Fixed-point roundoff error analysis of large feedforward neural networks
Digital implementations of neural nets must consider finite wordlength effects. For large sized nets, it is particularly important to investigate the roundoff errors in order to realize low-cost hardware implementations while satisfying precision constraints. This paper presents output error express...
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Tagungsbericht |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | 1950 vol.2 |
|---|---|
| container_issue | |
| container_start_page | 1947 |
| container_title | |
| container_volume | 2 |
| creator | Choi, H. Burleson, W.P. Phatak, D.S. |
| description | Digital implementations of neural nets must consider finite wordlength effects. For large sized nets, it is particularly important to investigate the roundoff errors in order to realize low-cost hardware implementations while satisfying precision constraints. This paper presents output error expressions for a large feedforward neural net, which are based on statistical error analysis. Weight quantization errors as well as arithmetic errors due to rounding of multiplier output and sigmoid output are modeled. The results indicate that for equal wordlengths, multiplier roundoff errors exceed weight quantization errors by about an order of magnitude. |
| doi_str_mv | 10.1109/IJCNN.1993.717037 |
| format | Conference Proceeding |
| fullrecord | <record><control><sourceid>proquest_6IE</sourceid><recordid>TN_cdi_proquest_miscellaneous_26053266</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>717037</ieee_id><sourcerecordid>26053266</sourcerecordid><originalsourceid>FETCH-LOGICAL-i203t-2a59d69c57c083dcf82ecd946ad7fd476d9ff40fcb74224daee8e40e4d53872a3</originalsourceid><addsrcrecordid>eNotkL1OwzAYRS0hJKD0AWDyxJbgv9jxiCJKi6qywByZ-DMyuHGwE5W-fSOVu5zl6AwXoTtKSkqJfty8NrtdSbXmpaKKcHWBboiqCaeCUXaFljl_k3lCVEyra7Re-T-wxRB9P-IUp95G5zCkFBM2vQnH7DOODgeTvgA7AOtiOphkcQ9TMmHGeIjpJ9-iS2dChuU_F-hj9fzerIvt28umedoWnhE-FsxU2krdVaojNbedqxl0VgtprHJWKGm1c4K47lMJxoQ1ADUIAsJWvFbM8AV6OHeHFH8nyGO797mDEEwPccotk6TiTMpZvD-LHgDaIfm9Scf2fAo_AWOZWT0</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype><pqid>26053266</pqid></control><display><type>conference_proceeding</type><title>Fixed-point roundoff error analysis of large feedforward neural networks</title><source>IEEE Electronic Library (IEL) Conference Proceedings</source><creator>Choi, H. ; Burleson, W.P. ; Phatak, D.S.</creator><creatorcontrib>Choi, H. ; Burleson, W.P. ; Phatak, D.S.</creatorcontrib><description>Digital implementations of neural nets must consider finite wordlength effects. For large sized nets, it is particularly important to investigate the roundoff errors in order to realize low-cost hardware implementations while satisfying precision constraints. This paper presents output error expressions for a large feedforward neural net, which are based on statistical error analysis. Weight quantization errors as well as arithmetic errors due to rounding of multiplier output and sigmoid output are modeled. The results indicate that for equal wordlengths, multiplier roundoff errors exceed weight quantization errors by about an order of magnitude.</description><identifier>ISBN: 0780314212</identifier><identifier>ISBN: 9780780314214</identifier><identifier>DOI: 10.1109/IJCNN.1993.717037</identifier><language>eng</language><publisher>IEEE</publisher><subject>Aggregates ; Arithmetic ; Error analysis ; Feedforward neural networks ; Hardware ; Multi-layer neural network ; Neural networks ; Nonhomogeneous media ; Quantization ; Roundoff errors</subject><ispartof>International Joint Conference on Neural Networks, Nagoya, 1993, 1993, Vol.2, p.1947-1950 vol.2</ispartof><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/717037$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,314,780,784,789,790,2058,4050,4051,27924,27925,54920</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/717037$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Choi, H.</creatorcontrib><creatorcontrib>Burleson, W.P.</creatorcontrib><creatorcontrib>Phatak, D.S.</creatorcontrib><title>Fixed-point roundoff error analysis of large feedforward neural networks</title><title>International Joint Conference on Neural Networks, Nagoya, 1993</title><addtitle>IJCNN</addtitle><description>Digital implementations of neural nets must consider finite wordlength effects. For large sized nets, it is particularly important to investigate the roundoff errors in order to realize low-cost hardware implementations while satisfying precision constraints. This paper presents output error expressions for a large feedforward neural net, which are based on statistical error analysis. Weight quantization errors as well as arithmetic errors due to rounding of multiplier output and sigmoid output are modeled. The results indicate that for equal wordlengths, multiplier roundoff errors exceed weight quantization errors by about an order of magnitude.