Stagewise Newton, differential dynamic programming, and neighboring optimum control for neural-network learning
The theory of optimal control is applied to multi-stage (i.e., multiple-layered) neural-network (NN) learning for developing efficient second-order algorithms, expressed in NN notation. In particular, we compare differential dynamic programming, neighboring optimum control, and stagewise Newton meth...
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 | 1336 vol. 2 |
|---|---|
| container_issue | |
| container_start_page | 1331 |
| container_title | |
| container_volume | |
| creator | Mizutani, E. Dreyfus, S.E. |
| description | The theory of optimal control is applied to multi-stage (i.e., multiple-layered) neural-network (NN) learning for developing efficient second-order algorithms, expressed in NN notation. In particular, we compare differential dynamic programming, neighboring optimum control, and stagewise Newton methods. Understanding their strengths and weaknesses would prove useful in pursuit of an effective intermediate step between the steepest descent and the Newton directions, arising in supervised NN-learning as well as reinforcement learning with function approximators. |
| doi_str_mv | 10.1109/ACC.2005.1470149 |
| format | Conference Proceeding |
| fullrecord | <record><control><sourceid>proquest_6IE</sourceid><recordid>TN_cdi_proquest_miscellaneous_29044553</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>1470149</ieee_id><sourcerecordid>29044553</sourcerecordid><originalsourceid>FETCH-LOGICAL-i206t-22823b07bb6314cba1206ea99b12c9e407f2b35c0b271353430a097601899dd63</originalsourceid><addsrcrecordid>eNo1kEtrwzAQhNUXNElzL_SiU09xupJsy3sMoS8I7aHt2cj22lVrS6nsEPLva0gLAwszH8swjF0LWAoBeLdar5cSIFmKWIOI8YTNUWcwSiEg6lM2kUpnUZKl4oxN_4MMz9kEdKwikQq8ZNO-_wIQiClMmH8bTEN72xN_of3g3YJXtq4pkBusaXl1cKazJd8G3wTTddY1C25cxR3Z5rPwYTS43w6223W89G4IvuW1D2O-C6aNHA17H755Sya4kb1iF7Vpe5r_3Rn7eLh_Xz9Fm9fH5_VqE1kJ6RBJmUlVgC6KVIm4LIwYbTKIhZAlUgy6loVKSiikFipRsQIDqFMQGWJVpWrGbo9_x-I_O-qHvLN9SW1rHPldn0uEOE4SNYI3R9ASUb4NtjPhkP8trH4BJ1hryg</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype><pqid>29044553</pqid></control><display><type>conference_proceeding</type><title>Stagewise Newton, differential dynamic programming, and neighboring optimum control for neural-network learning</title><source>IEEE Electronic Library (IEL) Conference Proceedings</source><creator>Mizutani, E. ; Dreyfus, S.E.</creator><creatorcontrib>Mizutani, E. ; Dreyfus, S.E.</creatorcontrib><description>The theory of optimal control is applied to multi-stage (i.e., multiple-layered) neural-network (NN) learning for developing efficient second-order algorithms, expressed in NN notation. In particular, we compare differential dynamic programming, neighboring optimum control, and stagewise Newton methods. Understanding their strengths and weaknesses would prove useful in pursuit of an effective intermediate step between the steepest descent and the Newton directions, arising in supervised NN-learning as well as reinforcement learning with function approximators.</description><identifier>ISSN: 0743-1619</identifier><identifier>ISBN: 0780390989</identifier><identifier>ISBN: 9780780390980</identifier><identifier>ISBN: 9780780390997</identifier><identifier>ISBN: 0780390997</identifier><identifier>EISSN: 2378-5861</identifier><identifier>EISBN: 9780780390997</identifier><identifier>EISBN: 0780390997</identifier><identifier>DOI: 10.1109/ACC.2005.1470149</identifier><language>eng</language><publisher>IEEE</publisher><subject>Boundary conditions ; Costs ; Difference equations ; Dynamic programming ; Lagrangian functions ; Learning ; Neural networks ; Newton method ; Optimal control ; Performance analysis</subject><ispartof>2005 American Control Conference, 2005, p.1331-1336 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/1470149$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,776,780,785,786,2052,4036,4037,27904,54898</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/1470149$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Mizutani, E.</creatorcontrib><creatorcontrib>Dreyfus, S.E.</creatorcontrib><title>Stagewise Newton, differential dynamic programming, and neighboring optimum control for neural-network learning</title><title>2005 American Control Conference</title><addtitle>ACC</addtitle><description>The theory of optimal control is applied to multi-stage (i.e., multiple-layered) neural-network (NN) learning for developing efficient second-order algorithms, expressed in NN notation. In particular, we compare differential dynamic programming, neighboring optimum control, and stagewise Newton methods. Understanding their strengths and weaknesses would prove useful in pursuit of an effective intermediate step between the steepest descent and the Newton directions, arising in supervised NN-learning as well as reinforcement learning with function approximators.