{"id":"d3cdcfad875e1227da74ec39f43b4b7877f80452b9922fbe07dcb48587ec0dd0","title":"arXiv cs.AI: Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing","argument":"The authors propose Spectral Neural Networks (SNN), a matrix network architecture in which each neuron applies nonlinear activation to an input matrix’s singular values. They describe this as representing matrices through singular values and singular vectors rather than entries. In a matrix sensing model trained under gradient flow—a continuous approximation of discrete gradient descent—the authors give a theoretical result that squared loss decays exponentially fast and training implicitly selects an approximate solution minimizing nuclear norm. They also report empirical investigations of SNN for matrix sensing and say the architecture has potential across a broader class of matrix learning scenarios. The paper connects the architecture’s training dynamics to a particular approximate solution, while its empirical investigation concerns matrix sensing. Its abstract does not establish how far those results extend beyond that setting.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The stated theoretical results concern an SNN parameterized matrix sensing model under gradient flow; the abstract does not specify how broadly the empirical investigations extend. The authors describe wider applicability as potential, not as a demonstrated result across a broad class of matrix learning scenarios.","speaker":{"name":"Chu, Hong T. M., Dong, Xiaoyu, Ghosh, Subhro, Lam, Chi Thanh, Mukherjee, Soumendu Sundar","roleAtSource":"Article author"},"source":{"title":"Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2402.17595","publishedAt":"2024-02-27","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose Spectral Neural Networks, where each neuron applies nonlinear activation to the singular values of an input matrix.","excerpt":"each neuron applies nonlinear activation to the singular values of the input matrix","url":"https://arxiv.org/abs/2402.17595","relationship":"Source-stated"},{"text":"For an SNN parameterized matrix sensing model under gradient flow, the authors prove exponentially fast decay of squared loss and implicit selection of an approximate solution minimizing nuclear norm.","excerpt":"We prove the exponentially fast decay of squared loss during the training dynamics of an SNN parameterized matrix sensing model under gradient flow","url":"https://arxiv.org/abs/2402.17595","relationship":"Source-stated"},{"text":"The authors report empirical investigations of the SNN architecture in matrix sensing and say it has potential for broader matrix learning scenarios.","excerpt":"We also demonstrate the effectiveness of the SNN architecture in the context of matrix sensing by empirical investigations.","url":"https://arxiv.org/abs/2402.17595","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2402.17595","checkedAt":"2026-11-09T01:48:08.654Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What evidence across other matrix learning scenarios would show whether the authors’ proposed SNN architecture has the wider applicability they describe as potential?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv cs.AI: Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing","summary":"The authors propose Spectral Neural Networks (SNN), a matrix network architecture in which each neuron applies nonlinear activation to an input matrix’s singular values. They describe this as representing matrices through singular values and singular vectors rather than entries. In a matrix sensing model trained under gradient flow—a continuous approximation of discrete gradient descent—the authors give a theoretical result that squared loss decays exponentially fast and training implicitly selects an approximate solution minimizing nuclear norm. They also report empirical investigations of SNN for matrix sensing and say the architecture has potential across a broader class of matrix learning scenarios. The paper connects the architecture’s training dynamics to a particular approximate solution, while its empirical investigation concerns matrix sensing. Its abstract does not establish how far those results extend beyond that setting.","question":"What evidence across other matrix learning scenarios would show whether the authors’ proposed SNN architecture has the wider applicability they describe as potential?","claims":[{"text":"The authors propose Spectral Neural Networks, where each neuron applies nonlinear activation to the singular values of an input matrix.","excerpt":"each neuron applies nonlinear activation to the singular values of the input matrix","support":"source-stated"},{"text":"For an SNN parameterized matrix sensing model under gradient flow, the authors prove exponentially fast decay of squared loss and implicit selection of an approximate solution minimizing nuclear norm.","excerpt":"We prove the exponentially fast decay