{"id":"https://openalex.org/W4416969256","doi":"https://doi.org/10.48550/arxiv.2512.01980","title":"Low-Rank Prehab: Preparing Neural Networks for SVD Compression","display_name":"Low-Rank Prehab: Preparing Neural Networks for SVD Compression","publication_year":2025,"publication_date":"2025-12-01","ids":{"openalex":"https://openalex.org/W4416969256","doi":"https://doi.org/10.48550/arxiv.2512.01980"},"language":null,"primary_location":{"id":"pmh:oai:arXiv.org:2512.01980","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2512.01980","pdf_url":"https://arxiv.org/pdf/2512.01980","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2512.01980","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101558908","display_name":"Haoran Qin","orcid":"https://orcid.org/0009-0007-4789-5488"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Qin, Haoran","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Sharma, Shansita","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sharma, Shansita","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101476807","display_name":"Ali Abbasi","orcid":"https://orcid.org/0009-0008-4116-0920"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Abbasi, Ali","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5093329996","display_name":"Chayne Thrash","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Thrash, Chayne","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5068682350","display_name":"Soheil Kolouri","orcid":"https://orcid.org/0000-0001-8495-5362"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kolouri, Soheil","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5101558908"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.37869998812675476,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.37869998812675476,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10500","display_name":"Sparse and Compressive Sensing Techniques","score":0.15719999372959137,"subfield":{"id":"https://openalex.org/subfields/2206","display_name":"Computational Mechanics"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.08630000054836273,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/singular-value-decomposition","display_name":"Singular value decomposition","score":0.7009999752044678},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.6552000045776367},{"id":"https://openalex.org/keywords/compression","display_name":"Compression (physics)","score":0.5835000276565552},{"id":"https://openalex.org/keywords/data-compression","display_name":"Data compression","score":0.5303999781608582},{"id":"https://openalex.org/keywords/decomposition","display_name":"Decomposition","score":0.47589999437332153},{"id":"https://openalex.org/keywords/source-code","display_name":"Source code","score":0.3772999942302704},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.35989999771118164},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.3499000072479248}],"concepts":[{"id":"https://openalex.org/C22789450","wikidata":"https://www.wikidata.org/wiki/Q420904","display_name":"Singular value decomposition","level":2,"score":0.7009999752044678},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6607999801635742},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.6552000045776367},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.5835000276565552},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.5303999781608582},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.48420000076293945},{"id":"https://openalex.org/C124681953","wikidata":"https://www.wikidata.org/wiki/Q339062","display_name":"Decomposition","level":2,"score":0.47589999437332153},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.44279998540878296},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.3772999942302704},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.35989999771118164},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.3499000072479248},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.34790000319480896},{"id":"https://openalex.org/C25797200","wikidata":"https://www.wikidata.org/wiki/Q828137","display_name":"Compression ratio","level":3,"score":0.34470000863075256},{"id":"https://openalex.org/C13481523","wikidata":"https://www.wikidata.org/wiki/Q412438","display_name":"Image compression","level":4,"score":0.33079999685287476},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3043999969959259},{"id":"https://openalex.org/C42355184","wikidata":"https://www.wikidata.org/wiki/Q1361088","display_name":"Matrix decomposition","level":3,"score":0.2786000072956085},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.27399998903274536},{"id":"https://openalex.org/C148764684","wikidata":"https://www.wikidata.org/wiki/Q621751","display_name":"Approximation algorithm","level":2,"score":0.26649999618530273},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.25200000405311584},{"id":"https://openalex.org/C94835093","wikidata":"https://www.wikidata.org/wiki/Q3113333","display_name":"Data compression ratio","level":5,"score":0.25049999356269836}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2512.01980","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2512.01980","pdf_url":"https://arxiv.org/pdf/2512.01980","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2512.01980","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2512.01980","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2512.01980","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2512.01980","pdf_url":"https://arxiv.org/pdf/2512.01980","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2867940020","display_name":null,"funder_award_id":"2339898","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G6671297155","display_name":null,"funder_award_id":"CAREER","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416969256.pdf","grobid_xml":"https://content.openalex.org/works/W4416969256.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Low-rank":[0],"approximation":[1,91],"methods":[2],"such":[3,138],"as":[4,20,32,41,139],"singular":[5],"value":[6],"decomposition":[7,30],"(SVD)":[8],"and":[9,92,101,120,155],"its":[10],"variants":[11],"(e.g.,":[12],"Fisher-weighted":[13],"SVD,":[14,76],"Activation":[15],"SVD)":[16],"have":[17],"recently":[18],"emerged":[19],"effective":[21],"tools":[22],"for":[23,147],"neural":[24],"network":[25],"compression.":[26],"In":[27],"this":[28],"setting,":[29],"acts":[31],"a":[33,53,126],"\"surgical\"":[34],"intervention,":[35],"followed":[36],"by":[37,47],"fine-tuning":[38,55],"that":[39,59,110],"serves":[40],"\"rehab\"":[42],"to":[43],"recover":[44],"accuracy.":[45],"Inspired":[46],"prehabilitation":[48],"in":[49,64],"surgery,":[50],"we":[51],"introduce":[52],"pre-compression":[54],"stage,":[56],"Low-Rank":[57],"Prehab,":[58],"explicitly":[60],"encourages":[61],"low-rank":[62,90],"structure":[63],"weight":[65],"matrices":[66],"while":[67],"preserving":[68],"task":[69],"performance.":[70,124],"By":[71],"conditioning":[72],"the":[73,85,114,142,153],"model":[74],"before":[75],"Prehab":[77,111],"steers":[78],"weights":[79],"toward":[80],"spectrally":[81],"compact":[82],"regions":[83],"of":[84,129,144],"parameter":[86],"space,":[87],"enabling":[88],"smoother":[89],"improved":[93],"recovery.":[94],"Experiments":[95],"on":[96],"large":[97],"language":[98],"models":[99,146],"(LLMs)":[100],"other":[102],"Transformer-based":[103],"architectures,":[104],"including":[105],"Vision":[106],"Transformers":[107],"(ViTs),":[108],"show":[109],"substantially":[112],"reduces":[113],"immediate":[115],"accuracy":[116],"drop":[117],"after":[118],"compression":[119,130,148,154],"consistently":[121],"improves":[122],"post-finetuning":[123],"Across":[125],"wide":[127],"range":[128],"ratios,":[131],"our":[132],"method":[133],"outperforms":[134],"state-of-the-art":[135],"SVD-based":[136],"techniques":[137],"SVD-LLM,":[140],"highlighting":[141],"importance":[143],"preparing":[145],"rather":[149],"than":[150],"only":[151],"improving":[152],"recovery":[156],"stages.":[157],"Source":[158],"code":[159],"is":[160],"available":[161],"at":[162],"https://github.com/niqretnuh/PREHAB-SVD":[163]},"counts_by_year":[],"updated_date":"2026-04-21T08:09:41.155169","created_date":"2025-12-03T00:00:00"}