-inspired filters, e.g. receptive fields
4623with interaction between the responses of different filters
4628Integrating the filters into a hierarchical structure
G06N 3/04
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
G06N 3/06
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
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
G06N 3/10
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
10Simulation on general purpose computers
Applicants
  • QUALCOMM INCORPORATED [US]/[US]
Inventors
  • LIN, Dexu
  • ANNAPUREDDY, Venkata Sreekanta Reddy
  • HOWARD, David Edward
  • JULIAN, David Jonathan
  • MAJUMDAR, Somdeb
  • BELL, II, William Richard
Agents
  • LENKIN, Alan M.
Priority Data
14/920,09922.10.2015US
62/159,07908.05.2015US
Publication Language English (EN)
Filing Language English (EN)
Designated States
Title
(EN) FIXED POINT NEURAL NETWORK BASED ON FLOATING POINT NEURAL NETWORK QUANTIZATION
(FR) RÉSEAU NEURONAL À VIRGULE FIXE REPOSANT SUR UNE QUANTIFICATION DE RÉSEAU NEURONAL À VIRGULE FLOTTANTE
Abstract
(EN)
A method of quantizing a floating point machine learning network to obtain a fixed point machine learning network using a quantizer may include selecting at least one moment of an input distribution of the floating point machine learning network. The method may also include determining quantizer parameters for quantizing values of the floating point machine learning network based at least in part on the at least one selected moment of the input distribution of the floating point machine learning network to obtain corresponding values of the fixed point machine learning network.
(FR)
Un procédé de quantification d'un réseau d'apprentissage de machine à virgule flottante en vue d'obtenir un réseau d'apprentissage de machine à virgule fixe à l'aide d'un quantificateur peut consister à sélectionner au moins un moment d'une distribution d'entrée du réseau d'apprentissage de machine à virgule flottante. Le procédé peut également consister à déterminer des paramètres de quantificateur en vue de quantifier des valeurs du réseau d'apprentissage de machine à virgule flottante sur la base au moins en partie dudit moment sélectionné de la distribution d'entrée du réseau d'apprentissage de machine à virgule flottante de sorte à obtenir des valeurs correspondantes du réseau d'apprentissage de machine à virgule fixe.
Latest bibliographic data on file with the International Bureau