of squared loss during the training dynamics of an SNN parameterized matrix sensing model under gradient flow","support":"source-stated"},{"text":"The authors report empirical investigations of the SNN architecture in matrix sensing and say it has potential for broader matrix learning scenarios.","excerpt":"We also demonstrate the effectiveness of the SNN architecture in the context of matrix sensing by empirical investigations.","support":"source-stated"}],"limitations":["The stated theoretical results concern an SNN parameterized matrix sensing model under gradient flow; the abstract does not specify how broadly the empirical investigations extend.","The authors describe wider applicability as potential, not as a demonstrated result across a broad class of matrix learning scenarios."],"mediaId":"ad8bc169274632a30fbcdc2822c1c2e29a39fb5d521f06820e94efdd171329a6","correctionReason":null,"id":"d3cdcfad875e1227da74ec39f43b4b7877f80452b9922fbe07dcb48587ec0dd0","publicationId":"176434e8-d984-448f-8577-99f7515bc955","version":1,"sourceUrl":"https://arxiv.org/abs/2402.17595","sourceTitle":"Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2024-02-27","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-06T05:10:16.525Z","createdAt":"2026-11-09T01:17:19.825Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"ad8bc169274632a30fbcdc2822c1c2e29a39fb5d521f06820e94efdd171329a6","kind":"article","url":"https://arxiv.org/abs/2402.17595","sourceUrl":"https://arxiv.org/abs/2402.17595","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-09T01:16:39.343Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2402.17595","checkedAt":"2026-11-09T01:48:08.654Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"ad8bc169274632a30fbcdc2822c1c2e29a39fb5d521f06820e94efdd171329a6","confidence":0.68,"provider":"typesafe","model":"jev-1.13.0","inputHash":"bc2da8518a128ad4a5ee4a4f938ee99b200d0029c667cc0a5bf19efbbab42c4e","decidedAt":"2026-11-09T01:16:40.256Z"},"provenance":{"textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observationId":"9c37b2c15659a930dc6f5c4acd9388c49ed98b4e0347b6d848142082b1ad2e24","retrievedAt":"2026-11-09T01:16:39.197Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"a6f35515-3095-4664-88aa-e3b64e051992"},"reviewer":{"model":"gpt-6-luna","runId":"386373b4-b02b-4ef0-8f2d-d10ef52ae42d"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-09T01:17:19.825Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Chu, Hong T. M.","url":null,"kind":"unknown"},{"name":"Dong, Xiaoyu","url":null,"kind":"unknown"},{"name":"Ghosh, Subhro","url":null,"kind":"unknown"},{"name":"Lam, Chi Thanh","url":null,"kind":"unknown"},{"name":"Mukherjee, Soumendu Sundar","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Chu, Hong T. M.","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Chu, Hong T. M.","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2402.17595","locator":"head/meta[name=author|citation_author]"},"assertionId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T01:48:08.654Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","sourceUrl":"https://arxiv.org/abs/2402.17595","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z"}},{"actor":{"name":"Dong, Xiaoyu","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Dong, Xiaoyu","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2402.17595","locator":"head/meta[name=author|citation_author]"},"assertionId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T01:48:08.654Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","sourceUrl":"https://arxiv.org/abs/2402.17595","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z"}},{"actor":{"name":"Ghosh, Subhro","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Ghosh, Subhro","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2402.17595","locator":"head/meta[name=author|citation_author]"},"assertionId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T01:48:08.654Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","sourceUrl":"https://arxiv.org/abs/2402.17595","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z"}},{"actor":{"name":"Lam, Chi Thanh","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lam, Chi Thanh","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2402.17595","locator":"head/meta[name=author|citation_author]"},"assertionId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T01:48:08.654Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","sourceUrl":"https://arxiv.org/abs/2402.17595","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z"}},{"actor":{"name":"Mukherjee, Soumendu Sundar","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Mukherjee, Soumendu