</description><subject>Aggregates</subject><subject>Arithmetic</subject><subject>Error analysis</subject><subject>Feedforward neural networks</subject><subject>Hardware</subject><subject>Multi-layer neural network</subject><subject>Neural networks</subject><subject>Nonhomogeneous media</subject><subject>Quantization</subject><subject>Roundoff errors</subject><isbn>0780314212</isbn><isbn>9780780314214</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>1993</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><sourceid>RIE</sourceid><recordid>eNotkL1OwzAYRS0hJKD0AWDyxJbgv9jxiCJKi6qywByZ-DMyuHGwE5W-fSOVu5zl6AwXoTtKSkqJfty8NrtdSbXmpaKKcHWBboiqCaeCUXaFljl_k3lCVEyra7Re-T-wxRB9P-IUp95G5zCkFBM2vQnH7DOODgeTvgA7AOtiOphkcQ9TMmHGeIjpJ9-iS2dChuU_F-hj9fzerIvt28umedoWnhE-FsxU2krdVaojNbedqxl0VgtprHJWKGm1c4K47lMJxoQ1ADUIAsJWvFbM8AV6OHeHFH8nyGO797mDEEwPccotk6TiTMpZvD-LHgDaIfm9Scf2fAo_AWOZWT0</recordid><startdate>1993</startdate><enddate>1993</enddate><creator>Choi, H.</creator><creator>Burleson, W.P.</creator><creator>Phatak, D.S.</creator><general>IEEE</general><scope>6IE</scope><scope>6IH</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIO</scope><scope>7SC</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope></search><sort><creationdate>1993</creationdate><title>Fixed-point roundoff error analysis of large feedforward neural networks</title><author>Choi, H. ; Burleson, W.P. ; Phatak, D.S.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-i203t-2a59d69c57c083dcf82ecd946ad7fd476d9ff40fcb74224daee8e40e4d53872a3</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>1993</creationdate><topic>Aggregates</topic><topic>Arithmetic</topic><topic>Error analysis</topic><topic>Feedforward neural networks</topic><topic>Hardware</topic><topic>Multi-layer neural network</topic><topic>Neural networks</topic><topic>Nonhomogeneous media</topic><topic>Quantization</topic><topic>Roundoff errors</topic><toplevel>online_resources</toplevel><creatorcontrib>Choi, H.</creatorcontrib><creatorcontrib>Burleson, W.P.</creatorcontrib><creatorcontrib>Phatak, D.S.</creatorcontrib><collection>IEEE Electronic Library (IEL) Conference Proceedings</collection><collection>IEEE Proceedings Order Plan (POP) 1998-present by volume</collection><collection>IEEE Xplore All Conference Proceedings</collection><collection>IEEE Xplore</collection><collection>IEEE Proceedings Order Plans (POP) 1998-present</collection><collection>Computer and Information Systems Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Choi, H.</au><au>Burleson, W.P.</au><au>Phatak, D.S.</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>Fixed-point roundoff error analysis of large feedforward neural networks</atitle><btitle>International Joint Conference on Neural Networks, Nagoya, 1993</btitle><stitle>IJCNN</stitle><date>1993</date><risdate>1993</risdate><volume>2</volume><spage>1947</spage><epage>1950 vol.2</epage><pages>1947-1950 vol.2</pages><isbn>0780314212</isbn><isbn>9780780314214</isbn><abstract>Digital implementations of neural nets must consider finite wordlength effects. For large sized nets, it is particularly important to investigate the roundoff errors in order to realize low-cost hardware implementations while satisfying precision constraints. This paper presents output error expressions for a large feedforward neural net, which are based on statistical error analysis. Weight quantization errors as well as arithmetic errors due to rounding of multiplier output and sigmoid output are modeled. The results indicate that for equal wordlengths, multiplier roundoff errors exceed weight quantization errors by about an order of magnitude.</abstract><pub>IEEE</pub><doi>10.1109/IJCNN.1993.717037</doi></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | ISBN: 0780314212 |
| ispartof | International Joint Conference on Neural Networks, Nagoya, 1993, 1993, Vol.2, p.1947-1950 vol.2 |
| issn | |
| language | eng |
| recordid | cdi_proquest_miscellaneous_26053266 |
| source | IEEE Electronic Library (IEL) Conference Proceedings |
| subjects | Aggregates Arithmetic Error analysis Feedforward neural networks Hardware Multi-layer neural network Neural networks Nonhomogeneous media Quantization Roundoff errors |
| title | Fixed-point roundoff error analysis of large feedforward neural networks |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-07T20%3A03%3A15IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_6IE&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=proceeding&rft.atitle=Fixed-point%20roundoff%20error%20analysis%20of%20large%20feedforward%20neural%20networks&rft.btitle=International%20Joint%20Conference%20on%20Neural%20Networks,%20Nagoya,%201993&rft.au=Choi,%20H.&rft.date=1993&rft.volume=2&rft.spage=1947&rft.epage=1950%20vol.2&rft.pages=1947-1950%20vol.2&rft.isbn=0780314212&rft.isbn_list=9780780314214&rft_id=info:doi/10.1109/IJCNN.1993.717037&rft_dat=%3Cproquest_6IE%3E26053266%3C/proquest_6IE%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=26053266&rft_id=info:pmid/&rft_ieee_id=717037&rfr_iscdi=true |