{"id":"https://openalex.org/W7225423378","doi":"https://doi.org/10.20935/acadai8620","title":"Heat-kernel context compression for transformer representations using graph diffusion","display_name":"Heat-kernel context compression for transformer representations using graph diffusion","publication_year":2026,"publication_date":"2026-11-06","ids":{"openalex":"https://openalex.org/W7225423378","doi":"https://doi.org/10.20935/acadai8620"},"language":"en","primary_location":{"id":"doi:10.20935/acadai8620","doi":"https://doi.org/10.20935/acadai8620","is_oa":true,"landing_page_url":"https://doi.org/10.20935/acadai8620","pdf_url":"https://www.academia.edu/3071-0286/2/4/10.20935/AcadAI8620.pdf","source":{"id":"https://openalex.org/S7407056243","display_name":"Academia AI and Applications","issn_l":"3071-0286","issn":["3071-0286"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Academia AI and Applications","raw_type":"journal-article"},"volume":null,"type":"article","is_traditional":true,"indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://www.academia.edu/3071-0286/2/4/10.20935/AcadAI8620.pdf","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5003346134","display_name":"Shashank Reddy Vadyala","orcid":"https://orcid.org/0000-0001-6769-6434","observed_orcids":["https://orcid.org/0000-0001-6769-6434"]},"institutions":[{"id":"https://openalex.org/I1283103587","display_name":"AT&T (United States)","ror":"https://ror.org/02bbd5539","country_code":"US","type":"company","lineage":["https://openalex.org/I1283103587"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Shashank Reddy Vadyala","raw_affiliation_strings":["AT&T, Dallas, TX, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AT&T, Dallas, TX, USA","institution_ids":["https://openalex.org/I1283103587"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5045774308","display_name":"Sai Nethra Betgeri","orcid":"https://orcid.org/0009-0003-5946-8847","observed_orcids":["https://orcid.org/0009-0003-5946-8847","https://orcid.org/0000-0001-7113-7141"]},"institutions":[{"id":"https://openalex.org/I142740786","display_name":"University of Louisville","ror":"https://ror.org/01ckdn478","country_code":"US","type":"education","lineage":["https://openalex.org/I142740786"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Sai Nethra Betgeri","raw_affiliation_strings":["Computer Science and Engineering Department, University of Louisville, Louisville, KY, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Computer Science and Engineering Department, University of Louisville, Louisville, KY, USA","institution_ids":["https://openalex.org/I142740786"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5045774308"],"corresponding_institution_ids":["https://openalex.org/I142740786"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.0,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"2","issue":"4","first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.45899999141693115,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.45899999141693115,"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"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.4458000063896179,"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.005499999970197678,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/heat-kernel","display_name":"heat kernel","score":0.609000027179718},{"id":"https://openalex.org/keywords/context-compression","display_name":"context compression","score":0.8510000109672546},{"id":"https://openalex.org/keywords/transformer-representations","display_name":"transformer representations","score":0.8299999833106995},{"id":"https://openalex.org/keywords/graph-diffusion","display_name":"graph diffusion","score":0.9459999799728394},{"id":"https://openalex.org/keywords/representation-compression","display_name":"representation compression","score":0.4790000021457672},{"id":"https://openalex.org/keywords/chebyshev-spectral-method","display_name":"Chebyshev spectral