{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2016:WODWTCHNUZ6QTCEAY5IS7R25IO","short_pith_number":"pith:WODWTCHN","schema_version":"1.0","canonical_sha256":"b3876988eda67d098880c7512fc75d438bcc22e62ecdf263272e8f13da9d95e4","source":{"kind":"arxiv","id":"1606.01981","version":1},"attestation_state":"computed","paper":{"title":"Deep neural networks are robust to weight binarization and other non-linear distortions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.LG"],"primary_cat":"cs.NE","authors_text":"Dharmendra Modha, John Arthur, Paul Merolla, Rathinakumar Appuswamy, Steve K. Esser","submitted_at":"2016-06-07T00:28:42Z","abstract_excerpt":"Recent results show that deep neural networks achieve excellent performance even when, during training, weights are quantized and projected to a binary representation. Here, we show that this is just the tip of the iceberg: these same networks, during testing, also exhibit a remarkable robustness to distortions beyond quantization, including additive and multiplicative noise, and a class of non-linear projections where binarization is just a special case. To quantify this robustness, we show that one such network achieves 11% test error on CIFAR-10 even with 0.68 effective bits per weight. Fur"},"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":"1606.01981","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.NE","submitted_at":"2016-06-07T00:28:42Z","cross_cats_sorted":["cs.CV","cs.LG"],"title_canon_sha256":"b9a4127e8676f5df7ff77bbf8517b40d226f19b210c44ef6a8011a6d3688011d","abstract_canon_sha256":"686caa93132dcfc58f90e5787684f95751fb453905a246063f72c119fbe5c928"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T01:12:44.888687Z","signature_b64":"r/oZlkcID2BSD2q3QW/VhKBANKdJrt+wqEnUw3Tn641ZHVxpcamjqNuPCZWuNHvw5d3YeWPMYLIIf0RAqdfCBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b3876988eda67d098880c7512fc75d438bcc22e62ecdf263272e8f13da9d95e4","last_reissued_at":"2026-05-18T01:12:44.888301Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T01:12:44.888301Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Deep neural networks are robust to weight binarization and other non-linear distortions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.LG"],"primary_cat":"cs.NE","authors_text":"Dharmendra Modha, John Arthur, Paul Merolla, Rathinakumar Appuswamy, Steve K. Esser","submitted_at":"2016-06-07T00:28:42Z","abstract_excerpt":"Recent results show that deep neural networks achieve excellent performance even when, during training, weights are quantized and projected to a binary representation. Here, we show that this is just the tip of the iceberg: these same networks, during testing, also exhibit a remarkable robustness to distortions beyond quantization, including additive and multiplicative noise, and a class of non-linear projections where binarization is just a special case. To quantify this robustness, we show that one such network achieves 11% test error on CIFAR-10 even with 0.68 effective bits per weight. Fur"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1606.01981","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":""},"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":"1606.01981","created_at":"2026-05-18T01:12:44.888371+00:00"},{"alias_kind":"arxiv_version","alias_value":"1606.01981v1","created_at":"2026-05-18T01:12:44.888371+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1606.01981","created_at":"2026-05-18T01:12:44.888371+00:00"},{"alias_kind":"pith_short_12","alias_value":"WODWTCHNUZ6Q","created_at":"2026-05-18T12:30:51.357362+00:00"},{"alias_kind":"pith_short_16","alias_value":"WODWTCHNUZ6QTCEA","created_at":"2026-05-18T12:30:51.357362+00:00"},{"alias_kind":"pith_short_8","alias_value":"WODWTCHN","created_at":"2026-05-18T12:30:51.357362+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"1907.06916","citing_title":"Single-bit-per-weight deep convolutional neural networks without batch-normalization layers for embedded systems","ref_index":9,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WODWTCHNUZ6QTCEAY5IS7R25IO","json":"https://pith.science/pith/WODWTCHNUZ6QTCEAY5IS7R25IO.json","graph_json":"https://pith.science/api/pith-number/WODWTCHNUZ6QTCEAY5IS7R25IO/graph.json","events_json":"https://pith.science/api/pith-number/WODWTCHNUZ6QTCEAY5IS7R25IO/events.json","paper":"https://pith.science/paper/WODWTCHN"},"agent_actions":{"view_html":"https://pith.science/pith/WODWTCHNUZ6QTCEAY5IS7R25IO","download_json":"https://pith.science/pith/WODWTCHNUZ6QTCEAY5IS7R25IO.json","view_paper":"https://pith.science/paper/WODWTCHN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1606.01981&json=true","fetch_graph":"https://pith.science/api/pith-number/WODWTCHNUZ6QTCEAY5IS7R25IO/graph.json","fetch_events":"https://pith.science/api/pith-number/WODWTCHNUZ6QTCEAY5IS7R25IO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WODWTCHNUZ6QTCEAY5IS7R25IO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WODWTCHNUZ6QTCEAY5IS7R25IO/action/storage_attestation","attest_author":"https://pith.science/pith/WODWTCHNUZ6QTCEAY5IS7R25IO/action/author_attestation","sign_citation":"https://pith.science/pith/WODWTCHNUZ6QTCEAY5IS7R25IO/action/citation_signature","submit_replication":"https://pith.science/pith/WODWTCHNUZ6QTCEAY5IS7R25IO/action/replication_record"}},"created_at":"2026-05-18T01:12:44.888371+00:00","updated_at":"2026-05-18T01:12:44.888371+00:00"}