Sundar","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2402.17595","locator":"head/meta[name=author|citation_author]"},"assertionId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T01:48:08.654Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","sourceUrl":"https://arxiv.org/abs/2402.17595","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z"}}],"participationCoverage":{"total":5,"returned":5,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2402.17595","snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z","basis":"publisher-page-metadata","sourceLinks":50,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/d3cdcfad875e1227da74ec39f43b4b7877f80452b9922fbe07dcb48587ec0dd0/source"}},"canonicalUrl":"https://escapemeridian.com/readings/arxiv-cs-ai-implicit-regularization-via-spectral-neural-d3cdcfad","formats":{"html":"/readings/arxiv-cs-ai-implicit-regularization-via-spectral-neural-d3cdcfad","markdown":"/readings/arxiv-cs-ai-implicit-regularization-via-spectral-neural-d3cdcfad.md","json":"/readings/arxiv-cs-ai-implicit-regularization-via-spectral-neural-d3cdcfad.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-neural-network","name":"Neural network","explanation":"A system of connected simple calculations, loosely inspired by the brain, that learns patterns from data.","referenceUrl":"https://developers.google.com/machine-learning/glossary#neural-network"}],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"d3cdcfad875e1227da74ec39f43b4b7877f80452b9922fbe07dcb48587ec0dd0","revision":"176434e8-d984-448f-8577-99f7515bc955","revisionKind":"publication-id","title":"arXiv cs.AI: Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing","url":"https://escapemeridian.com/readings/arxiv-cs-ai-implicit-regularization-via-spectral-neural-d3cdcfad"},"source":{"type":"source","id":"https://arxiv.org/abs/2402.17595","revision":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","revisionKind":"sha256-source-text","title":"Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing","url":"https://arxiv.org/abs/2402.17595"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"d3cdcfad875e1227da74ec39f43b4b7877f80452b9922fbe07dcb48587ec0dd0","title":"arXiv cs.AI: Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing","argument":"The authors propose Spectral Neural Networks (SNN), a matrix network architecture in which each neuron applies nonlinear activation to an input matrix’s singular values. They describe this as representing matrices through singular values and singular vectors rather than entries. In a matrix sensing model trained under gradient flow—a continuous approximation of discrete gradient descent—the authors give a theoretical result that squared loss decays exponentially fast and training implicitly selects an approximate solution minimizing nuclear norm. They also report empirical investigations of SNN for matrix sensing and say the architecture has potential across a broader class of matrix learning scenarios. The paper connects the architecture’s training dynamics to a particular approximate solution, while its empirical investigation concerns matrix sensing. Its abstract does not establish how far those results extend beyond that setting.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The stated theoretical results concern an SNN parameterized matrix sensing model under gradient flow; the abstract does not specify how broadly the empirical investigations extend. The authors describe wider applicability as potential, not as a demonstrated result across a broad class of matrix learning scenarios.","speaker":{"name":"Chu, Hong T. M., Dong, Xiaoyu, Ghosh, Subhro, Lam, Chi Thanh, Mukherjee, Soumendu Sundar","roleAtSource":"Article author"},"source":{"title":"Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2402.17595","publishedAt":"2024-02-27","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose Spectral Neural Networks, where each neuron applies nonlinear activation to the singular values of an input matrix.","excerpt":"each neuron applies nonlinear activation to the singular values of the input matrix","url":"https://arxiv.org/abs/2402.17595","relationship":"Source-stated"},{"text":"For an SNN parameterized matrix sensing model under gradient flow, the authors prove exponentially fast decay of squared loss and implicit selection of an approximate solution minimizing nuclear norm.","excerpt":"We prove the exponentially fast decay of squared loss during the training dynamics of an SNN parameterized matrix sensing model under gradient flow","url":"https://arxiv.org/abs/2402.17595","relationship":"Source-stated"},{"text":"The authors report empirical investigations of the SNN architecture in matrix sensing and say it has potential for broader matrix learning scenarios.","excerpt":"We also demonstrate the effectiveness of the SNN architecture in the context of matrix sensing by empirical investigations.","url":"https://arxiv.org/abs/2402.17595","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2402.17595","checkedAt":"2026-11-09T01:48:08.654Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What evidence across other matrix learning scenarios would show whether the authors’ proposed SNN architecture has the wider applicability they describe as potential?