{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2014:J6HABZN4HVQILV53T7EF6F4PFK","short_pith_number":"pith:J6HABZN4","schema_version":"1.0","canonical_sha256":"4f8e00e5bc3d6085d7bb9fc85f178f2a98ab05aa6979093be5a1ea5b7874c2fe","source":{"kind":"arxiv","id":"1402.3337","version":5},"attestation_state":"computed","paper":{"title":"Zero-bias autoencoders and the benefits of co-adapting features","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.LG","cs.NE"],"primary_cat":"stat.ML","authors_text":"David Krueger, Kishore Konda, Roland Memisevic","submitted_at":"2014-02-13T23:37:39Z","abstract_excerpt":"Regularized training of an autoencoder typically results in hidden unit biases that take on large negative values. We show that negative biases are a natural result of using a hidden layer whose responsibility is to both represent the input data and act as a selection mechanism that ensures sparsity of the representation. We then show that negative biases impede the learning of data distributions whose intrinsic dimensionality is high. We also propose a new activation function that decouples the two roles of the hidden layer and that allows us to learn representations on data with very high in"},"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":"1402.3337","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2014-02-13T23:37:39Z","cross_cats_sorted":["cs.CV","cs.LG","cs.NE"],"title_canon_sha256":"f20388113d6e31c79e015254dec841cdd083a1bada3a3dccbb317e949c85043f","abstract_canon_sha256":"956cec7e64892b27889a64ba692af87efb7125a42a9a800d9e9f9016eb172af0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:19:25.711787Z","signature_b64":"xtjYFIDP5xrlt8otOb2jGpS28M2+xwO6bLCUuTD2rWNdD19z5S7GG5J22zaWLION+2xAvq9k0VFFlykJN0ZOBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4f8e00e5bc3d6085d7bb9fc85f178f2a98ab05aa6979093be5a1ea5b7874c2fe","last_reissued_at":"2026-05-18T02:19:25.711084Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:19:25.711084Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Zero-bias autoencoders and the benefits of co-adapting features","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.LG","cs.NE"],"primary_cat":"stat.ML","authors_text":"David Krueger, Kishore Konda, Roland Memisevic","submitted_at":"2014-02-13T23:37:39Z","abstract_excerpt":"Regularized training of an autoencoder typically results in hidden unit biases that take on large negative values. We show that negative biases are a natural result of using a hidden layer whose responsibility is to both represent the input data and act as a selection mechanism that ensures sparsity of the representation. We then show that negative biases impede the learning of data distributions whose intrinsic dimensionality is high. We also propose a new activation function that decouples the two roles of the hidden layer and that allows us to learn representations on data with very high in"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1402.3337","kind":"arxiv","version":5},"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":"1402.3337","created_at":"2026-05-18T02:19:25.711186+00:00"},{"alias_kind":"arxiv_version","alias_value":"1402.3337v5","created_at":"2026-05-18T02:19:25.711186+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1402.3337","created_at":"2026-05-18T02:19:25.711186+00:00"},{"alias_kind":"pith_short_12","alias_value":"J6HABZN4HVQI","created_at":"2026-05-18T12:28:33.132498+00:00"},{"alias_kind":"pith_short_16","alias_value":"J6HABZN4HVQILV53","created_at":"2026-05-18T12:28:33.132498+00:00"},{"alias_kind":"pith_short_8","alias_value":"J6HABZN4","created_at":"2026-05-18T12:28:33.132498+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.23040","citing_title":"Steered Generation via Gradient-Based Optimization on Sparse Query Features","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2406.04093","citing_title":"Scaling and evaluating sparse autoencoders","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08183","citing_title":"Sparsity Hurts: Simple Linear Adapter Can Boost Generalized Category Discovery","ref_index":38,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/J6HABZN4HVQILV53T7EF6F4PFK","json":"https://pith.science/pith/J6HABZN4HVQILV53T7EF6F4PFK.json","graph_json":"https://pith.science/api/pith-number/J6HABZN4HVQILV53T7EF6F4PFK/graph.json","events_json":"https://pith.science/api/pith-number/J6HABZN4HVQILV53T7EF6F4PFK/events.json","paper":"https://pith.science/paper/J6HABZN4"},"agent_actions":{"view_html":"https://pith.science/pith/J6HABZN4HVQILV53T7EF6F4PFK","download_json":"https://pith.science/pith/J6HABZN4HVQILV53T7EF6F4PFK.json","view_paper":"https://pith.science/paper/J6HABZN4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1402.3337&json=true","fetch_graph":"https://pith.science/api/pith-number/J6HABZN4HVQILV53T7EF6F4PFK/graph.json","fetch_events":"https://pith.science/api/pith-number/J6HABZN4HVQILV53T7EF6F4PFK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J6HABZN4HVQILV53T7EF6F4PFK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J6HABZN4HVQILV53T7EF6F4PFK/action/storage_attestation","attest_author":"https://pith.science/pith/J6HABZN4HVQILV53T7EF6F4PFK/action/author_attestation","sign_citation":"https://pith.science/pith/J6HABZN4HVQILV53T7EF6F4PFK/action/citation_signature","submit_replication":"https://pith.science/pith/J6HABZN4HVQILV53T7EF6F4PFK/action/replication_record"}},"created_at":"2026-05-18T02:19:25.711186+00:00","updated_at":"2026-05-18T02:19:25.711186+00:00"}