Pruning method of convolutional neural network
The embodiment of the invention relates to the technical field of network compression, in particular to a convolutional neural network pruning method, and the method comprises the steps: inputting sample data into a target model, and obtaining a feature map outputted by each filter of each layer of...
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Patent |
| Sprache: | chi ; eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | |
|---|---|
| container_issue | |
| container_start_page | |
| container_title | |
| container_volume | |
| creator | HAN JUNWEI FENG XIAOXU TANG HENGREN YAO XIWEN GAO CHENYANG CHENG GONG |
| description | The embodiment of the invention relates to the technical field of network compression, in particular to a convolutional neural network pruning method, and the method comprises the steps: inputting sample data into a target model, and obtaining a feature map outputted by each filter of each layer of the target model; traversing the filter, flattening the feature map output by the current filter, and calculating the internal feature activeness of the current filter according to the flattened feature map corresponding to the current filter; calculating the feature difference degree between every two flattened feature maps, and determining the substitutability score of the current filter according to the feature difference degree between the flattened feature map corresponding to the current filter and the flattened feature maps corresponding to other filters; and determining an importance score based on the internal feature activeness and the substitutability score, and pruning the current layer according to a p |
| format | Patent |
| fullrecord | <record><control><sourceid>epo_EVB</sourceid><recordid>TN_cdi_epo_espacenet_CN117829241A</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>CN117829241A</sourcerecordid><originalsourceid>FETCH-epo_espacenet_CN117829241A3</originalsourceid><addsrcrecordid>eNrjZNALKCrNy8xLV8hNLcnIT1HIT1NIzs8ry88pLcnMz0vMUchLLS0CUyXl-UXZPAysaYk5xam8UJqbQdHNNcTZQze1ID8-tbggMTkVqDLe2c_Q0NzCyNLIxNDRmBg1AC_qKxI</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>patent</recordtype></control><display><type>patent</type><title>Pruning method of convolutional neural network</title><source>esp@cenet</source><creator>HAN JUNWEI ; FENG XIAOXU ; TANG HENGREN ; YAO XIWEN ; GAO CHENYANG ; CHENG GONG</creator><creatorcontrib>HAN JUNWEI ; FENG XIAOXU ; TANG HENGREN ; YAO XIWEN ; GAO CHENYANG ; CHENG GONG</creatorcontrib><description>The embodiment of the invention relates to the technical field of network compression, in particular to a convolutional neural network pruning method, and the method comprises the steps: inputting sample data into a target model, and obtaining a feature map outputted by each filter of each layer of the target model; traversing the filter, flattening the feature map output by the current filter, and calculating the internal feature activeness of the current filter according to the flattened feature map corresponding to the current filter; calculating the feature difference degree between every two flattened feature maps, and determining the substitutability score of the current filter according to the feature difference degree between the flattened feature map corresponding to the current filter and the flattened feature maps corresponding to other filters; and determining an importance score based on the internal feature activeness and the substitutability score, and pruning the current layer according to a p</description><language>chi ; eng</language><subject>CALCULATING ; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS ; COMPUTING ; COUNTING ; ELECTRIC DIGITAL DATA PROCESSING ; PHYSICS</subject><creationdate>2024</creationdate><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20240405&DB=EPODOC&CC=CN&NR=117829241A$$EHTML$$P50$$Gepo$$Hfree_for_read</linktohtml><link.rule.ids>230,308,776,881,25542,76289</link.rule.ids><linktorsrc>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20240405&DB=EPODOC&CC=CN&NR=117829241A$$EView_record_in_European_Patent_Office$$FView_record_in_$$GEuropean_Patent_Office$$Hfree_for_read</linktorsrc></links><search><creatorcontrib>HAN JUNWEI</creatorcontrib><creatorcontrib>FENG XIAOXU</creatorcontrib><creatorcontrib>TANG HENGREN</creatorcontrib><creatorcontrib>YAO XIWEN</creatorcontrib><creatorcontrib>GAO CHENYANG</creatorcontrib><creatorcontrib>CHENG GONG</creatorcontrib><title>Pruning method of convolutional neural network</title><description>The embodiment of the invention relates to the technical field of network compression, in particular to a convolutional neural network pruning