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv cs.AI: Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing","summary":"The authors propose Spectral Neural Networks (SNN), a matrix network architecture in which each neuron applies nonlinear activation to an input matrix’s singular values. They describe this as representing matrices through singular values and singular vectors rather than entries. In a matrix sensing model trained under gradient flow—a continuous approximation of discrete gradient descent—the authors give a theoretical result that squared loss decays exponentially fast and training implicitly selects an approximate solution minimizing nuclear norm. They also report empirical investigations of SNN for matrix sensing and say the architecture has potential across a broader class of matrix learning scenarios. The paper connects the architecture’s training dynamics to a particular approximate solution, while its empirical investigation concerns matrix sensing. Its abstract does not establish how far those results extend beyond that setting.","question":"What evidence across other matrix learning scenarios would show whether the authors’ proposed SNN architecture has the wider applicability they describe as potential?","claims":[{"text":"The authors propose Spectral Neural Networks, where each neuron applies nonlinear activation to the singular values of an input matrix.","excerpt":"each neuron applies nonlinear activation to the singular values of the input matrix","support":"source-stated"},{"text":"For an SNN parameterized matrix sensing model under gradient flow, the authors prove exponentially fast decay of squared loss and implicit selection of an approximate solution minimizing nuclear norm.","excerpt":"We prove the exponentially fast decay of squared loss during the training dynamics of an SNN parameterized matrix sensing model under gradient flow","support":"source-stated"},{"text":"The authors report empirical investigations of the SNN architecture in matrix sensing and say it has potential for broader matrix learning scenarios.","excerpt":"We also demonstrate the effectiveness of the SNN architecture in the context of matrix sensing by empirical investigations.","support":"source-stated"}],"limitations":["The stated theoretical results concern an SNN parameterized matrix sensing model under gradient flow; the abstract does not specify how broadly the empirical investigations extend.","The authors describe wider applicability as potential, not as a demonstrated result across a broad class of matrix learning scenarios."],"mediaId":"ad8bc169274632a30fbcdc2822c1c2e29a39fb5d521f06820e94efdd171329a6","correctionReason":null,"id":"d3cdcfad875e1227da74ec39f43b4b7877f80452b9922fbe07dcb48587ec0dd0","publicationId":"176434e8-d984-448f-8577-99f7515bc955","version":1,"sourceUrl":"https://arxiv.org/abs/2402.17595","sourceTitle":"Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2024-02-27","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-06T05:10:16.525Z","createdAt":"2026-11-09T01:17:19.825Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"ad8bc169274632a30fbcdc2822c1c2e29a39fb5d521f06820e94efdd171329a6","kind":"article","url":"https://arxiv.org/abs/2402.17595","sourceUrl":"https://arxiv.org/abs/2402.17595","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-09T01:16:39.343Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2402.17595","checkedAt":"2026-11-09T01:48:08.654Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"ad8bc169274632a30fbcdc2822c1c2e29a39fb5d521f06820e94efdd171329a6","confidence":0.68,"provider":"typesafe","model":"jev-1.13.0","inputHash":"bc2da8518a128ad4a5ee4a4f938ee99b200d0029c667cc0a5bf19efbbab42c4e","decidedAt":"2026-11-09T01:16:40.256Z"},"provenance":{"textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observationId":"9c37b2c15659a930dc6f5c4acd9388c49ed98b4e0347b6d848142082b1ad2e24","retrievedAt":"2026-11-09T01:16:39.197Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"a6f35515-3095-4664-88aa-e3b64e051992"},"reviewer":{"model":"gpt-6-luna","runId":"386373b4-b02b-4ef0-8f2d-d10ef52ae42d"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-09T01:17:19.825Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Chu, Hong T. M.","url":null,"kind":"unknown"},{"name":"Dong, Xiaoyu","url":null,"kind":"unknown"},{"name":"Ghosh, Subhro","url":null,"kind":"unknown"},{"name":"Lam, Chi Thanh","url":null,"kind":"unknown"},{"name":"Mukherjee, Soumendu Sundar","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Chu, Hong T. M.","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Chu, Hong