[go: up one dir, main page]

GB2624564B - Identifying one or more quantisation parameters for quantising values to be processed by a neural network - Google Patents

Identifying one or more quantisation parameters for quantising values to be processed by a neural network Download PDF

Info

Publication number
GB2624564B
GB2624564B GB2402394.7A GB202402394A GB2624564B GB 2624564 B GB2624564 B GB 2624564B GB 202402394 A GB202402394 A GB 202402394A GB 2624564 B GB2624564 B GB 2624564B
Authority
GB
United Kingdom
Prior art keywords
quantising
identifying
processed
values
neural network
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
GB2402394.7A
Other versions
GB2624564A (en
Inventor
Cséfalvay Szabolcs
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Imagination Technologies Ltd
Original Assignee
Imagination Technologies Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Imagination Technologies Ltd filed Critical Imagination Technologies Ltd
Priority to GB2402394.7A priority Critical patent/GB2624564B/en
Publication of GB2624564A publication Critical patent/GB2624564A/en
Application granted granted Critical
Publication of GB2624564B publication Critical patent/GB2624564B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0495—Quantised networks; Sparse networks; Compressed networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
    • G06N3/063—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Neurology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Error Detection And Correction (AREA)
  • Compression, Expansion, Code Conversion, And Decoders (AREA)
GB2402394.7A 2022-06-30 2022-06-30 Identifying one or more quantisation parameters for quantising values to be processed by a neural network Active GB2624564B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
GB2402394.7A GB2624564B (en) 2022-06-30 2022-06-30 Identifying one or more quantisation parameters for quantising values to be processed by a neural network

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
GB2402394.7A GB2624564B (en) 2022-06-30 2022-06-30 Identifying one or more quantisation parameters for quantising values to be processed by a neural network
GB2209612.7A GB2620172B (en) 2022-06-30 2022-06-30 Identifying one or more quantisation parameters for quantising values to be processed by a neural network

Publications (2)

Publication Number Publication Date
GB2624564A GB2624564A (en) 2024-05-22
GB2624564B true GB2624564B (en) 2025-01-01

Family

ID=82802604

Family Applications (2)

Application Number Title Priority Date Filing Date
GB2209612.7A Active GB2620172B (en) 2022-06-30 2022-06-30 Identifying one or more quantisation parameters for quantising values to be processed by a neural network
GB2402394.7A Active GB2624564B (en) 2022-06-30 2022-06-30 Identifying one or more quantisation parameters for quantising values to be processed by a neural network

Family Applications Before (1)

Application Number Title Priority Date Filing Date
GB2209612.7A Active GB2620172B (en) 2022-06-30 2022-06-30 Identifying one or more quantisation parameters for quantising values to be processed by a neural network

Country Status (1)

Country Link
GB (2) GB2620172B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119312851B (en) * 2024-11-25 2025-10-24 四川大学 A low-bitwidth adaptive quantization method for convolutional neural networks for image classification

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2580171A (en) * 2018-12-21 2020-07-15 Imagination Tech Ltd Methods and systems for selecting quantisation parameters for deep neural networks using back-propagation

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11106973B2 (en) * 2016-03-16 2021-08-31 Hong Kong Applied Science and Technology Research Institute Company Limited Method and system for bit-depth reduction in artificial neural networks

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2580171A (en) * 2018-12-21 2020-07-15 Imagination Tech Ltd Methods and systems for selecting quantisation parameters for deep neural networks using back-propagation

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
WANG et al, August 2020, "Differentiable Joint Pruning and Quantization for Hardware Efficiency", 16th European Conference on Computer Vision, Lecture Notes in Computer Science, Springer. *

Also Published As

Publication number Publication date
GB2624564A (en) 2024-05-22
GB2620172A (en) 2024-01-03
GB2620172B (en) 2024-05-29
GB202209612D0 (en) 2022-08-17

Similar Documents

Publication Publication Date Title
GB202209080D0 (en) Pretraining framework for neural networks
GB2580171B (en) Methods and systems for selecting quantisation parameters for deep neural networks using back-propagation
GB2620172B (en) Identifying one or more quantisation parameters for quantising values to be processed by a neural network
GB202112184D0 (en) Techniques for pruning neural networks
GB202012998D0 (en) Waveguide and method for fabricating a waveguide master grating tool
GB202118140D0 (en) Techniques for optimizing neural networks
EP4172862A4 (en) Object recognition neural network for amodal center prediction
GB202203553D0 (en) Pruning neural networks
GB2584751B (en) Resin composition, inorganic filler, direct-current power cable, and method for manufacturing direct-current power cable
DE602008006778D1 (en) PROCESS FOR REMOVING ZEARALENONE IN A FEEDING PRODUCT USING LACCASE
PL4177011T3 (en) Axe and a method for manufacturing an axe
IL288021B1 (en) Cluster-connected neural network
GB2602294B (en) Compressing a set of coefficients for subsequent use in a neural network
PL3936473T3 (en) Method for producing sulfide solid electrolyte
GB202103967D0 (en) Method for improved binarized neural networks
IL289360A (en) Strain and method for producing rosmarinic acid
LT3784036T (en) Method for influencing arthropods
MX2017003649A (en) Component which is at least partly made of a layer structure, and method for producing same.
GB202216948D0 (en) Identifying onr ot more quantising values to be processed by a nerual network
PH12021550753A1 (en) A method for treating effluent produced from palm oil milling process
GB202306770D0 (en) Action pruning by logical neural network
GB2614237B (en) Method of analysing a spectral peak using a neural network
GB202002279D0 (en) Multiple Spectra imput Neural Network for Prosthesis
GB202009303D0 (en) Method for processing an additively manufactured investment casting model
TWI914412B (en) Multiple neural network models for filtering during video coding

Legal Events

Date Code Title Description
732E Amendments to the register in respect of changes of name or changes affecting rights (sect. 32/1977)

Free format text: REGISTERED BETWEEN 20240822 AND 20240828