{"id":"https://openalex.org/W7162450667","doi":"https://doi.org/10.48550/arxiv.2605.25939","title":"From Latent Space to Training Data: Explainable Specialization in Minimal MLPs","display_name":"From Latent Space to Training Data: Explainable Specialization in Minimal MLPs","publication_year":2026,"publication_date":"2026-05-25","ids":{"openalex":"https://openalex.org/W7162450667","doi":"https://doi.org/10.48550/arxiv.2605.25939"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.25939","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25939","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":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.25939","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137016210","display_name":"Enrique Alba","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Alba, Enrique","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137022355","display_name":"Ezequiel Lopez-Rubio","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lopez-Rubio, Ezequiel","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.3357999920845032,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.3357999920845032,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.10530000180006027,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.09910000115633011,"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/hyperparameter","display_name":"Hyperparameter","score":0.6794000267982483},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.4747999906539917},{"id":"https://openalex.org/keywords/regularization","display_name":"Regularization (linguistics)","score":0.4722000062465668},{"id":"https://openalex.org/keywords/convex-hull","display_name":"Convex hull","score":0.46230000257492065},{"id":"https://openalex.org/keywords/regular-polygon","display_name":"Regular polygon","score":0.4050000011920929},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.3991999924182892},{"id":"https://openalex.org/keywords/hull","display_name":"Hull","score":0.38429999351501465},{"id":"https://openalex.org/keywords/space","display_name":"Space (punctuation)","score":0.3749000132083893}],"concepts":[{"id":"https://openalex.org/C8642999","wikidata":"https://www.wikidata.org/wiki/Q4171168","display_name":"Hyperparameter","level":2,"score":0.6794000267982483},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.49149999022483826},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.48170000314712524},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.4747999906539917},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.4722000062465668},{"id":"https://openalex.org/C206194317","wikidata":"https://www.wikidata.org/wiki/Q1138624","display_name":"Convex hull","level":3,"score":0.46230000257492065},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.43860000371932983},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.42100000381469727},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4059000015258789},{"id":"https://openalex.org/C112680207","wikidata":"https://www.wikidata.org/wiki/Q714886","display_name":"Regular polygon","level":2,"score":0.4050000011920929},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3991999924182892},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.38760000467300415},{"id":"https://openalex.org/C37423430","wikidata":"https://www.wikidata.org/wiki/Q6750281","display_name":"Hull","level":2,"score":0.38429999351501465},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.3749000132083893},{"id":"https://openalex.org/C2776061190","wikidata":"https://www.wikidata.org/wiki/Q7451805","display_name":"Separation (statistics)","level":2,"score":0.3538999855518341},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.3422999978065491},{"id":"https://openalex.org/C72319582","wikidata":"https://www.wikidata.org/wiki/Q584304","display_name":"Degenerate energy levels","level":2,"score":0.3273000121116638},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3082999885082245},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.29739999771118164},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.2924000024795532},{"id":"https://openalex.org/C2780069185","wikidata":"https://www.wikidata.org/wiki/Q7977945","display_name":"Equivalence (formal languages)","level":2,"score":0.2847000062465668},{"id":"https://openalex.org/C139945424","wikidata":"https://www.wikidata.org/wiki/Q1940696","display_name":"Mean squared error","level":2,"score":0.2784999907016754},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.2703000009059906},{"id":"https://openalex.org/C167085575","wikidata":"https://www.wikidata.org/wiki/Q6803654","display_name":"Mean squared prediction error","level":2,"score":0.26829999685287476},{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.26649999618530273},{"id":"https://openalex.org/C72134830","wikidata":"https://www.wikidata.org/wiki/Q5166524","display_name":"Convexity","level":2,"score":0.25360000133514404}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.25939","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25939","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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.25939","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25939","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":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"score":0.517366349697113,"display_name":"Peace, Justice and strong institutions","id":"https://metadata.un.org/sdg/16"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"We":[0,30,124],"here":[1],"study":[2],"whether":[3,16],"training":[4,24,51,164],"biases":[5],"can":[6],"make":[7],"hidden":[8,61],"neurons":[9],"specialize":[10],"in":[11,152],"minimal":[12],"one-hidden-layer":[13],"MLPs,":[14],"and":[15,40,57,102,118,171,183,199],"such":[17],"specialization":[18,106],"improves":[19],"prototype-based":[20],"reconstruction":[21,96],"of":[22,34,49,60,162],"the":[23,27,50,64,93,104,110,127,137,146,159,163,188,196,208,235],"dataset":[25,38],"from":[26,78],"learned":[28],"weights.":[29],"consider":[31],"Gaussianactivation":[32],"MLPs":[33],"width":[35],"equal":[36],"to":[37,82,109,148,241],"size":[39,101],"compare":[41],"three":[42],"structural":[43,222],"losses":[44],"that":[45,126],"respectively":[46],"encourage":[47],"coverage":[48],"samples,":[52],"separation":[53,114,176],"between":[54],"neuron-induced":[55],"prototypes,":[56],"low":[58],"overlap":[59,119,184],"responses,":[62],"against":[63],"standard":[65,111],"fitting":[66],"baseline.":[67],"Experiments":[68],"on":[69,195],"uniformly":[70],"sampled":[71],"one-dimensional":[72],"datasets":[73],"show":[74,125],"a":[75,149,200,213,228],"stable":[76],"pattern":[77],"N":[79,83,204],"=":[80,84,205],"3":[81],"100":[85,206],"across":[86],"480":[87],"controlled":[88],"runs.":[89],"Coverage":[90,166],"regularization":[91],"gives":[92],"lowest":[94],"mean":[95],"error":[97],"at":[98,180,187,203],"every":[99,220],"tested":[100],"raises":[103],"prototype-usage":[105],"ratio":[107],"relative":[108],"baseline,":[112],"while":[113],"has":[115],"mixed":[116],"effects":[117],"penalties":[120],"are":[121,156],"systematically":[122],"harmful.":[123],"harm":[128],"is":[129],"not":[130],"an":[131,174],"optimization":[132],"failure:":[133],"overlap-active":[134],"approaches":[135],"fit":[136],"data":[138],"as":[139,141,173],"well":[140],"overlap-free":[142],"ones":[143],"but":[144],"route":[145],"optimizer":[147],"degenerate":[150],"equilibrium":[151],"which":[153],"prototype":[154],"centers":[155],"pushed":[157],"outside":[158],"convex":[160],"hull":[161],"inputs.":[165],"cannot":[167],"reward":[168],"this":[169],"expulsion":[170],"acts":[172],"attractor:":[175],"admits":[177,185],"it":[178,186,232,238],"only":[179],"large":[181],"temperature":[182],"nominal":[189],"hyperparameter":[190],"choice.":[191],"A":[192],"direct":[193],"\u03c4-sweep":[194],"separation-only":[197],"mask":[198],"prototype-position":[201],"visualization":[202],"confirm":[207],"mechanism.":[209],"The":[210],"findings":[211],"yield":[212],"simple":[214],"design":[215],"principle":[216],"for":[217],"prototype-recoverability-aware":[218],"training:":[219],"repulsive":[221],"loss":[223],"must":[224],"be":[225],"compensated":[226],"by":[227],"compatible":[229],"attractor,":[230],"or":[231],"will":[233],"collapse":[234],"latent":[236],"geometry":[237],"was":[239],"meant":[240],"refine.":[242]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-27T00:00:00"}