T. M.","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2402.17595","locator":"head/meta[name=author|citation_author]"},"assertionId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T01:48:08.654Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","sourceUrl":"https://arxiv.org/abs/2402.17595","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z"}},{"actor":{"name":"Dong, Xiaoyu","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Dong, Xiaoyu","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2402.17595","locator":"head/meta[name=author|citation_author]"},"assertionId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T01:48:08.654Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","sourceUrl":"https://arxiv.org/abs/2402.17595","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z"}},{"actor":{"name":"Ghosh, Subhro","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Ghosh, Subhro","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2402.17595","locator":"head/meta[name=author|citation_author]"},"assertionId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T01:48:08.654Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","sourceUrl":"https://arxiv.org/abs/2402.17595","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z"}},{"actor":{"name":"Lam, Chi Thanh","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lam, Chi Thanh","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2402.17595","locator":"head/meta[name=author|citation_author]"},"assertionId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T01:48:08.654Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","sourceUrl":"https://arxiv.org/abs/2402.17595","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z"}},{"actor":{"name":"Mukherjee, Soumendu Sundar","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Mukherjee, Soumendu Sundar","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2402.17595","locator":"head/meta[name=author|citation_author]"},"assertionId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T01:48:08.654Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","sourceUrl":"https://arxiv.org/abs/2402.17595","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z"}}],"participationCoverage":{"total":5,"returned":5,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2402.17595","snapshotId":"f40f48908943332ffbf171008650b6e96c28ee89631c953bd9340f06a6b4579d","contentSha256":"9e62e78eda0119cffdef8a3eb8031e9e175c0371ecc6414afeb284804fb2d88c","textHash":"7082b0c3a835fdb2d7a2c8d447791f2f7aee14aa7421af5c132498ce927fac83","observedAt":"2026-11-09T01:48:08.654Z","basis":"publisher-page-metadata","sourceLinks":50,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/d3cdcfad875e1227da74ec39f43b4b7877f80452b9922fbe07dcb48587ec0dd0/source"}}}],"audits":[{"publicationId":"176434e8-d984-448f-8577-99f7515bc955","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading attributes the architecture, proofs, empirical work and the claim of broader potential to the paper's authors (Chu, Dong, Ghosh, Lam, Mukherjee), as listed on the arXiv page. It keeps the authors' claims as their claims rather than presenting them as established facts.","excerpt":"Authors:Hong T.M. Chu, Xiaoyu Dong, Subhro Ghosh, Chi Thanh Lam, Soumendu Sundar Mukherjee"},{"dimension":"summary","status":"uncertain","explanation":"The reading leaves out the abstract's central framing: implicit bias, where gradient descent without an explicit regularizer still converges to a regularized solution and is tied to neural networks' ability to generalize. Without that framing, a non-specialist cannot see why the nuclear-norm result matters. Central terms are also unexplained: matrix sensing, singular values, nuclear norm and implicit regularization. The line about connecting training dynamics to 'a particular approximate solution' is vague. Smallest repair: add one plain sentence on implicit bias and why it matters, and flag these terms for tooltips.","excerpt":"it entails that gradient descent dynamics in many neural nets, even without any explicit regularizer in the loss function, converges to the solution of a regularized learning problem"},{"dimension":"date","status":"pass","explanation":"The originalPublished date of 2024-02-27 matches the v1 submission date. It is correctly used instead of the v2 revision (5 Nov 2026) or the feed date.","excerpt":"Submitted on 27 Feb 2024 (v1), last revised 5 Nov 2026 (this version, v2)"},{"dimension":"support","status":"pass","explanation":"Each claim matches the abstract's statements: the neuron design, the proof of exponential loss decay and approximate nuclear-norm minimization under gradient flow, the empirical matrix sensing work, and potential wider applicability. The limitations and the question correctly treat broader applicability as the authors' belief rather than a demonstrated result. No benefits are invented.","excerpt":"We believe that the SNN architecture has the potential to be of wide applicability in a broad class of matrix learning scenarios."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper, which is the most direct access to the source.","excerpt":null}],"createdAt":"2026-11-09T01:17:19.825Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}