{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:EKO4D2QGBX6BSSL2LLQABM3JCO","short_pith_number":"pith:EKO4D2QG","schema_version":"1.0","canonical_sha256":"229dc1ea060dfc19497a5ae000b3691396ca3379b3b26c1d30b29cd921f0afed","source":{"kind":"arxiv","id":"2006.03463","version":2},"attestation_state":"computed","paper":{"title":"Sponge Examples: Energy-Latency Attacks on Neural Networks","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.CR","stat.ML"],"primary_cat":"cs.LG","authors_text":"Daniel Bates, Ilia Shumailov, Nicolas Papernot, Robert Mullins, Ross Anderson, Yiren Zhao","submitted_at":"2020-06-05T14:10:09Z","abstract_excerpt":"The high energy costs of neural network training and inference led to the use of acceleration hardware such as GPUs and TPUs. While this enabled us to train large-scale neural networks in datacenters and deploy them on edge devices, the focus so far is on average-case performance. In this work, we introduce a novel threat vector against neural networks whose energy consumption or decision latency are critical. We show how adversaries can exploit carefully crafted $\\boldsymbol{sponge}~\\boldsymbol{examples}$, which are inputs designed to maximise energy consumption and latency.\n  We mount two va"},"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":"2006.03463","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2020-06-05T14:10:09Z","cross_cats_sorted":["cs.CL","cs.CR","stat.ML"],"title_canon_sha256":"ef4cf38828ee371017969ef6e7dbb787cb1d2571aa7d25ef7e793ab59d0e6beb","abstract_canon_sha256":"4382783b5e3fdd8a893290a41b0bdb9970e3aa2a500afe91992f96fefbeb7eb2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:39:41.411878Z","signature_b64":"fpvj6qJfMw1WHmpVTuOY14niNPU++0ImcVhWllLQgkh1Rw7pMMuc6gb8SgajiePSWI3KTX7pkZ2eAdYvLz1CAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"229dc1ea060dfc19497a5ae000b3691396ca3379b3b26c1d30b29cd921f0afed","last_reissued_at":"2026-07-05T02:39:41.411453Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:39:41.411453Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sponge Examples: Energy-Latency Attacks on Neural Networks","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.CR","stat.ML"],"primary_cat":"cs.LG","authors_text":"Daniel Bates, Ilia Shumailov, Nicolas Papernot, Robert Mullins, Ross Anderson, Yiren Zhao","submitted_at":"2020-06-05T14:10:09Z","abstract_excerpt":"The high energy costs of neural network training and inference led to the use of acceleration hardware such as GPUs and TPUs. While this enabled us to train large-scale neural networks in datacenters and deploy them on edge devices, the focus so far is on average-case performance. In this work, we introduce a novel threat vector against neural networks whose energy consumption or decision latency are critical. We show how adversaries can exploit carefully crafted $\\boldsymbol{sponge}~\\boldsymbol{examples}$, which are inputs designed to maximise energy consumption and latency.\n  We mount two va"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2006.03463","kind":"arxiv","version":2},"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/2006.03463/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":"2006.03463","created_at":"2026-07-05T02:39:41.411511+00:00"},{"alias_kind":"arxiv_version","alias_value":"2006.03463v2","created_at":"2026-07-05T02:39:41.411511+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.03463","created_at":"2026-07-05T02:39:41.411511+00:00"},{"alias_kind":"pith_short_12","alias_value":"EKO4D2QGBX6B","created_at":"2026-07-05T02:39:41.411511+00:00"},{"alias_kind":"pith_short_16","alias_value":"EKO4D2QGBX6BSSL2","created_at":"2026-07-05T02:39:41.411511+00:00"},{"alias_kind":"pith_short_8","alias_value":"EKO4D2QG","created_at":"2026-07-05T02:39:41.411511+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.18893","citing_title":"Reality Check: A New Evaluation Ecosystem Is Necessary to Understand AI's Real World Effects","ref_index":88,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EKO4D2QGBX6BSSL2LLQABM3JCO","json":"https://pith.science/pith/EKO4D2QGBX6BSSL2LLQABM3JCO.json","graph_json":"https://pith.science/api/pith-number/EKO4D2QGBX6BSSL2LLQABM3JCO/graph.json","events_json":"https://pith.science/api/pith-number/EKO4D2QGBX6BSSL2LLQABM3JCO/events.json","paper":"https://pith.science/paper/EKO4D2QG"},"agent_actions":{"view_html":"https://pith.science/pith/EKO4D2QGBX6BSSL2LLQABM3JCO","download_json":"https://pith.science/pith/EKO4D2QGBX6BSSL2LLQABM3JCO.json","view_paper":"https://pith.science/paper/EKO4D2QG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2006.03463&json=true","fetch_graph":"https://pith.science/api/pith-number/EKO4D2QGBX6BSSL2LLQABM3JCO/graph.json","fetch_events":"https://pith.science/api/pith-number/EKO4D2QGBX6BSSL2LLQABM3JCO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EKO4D2QGBX6BSSL2LLQABM3JCO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EKO4D2QGBX6BSSL2LLQABM3JCO/action/storage_attestation","attest_author":"https://pith.science/pith/EKO4D2QGBX6BSSL2LLQABM3JCO/action/author_attestation","sign_citation":"https://pith.science/pith/EKO4D2QGBX6BSSL2LLQABM3JCO/action/citation_signature","submit_replication":"https://pith.science/pith/EKO4D2QGBX6BSSL2LLQABM3JCO/action/replication_record"}},"created_at":"2026-07-05T02:39:41.411511+00:00","updated_at":"2026-07-05T02:39:41.411511+00:00"}