</description><subject>Boundary conditions</subject><subject>Costs</subject><subject>Difference equations</subject><subject>Dynamic programming</subject><subject>Lagrangian functions</subject><subject>Learning</subject><subject>Neural networks</subject><subject>Newton method</subject><subject>Optimal control</subject><subject>Performance analysis</subject><issn>0743-1619</issn><issn>2378-5861</issn><isbn>0780390989</isbn><isbn>9780780390980</isbn><isbn>9780780390997</isbn><isbn>0780390997</isbn><isbn>9780780390997</isbn><isbn>0780390997</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>2005</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><sourceid>RIE</sourceid><recordid>eNo1kEtrwzAQhNUXNElzL_SiU09xupJsy3sMoS8I7aHt2cj22lVrS6nsEPLva0gLAwszH8swjF0LWAoBeLdar5cSIFmKWIOI8YTNUWcwSiEg6lM2kUpnUZKl4oxN_4MMz9kEdKwikQq8ZNO-_wIQiClMmH8bTEN72xN_of3g3YJXtq4pkBusaXl1cKazJd8G3wTTddY1C25cxR3Z5rPwYTS43w6223W89G4IvuW1D2O-C6aNHA17H755Sya4kb1iF7Vpe5r_3Rn7eLh_Xz9Fm9fH5_VqE1kJ6RBJmUlVgC6KVIm4LIwYbTKIhZAlUgy6loVKSiikFipRsQIDqFMQGWJVpWrGbo9_x-I_O-qHvLN9SW1rHPldn0uEOE4SNYI3R9ASUb4NtjPhkP8trH4BJ1hryg</recordid><startdate>2005</startdate><enddate>2005</enddate><creator>Mizutani, E.</creator><creator>Dreyfus, S.E.</creator><general>IEEE</general><scope>6IE</scope><scope>6IH</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIO</scope><scope>7SC</scope><scope>7SP</scope><scope>7TB</scope><scope>8FD</scope><scope>FR3</scope><scope>H8D</scope><scope>JQ2</scope><scope>KR7</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope></search><sort><creationdate>2005</creationdate><title>Stagewise Newton, differential dynamic programming, and neighboring optimum control for neural-network learning</title><author>Mizutani, E. ; Dreyfus, S.E.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-i206t-22823b07bb6314cba1206ea99b12c9e407f2b35c0b271353430a097601899dd63</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>2005</creationdate><topic>Boundary conditions</topic><topic>Costs</topic><topic>Difference equations</topic><topic>Dynamic programming</topic><topic>Lagrangian functions</topic><topic>Learning</topic><topic>Neural networks</topic><topic>Newton method</topic><topic>Optimal control</topic><topic>Performance analysis</topic><toplevel>online_resources</toplevel><creatorcontrib>Mizutani, E.</creatorcontrib><creatorcontrib>Dreyfus, S.E.</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 Electronic Library (IEL)</collection><collection>IEEE Proceedings Order Plans (POP) 1998-present</collection><collection>Computer and Information Systems Abstracts</collection><collection>Electronics & Communications Abstracts</collection><collection>Mechanical & Transportation Engineering Abstracts</collection><collection>Technology Research Database</collection><collection>Engineering Research Database</collection><collection>Aerospace Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Civil Engineering Abstracts</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>Mizutani, E.</au><au>Dreyfus, S.E.</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>Stagewise Newton, differential dynamic programming, and neighboring optimum control for neural-network learning</atitle><btitle>2005 American Control Conference</btitle><stitle>ACC</stitle><date>2005</date><risdate>2005</risdate><spage>1331</spage><epage>1336 vol. 2</epage><pages>1331-1336 vol. 2</pages><issn>0743-1619</issn><eissn>2378-5861</eissn><isbn>0780390989</isbn><isbn>9780780390980</isbn><isbn>9780780390997</isbn><isbn>0780390997</isbn><eisbn>9780780390997</eisbn><eisbn>0780390997</eisbn><abstract>The theory of optimal control is applied to multi-stage (i.e., multiple-layered) neural-network (NN) learning for developing efficient second-order algorithms, expressed in NN notation. In particular, we compare differential dynamic programming, neighboring optimum control, and stagewise Newton methods. Understanding their strengths and weaknesses would prove useful in pursuit of an effective intermediate step between the steepest descent and the Newton directions, arising in supervised NN-learning as well as reinforcement learning with function approximators.</abstract><pub>IEEE</pub><doi>10.1109/ACC.2005.1470149</doi></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | ISSN: 0743-1619 |
| ispartof | 2005 American Control Conference, 2005, p.1331-1336 vol. 2 |
| issn | 0743-1619 2378-5861 |
| language | eng |
| recordid | cdi_proquest_miscellaneous_29044553 |
| source | IEEE Electronic Library (IEL) Conference Proceedings |
| subjects | Boundary conditions Costs Difference equations Dynamic programming Lagrangian functions Learning Neural networks Newton method Optimal control Performance analysis |
| title | Stagewise Newton, differential dynamic programming, and neighboring optimum control for neural-network learning |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-27T18%3A59%3A27IST&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=Stagewise%20Newton,%20differential%20dynamic%20programming,%20and%20neighboring%20optimum%20control%20for%20neural-network%20learning&rft.btitle=2005%20American%20Control%20Conference&rft.au=Mizutani,%20E.&rft.date=2005&rft.spage=1331&rft.epage=1336%20vol.%202&rft.pages=1331-1336%20vol.%202&rft.issn=0743-1619&rft.eissn=2378-5861&rft.isbn=0780390989&rft.isbn_list=9780780390980&rft.isbn_list=9780780390997&rft.isbn_list=0780390997&rft_id=info:doi/10.1109/ACC.2005.1470149&rft_dat=%3Cproquest_6IE%3E29044553%3C/proquest_6IE%3E%3Curl%3E%3C/url%3E&rft.eisbn=9780780390997&rft.eisbn_list=0780390997&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=29044553&rft_id=info:pmid/&rft_ieee_id=1470149&rfr_iscdi=true |