method","score":0.6589999794960022},{"id":"https://openalex.org/keywords/training-free-compression","display_name":"training-free compression","score":0.5450000166893005}],"concepts":[{"id":"https://openalex.org/C183212220","wikidata":"https://www.wikidata.org/wiki/Q3345669","display_name":"Heat kernel","level":2,"score":0.621399998664856},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5932000279426575},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5748999714851379},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.498199999332428},{"id":"https://openalex.org/C136886441","wikidata":"https://www.wikidata.org/wiki/Q926129","display_name":"Normalization (sociology)","level":2,"score":0.4715999960899353},{"id":"https://openalex.org/C32900221","wikidata":"https://www.wikidata.org/wiki/Q181365","display_name":"Dot product","level":2,"score":0.42170000076293945},{"id":"https://openalex.org/C115178988","wikidata":"https://www.wikidata.org/wiki/Q772067","display_name":"Laplacian matrix","level":3,"score":0.388700008392334},{"id":"https://openalex.org/C155512373","wikidata":"https://www.wikidata.org/wiki/Q287450","display_name":"Residual","level":2,"score":0.38510000705718994},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3837999999523163},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.3781000077724457},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.36809998750686646},{"id":"https://openalex.org/C165700671","wikidata":"https://www.wikidata.org/wiki/Q203484","display_name":"Laplace operator","level":2,"score":0.3671000003814697},{"id":"https://openalex.org/C178009071","wikidata":"https://www.wikidata.org/wiki/Q93344","display_name":"Trigonometric functions","level":2,"score":0.3495999872684479},{"id":"https://openalex.org/C2780762811","wikidata":"https://www.wikidata.org/wiki/Q1784941","display_name":"Cosine similarity","level":3,"score":0.32710000872612},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.3089999854564667},{"id":"https://openalex.org/C2776459999","wikidata":"https://www.wikidata.org/wiki/Q2119376","display_name":"Fidelity","level":2,"score":0.30889999866485596},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.3005000054836273},{"id":"https://openalex.org/C21424316","wikidata":"https://www.wikidata.org/wiki/Q718621","display_name":"Chebyshev filter","level":2,"score":0.26579999923706055},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.2578999996185303},{"id":"https://openalex.org/C176809094","wikidata":"https://www.wikidata.org/wiki/Q15401496","display_name":"Traverse","level":2,"score":0.25600001215934753},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2551000118255615},{"id":"https://openalex.org/C129785596","wikidata":"https://www.wikidata.org/wiki/Q619511","display_name":"Chebyshev polynomials","level":2,"score":0.2540999948978424},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.2529999911785126},{"id":"https://openalex.org/C55526617","wikidata":"https://www.wikidata.org/wiki/Q719375","display_name":"Operand","level":2,"score":0.25130000710487366},{"id":"https://openalex.org/C54732982","wikidata":"https://www.wikidata.org/wiki/Q1415345","display_name":"Modulo","level":2,"score":0.25040000677108765}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.20935/acadai8620","doi":"https://doi.org/10.20935/acadai8620","is_oa":true,"landing_page_url":"https://doi.org/10.20935/acadai8620","pdf_url":"https://www.academia.edu/3071-0286/2/4/10.20935/AcadAI8620.pdf","source":{"id":"https://openalex.org/S7407056243","display_name":"Academia AI and Applications","issn_l":"3071-0286","issn":["3071-0286"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Academia AI and Applications","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.20935/acadai8620","doi":"https://doi.org/10.20935/acadai8620","is_oa":true,"landing_page_url":"https://doi.org/10.20935/acadai8620","pdf_url":"https://www.academia.edu/3071-0286/2/4/10.20935/AcadAI8620.pdf","source":{"id":"https://openalex.org/S7407056243","display_name":"Academia AI and