method, and the method comprises the steps: inputting sample data into a target model, and obtaining a feature map outputted by each filter of each layer of the target model; traversing the filter, flattening the feature map output by the current filter, and calculating the internal feature activeness of the current filter according to the flattened feature map corresponding to the current filter; calculating the feature difference degree between every two flattened feature maps, and determining the substitutability score of the current filter according to the feature difference degree between the flattened feature map corresponding to the current filter and the flattened feature maps corresponding to other filters; and determining an importance score based on the internal feature activeness and the substitutability score, and pruning the current layer according to a p</description><subject>CALCULATING</subject><subject>COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS</subject><subject>COMPUTING</subject><subject>COUNTING</subject><subject>ELECTRIC DIGITAL DATA PROCESSING</subject><subject>PHYSICS</subject><fulltext>true</fulltext><rsrctype>patent</rsrctype><creationdate>2024</creationdate><recordtype>patent</recordtype><sourceid>EVB</sourceid><recordid>eNrjZNALKCrNy8xLV8hNLcnIT1HIT1NIzs8ry88pLcnMz0vMUchLLS0CUyXl-UXZPAysaYk5xam8UJqbQdHNNcTZQze1ID8-tbggMTkVqDLe2c_Q0NzCyNLIxNDRmBg1AC_qKxI</recordid><startdate>20240405</startdate><enddate>20240405</enddate><creator>HAN JUNWEI</creator><creator>FENG XIAOXU</creator><creator>TANG HENGREN</creator><creator>YAO XIWEN</creator><creator>GAO CHENYANG</creator><creator>CHENG GONG</creator><scope>EVB</scope></search><sort><creationdate>20240405</creationdate><title>Pruning method of convolutional neural network</title><author>HAN JUNWEI ; FENG XIAOXU ; TANG HENGREN ; YAO XIWEN ; GAO CHENYANG ; CHENG GONG</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-epo_espacenet_CN117829241A3</frbrgroupid><rsrctype>patents</rsrctype><prefilter>patents</prefilter><language>chi ; eng</language><creationdate>2024</creationdate><topic>CALCULATING</topic><topic>COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS</topic><topic>COMPUTING</topic><topic>COUNTING</topic><topic>ELECTRIC DIGITAL DATA PROCESSING</topic><topic>PHYSICS</topic><toplevel>online_resources</toplevel><creatorcontrib>HAN JUNWEI</creatorcontrib><creatorcontrib>FENG XIAOXU</creatorcontrib><creatorcontrib>TANG HENGREN</creatorcontrib><creatorcontrib>YAO XIWEN</creatorcontrib><creatorcontrib>GAO CHENYANG</creatorcontrib><creatorcontrib>CHENG GONG</creatorcontrib><collection>esp@cenet</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>HAN JUNWEI</au><au>FENG XIAOXU</au><au>TANG HENGREN</au><au>YAO XIWEN</au><au>GAO CHENYANG</au><au>CHENG GONG</au><format>patent</format><genre>patent</genre><ristype>GEN</ristype><title>Pruning method of convolutional neural network</title><date>2024-04-05</date><risdate>2024</risdate><abstract>The embodiment of the invention relates to the technical field of network compression, in particular to a convolutional neural network pruning method, and the method comprises the steps: inputting sample data into a target model, and obtaining a feature map outputted by each filter of each layer of the target model; traversing the filter, flattening the feature map output by the current filter, and calculating the internal feature activeness of the current filter according to the flattened feature map corresponding to the current filter; calculating the feature difference degree between every two flattened feature maps, and determining the substitutability score of the current filter according to the feature difference degree between the flattened feature map corresponding to the current filter and the flattened feature maps corresponding to other filters; and determining an importance score based on the internal feature activeness and the substitutability score, and pruning the current layer according to a p</abstract><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | |
| ispartof | |
| issn | |
| language | chi ; eng |
| recordid | cdi_epo_espacenet_CN117829241A |
| source | esp@cenet |
| subjects | CALCULATING COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS COMPUTING COUNTING ELECTRIC DIGITAL DATA PROCESSING PHYSICS |
| title | Pruning method of convolutional neural network |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-08T11%3A23%3A21IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-epo_EVB&rft_val_fmt=info:ofi/fmt:kev:mtx:patent&rft.genre=patent&rft.au=HAN%20JUNWEI&rft.date=2024-04-05&rft_id=info:doi/&rft_dat=%3Cepo_EVB%3ECN117829241A%3C/epo_EVB%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |