{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:FURGR5AGEL4XWI3AAWCPP24YRQ","short_pith_number":"pith:FURGR5AG","schema_version":"1.0","canonical_sha256":"2d2268f40622f97b23600584f7eb988c24efc9919354a9764b525aa570fa5862","source":{"kind":"arxiv","id":"2507.23010","version":1},"attestation_state":"computed","paper":{"title":"Investigating the Invertibility of Multimodal Latent Spaces: Limitations of Optimization-Based Methods","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.SD","eess.AS"],"primary_cat":"cs.LG","authors_text":"Siwoo Park","submitted_at":"2025-07-30T18:19:11Z","abstract_excerpt":"This paper investigates the inverse capabilities and broader utility of multimodal latent spaces within task-specific AI (Artificial Intelligence) models. While these models excel at their designed forward tasks (e.g., text-to-image generation, audio-to-text transcription), their potential for inverse mappings remains largely unexplored. We propose an optimization-based framework to infer input characteristics from desired outputs, applying it bidirectionally across Text-Image (BLIP, Flux.1-dev) and Text-Audio (Whisper-Large-V3, Chatterbox-TTS) modalities.\n  Our central hypothesis posits that "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2507.23010","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-07-30T18:19:11Z","cross_cats_sorted":["cs.AI","cs.CV","cs.SD","eess.AS"],"title_canon_sha256":"7cf2d32680bfa9a0cf435a74eb8cf6d824f238c4eb071d58c893edebcb87b4f8","abstract_canon_sha256":"60287ffab2304448461556ee11c49da0fceb4b23508eda13a821a1fba09bf98a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:46:03.659166Z","signature_b64":"C/o004A2yDlSY+VLBHvavlfCjlYMHySg7Lw0hQzk++4rno5Efq4/qNLyDb2nQkaO0zPGyYOXmFVEru1aztioDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2d2268f40622f97b23600584f7eb988c24efc9919354a9764b525aa570fa5862","last_reissued_at":"2026-07-05T11:46:03.658632Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:46:03.658632Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Investigating the Invertibility of Multimodal Latent Spaces: Limitations of Optimization-Based Methods","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.SD","eess.AS"],"primary_cat":"cs.LG","authors_text":"Siwoo Park","submitted_at":"2025-07-30T18:19:11Z","abstract_excerpt":"This paper investigates the inverse capabilities and broader utility of multimodal latent spaces within task-specific AI (Artificial Intelligence) models. While these models excel at their designed forward tasks (e.g., text-to-image generation, audio-to-text transcription), their potential for inverse mappings remains largely unexplored. We propose an optimization-based framework to infer input characteristics from desired outputs, applying it bidirectionally across Text-Image (BLIP, Flux.1-dev) and Text-Audio (Whisper-Large-V3, Chatterbox-TTS) modalities.\n  Our central hypothesis posits that "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.23010","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2507.23010/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2507.23010","created_at":"2026-07-05T11:46:03.658691+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.23010v1","created_at":"2026-07-05T11:46:03.658691+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.23010","created_at":"2026-07-05T11:46:03.658691+00:00"},{"alias_kind":"pith_short_12","alias_value":"FURGR5AGEL4X","created_at":"2026-07-05T11:46:03.658691+00:00"},{"alias_kind":"pith_short_16","alias_value":"FURGR5AGEL4XWI3A","created_at":"2026-07-05T11:46:03.658691+00:00"},{"alias_kind":"pith_short_8","alias_value":"FURGR5AG","created_at":"2026-07-05T11:46:03.658691+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.15842","citing_title":"Informationally Compressive Anonymization: Non-Degrading Sensitive Input Protection for Privacy-Preserving Supervised Machine Learning","ref_index":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FURGR5AGEL4XWI3AAWCPP24YRQ","json":"https://pith.science/pith/FURGR5AGEL4XWI3AAWCPP24YRQ.json","graph_json":"https://pith.science/api/pith-number/FURGR5AGEL4XWI3AAWCPP24YRQ/graph.json","events_json":"https://pith.science/api/pith-number/FURGR5AGEL4XWI3AAWCPP24YRQ/events.json","paper":"https://pith.science/paper/FURGR5AG"},"agent_actions":{"view_html":"https://pith.science/pith/FURGR5AGEL4XWI3AAWCPP24YRQ","download_json":"https://pith.science/pith/FURGR5AGEL4XWI3AAWCPP24YRQ.json","view_paper":"https://pith.science/paper/FURGR5AG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.23010&json=true","fetch_graph":"https://pith.science/api/pith-number/FURGR5AGEL4XWI3AAWCPP24YRQ/graph.json","fetch_events":"https://pith.science/api/pith-number/FURGR5AGEL4XWI3AAWCPP24YRQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FURGR5AGEL4XWI3AAWCPP24YRQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FURGR5AGEL4XWI3AAWCPP24YRQ/action/storage_attestation","attest_author":"https://pith.science/pith/FURGR5AGEL4XWI3AAWCPP24YRQ/action/author_attestation","sign_citation":"https://pith.science/pith/FURGR5AGEL4XWI3AAWCPP24YRQ/action/citation_signature","submit_replication":"https://pith.science/pith/FURGR5AGEL4XWI3AAWCPP24YRQ/action/replication_record"}},"created_at":"2026-07-05T11:46:03.658691+00:00","updated_at":"2026-07-05T11:46:03.658691+00:00"}