Applications","issn_l":"3071-0286","issn":["3071-0286"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Academia AI and Applications","raw_type":"journal-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7225423378.pdf","grobid_xml":"https://content.openalex.org/works/W7225423378.grobid-xml"},"referenced_works_count":15,"referenced_works":["https://openalex.org/W1578099820","https://openalex.org/W2101491865","https://openalex.org/W2158787690","https://openalex.org/W2468907370","https://openalex.org/W2626778328","https://openalex.org/W4213367101","https://openalex.org/W4306886919","https://openalex.org/W4385801014","https://openalex.org/W4388979610","https://openalex.org/W4389519226","https://openalex.org/W4389524473","https://openalex.org/W4402716428","https://openalex.org/W4415798186","https://openalex.org/W7133221612","https://openalex.org/W7133224090"],"related_works":["https://openalex.org/W4416039180","https://openalex.org/W7160596962","https://openalex.org/W4389524288","https://openalex.org/W7101747908","https://openalex.org/W7224753881","https://openalex.org/W7225084683","https://openalex.org/W4407574719","https://openalex.org/W4416252571","https://openalex.org/W7162305076","https://openalex.org/W7133191540"],"abstract_inverted_index":{"Introduction:":[0],"Long":[1],"contexts":[2],"often":[3],"contain":[4],"geometrically":[5],"redundant":[6],"internal":[7,24],"representations.":[8],"Heat":[9,66],"Kernel":[10],"Context":[11,35],"Compression":[12],"(HKCC)":[13],"is":[14,139],"a":[15,40,142],"training-free":[16],"representation\u2013compression":[17,148],"method":[18,45],"for":[19,62,147,163],"transformer":[20,154],"systems":[21],"in":[22,149],"which":[23],"embeddings":[25],"and":[26,33,70,95,120,187,190],"attention":[27,114],"logits":[28],"can":[29],"be":[30],"modified.":[31],"Materials":[32],"methods:":[34],"representations":[36,72],"are":[37,73],"placed":[38],"on":[39],"sparse":[41],"similarity":[42,59],"graph.":[43],"The":[44,156,180],"uses":[46],"the":[47,87],"random-walk":[48,88],"Laplacian":[49],"throughout,":[50],"while":[51],"symmetric":[52,100],"normalization":[53],"appears":[54],"only":[55],"through":[56],"an":[57],"explicit":[58],"transform":[60],"used":[61],"stable":[63],"Chebyshev":[64],"computation.":[65],"residuals":[67],"select":[68],"anchors,":[69],"omitted":[71],"merged":[74],"into":[75],"multiplicity-weighted":[76],"representatives.":[77],"Results:":[78],"Corrected":[79],"simulations":[80],"with":[81,97],"340":[82],"unit-normalized":[83],"vectors":[84],"show":[85],"that":[86],"heat":[89,177],"kernel":[90],"preserves":[91],"row":[92],"sums":[93],"exactly":[94],"agrees":[96],"its":[98],"similarity-transformed":[99],"implementation":[101],"to":[102],"numerical":[103],"precision.":[104],"Across":[105],"six":[106],"controlled":[107,143],"contexts,":[108],"multiplicity-aware":[109],"merging":[110],"preserved":[111],"approximately":[112],"0.99999":[113],"output":[115],"cosine":[116],"fidelity":[117],"at":[118],"8%":[119],"12%":[121],"retention,":[122],"whereas":[123],"residual-only":[124],"pruning":[125],"degraded":[126],"sharply.":[127],"Anchor":[128],"rules":[129],"were":[130],"effectively":[131],"tied":[132],"under":[133],"this":[134],"synthetic":[135],"regime.":[136],"Conclusions:":[137],"HKCC":[138],"supported":[140],"as":[141],"proof":[144],"of":[145,176,183],"concept":[146],"self-hosted":[150],"or":[151,173],"otherwise":[152],"modifiable":[153],"stacks.":[155],"current":[157],"study":[158],"does":[159],"not":[160],"establish":[161],"savings":[162],"commercial":[164],"text":[165],"application":[166],"programming":[167],"interfaces":[168],"(APIs),":[169],"downstream":[170],"task":[171],"preservation,":[172],"universal":[174],"superiority":[175],"residual":[178],"anchors.":[179],"added":[181],"utility":[182],"diffusion,":[184],"real-model":[185],"key/value":[186],"positional":[188],"integration,":[189],"end-to-end":[191],"prefill":[192],"speedups":[193],"remain":[194],"unestablished.":[195]},"counts_by_year":[],"updated_date":"2026-11-08T12:12:28.789561","created_date":"2026-11-07T00:00:00"}