{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2015:CNS54ZTLBZPKQKUL4YCGISCJE3","short_pith_number":"pith:CNS54ZTL","schema_version":"1.0","canonical_sha256":"1365de666b0e5ea82a8be60464484926c00ad83670bf3a6b813cc7ac1203c8ce","source":{"kind":"arxiv","id":"1505.00853","version":2},"attestation_state":"computed","paper":{"title":"Empirical Evaluation of Rectified Activations in Convolutional Network","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Bing Xu, Mu Li, Naiyan Wang, Tianqi Chen","submitted_at":"2015-05-05T01:16:39Z","abstract_excerpt":"In this paper we investigate the performance of different types of rectified activation functions in convolutional neural network: standard rectified linear unit (ReLU), leaky rectified linear unit (Leaky ReLU), parametric rectified linear unit (PReLU) and a new randomized leaky rectified linear units (RReLU). We evaluate these activation function on standard image classification task. Our experiments suggest that incorporating a non-zero slope for negative part in rectified activation units could consistently improve the results. Thus our findings are negative on the common belief that sparsi"},"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":"1505.00853","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2015-05-05T01:16:39Z","cross_cats_sorted":["cs.CV","stat.ML"],"title_canon_sha256":"162e5fea8ee8ab7ffbc03baa1ccb6bd8dff086346b5ebadfb6fcb552a9bdb0e0","abstract_canon_sha256":"54347e0144509315263115d84424704f4defa0368e2e52d5c381cfca1bd37a95"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T01:25:55.353531Z","signature_b64":"XmEyWqSyoEXg0ylYEq3CmlZnThxTw2OnGzcg3+Gt4IiGA21UMnAgxa7Ovr/cLVm5YeNjJNyaBUkB1vqf5f5CBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1365de666b0e5ea82a8be60464484926c00ad83670bf3a6b813cc7ac1203c8ce","last_reissued_at":"2026-05-18T01:25:55.352852Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T01:25:55.352852Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Empirical Evaluation of Rectified Activations in Convolutional Network","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Bing Xu, Mu Li, Naiyan Wang, Tianqi Chen","submitted_at":"2015-05-05T01:16:39Z","abstract_excerpt":"In this paper we investigate the performance of different types of rectified activation functions in convolutional neural network: standard rectified linear unit (ReLU), leaky rectified linear unit (Leaky ReLU), parametric rectified linear unit (PReLU) and a new randomized leaky rectified linear units (RReLU). We evaluate these activation function on standard image classification task. Our experiments suggest that incorporating a non-zero slope for negative part in rectified activation units could consistently improve the results. Thus our findings are negative on the common belief that sparsi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1505.00853","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":""},"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":"1505.00853","created_at":"2026-05-18T01:25:55.352952+00:00"},{"alias_kind":"arxiv_version","alias_value":"1505.00853v2","created_at":"2026-05-18T01:25:55.352952+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1505.00853","created_at":"2026-05-18T01:25:55.352952+00:00"},{"alias_kind":"pith_short_12","alias_value":"CNS54ZTLBZPK","created_at":"2026-05-18T12:29:17.054201+00:00"},{"alias_kind":"pith_short_16","alias_value":"CNS54ZTLBZPKQKUL","created_at":"2026-05-18T12:29:17.054201+00:00"},{"alias_kind":"pith_short_8","alias_value":"CNS54ZTL","created_at":"2026-05-18T12:29:17.054201+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":18,"internal_anchor_count":12,"sample":[{"citing_arxiv_id":"1906.09610","citing_title":"Improving Description-based Person Re-identification by Multi-granularity Image-text Alignments","ref_index":40,"is_internal_anchor":true},{"citing_arxiv_id":"1907.00068","citing_title":"On Reducing Negative Jacobian Determinant of the Deformation Predicted by Deep Registration Networks","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"1907.02994","citing_title":"Deep learning in ultrasound imaging","ref_index":100,"is_internal_anchor":true},{"citing_arxiv_id":"1907.02959","citing_title":"High-throughput Onboard Hyperspectral Image Compression with Ground-based CNN Reconstruction","ref_index":36,"is_internal_anchor":true},{"citing_arxiv_id":"1907.05006","citing_title":"Two-stream Spatiotemporal Feature for Video QA Task","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"1907.07772","citing_title":"Modern CNNs for IoT Based Farms","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"1907.08070","citing_title":"Discriminative Embedding Autoencoder with a Regressor Feedback for Zero-Shot Learning","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2309.09550","citing_title":"Adaptive Reorganization of Neural Pathways for Continual Learning with Spiking Neural Networks","ref_index":67,"is_internal_anchor":true},{"citing_arxiv_id":"2410.09355","citing_title":"On Divergence Measures for Training GFlowNets","ref_index":98,"is_internal_anchor":true},{"citing_arxiv_id":"2412.20091","citing_title":"Gamma-Ray Burst Light Curve Reconstruction: A Comparative Machine and Deep Learning Analysis","ref_index":101,"is_internal_anchor":true},{"citing_arxiv_id":"2509.22562","citing_title":"Activation Function Design Sustains Plasticity in Continual Learning","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2510.04378","citing_title":"Score-based Greedy Search for Structure Identification of Partially Observed Linear Causal Models","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2604.16426","citing_title":"Functional Similarity Metric for Neural Networks: Overcoming Parametric Ambiguity via Activation Region Analysis","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"1511.06434","citing_title":"Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08183","citing_title":"Sparsity Hurts: Simple Linear Adapter Can Boost Generalized Category Discovery","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"1710.05941","citing_title":"Searching for Activation Functions","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21119","citing_title":"Materialistic RIR: Material Conditioned Realistic RIR Generation","ref_index":79,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06959","citing_title":"Locally Near Optimal Piecewise Linear Regression in High Dimensions via Difference of Max-Affine Functions","ref_index":184,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CNS54ZTLBZPKQKUL4YCGISCJE3","json":"https://pith.science/pith/CNS54ZTLBZPKQKUL4YCGISCJE3.json","graph_json":"https://pith.science/api/pith-number/CNS54ZTLBZPKQKUL4YCGISCJE3/graph.json","events_json":"https://pith.science/api/pith-number/CNS54ZTLBZPKQKUL4YCGISCJE3/events.json","paper":"https://pith.science/paper/CNS54ZTL"},"agent_actions":{"view_html":"https://pith.science/pith/CNS54ZTLBZPKQKUL4YCGISCJE3","download_json":"https://pith.science/pith/CNS54ZTLBZPKQKUL4YCGISCJE3.json","view_paper":"https://pith.science/paper/CNS54ZTL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1505.00853&json=true","fetch_graph":"https://pith.science/api/pith-number/CNS54ZTLBZPKQKUL4YCGISCJE3/graph.json","fetch_events":"https://pith.science/api/pith-number/CNS54ZTLBZPKQKUL4YCGISCJE3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CNS54ZTLBZPKQKUL4YCGISCJE3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CNS54ZTLBZPKQKUL4YCGISCJE3/action/storage_attestation","attest_author":"https://pith.science/pith/CNS54ZTLBZPKQKUL4YCGISCJE3/action/author_attestation","sign_citation":"https://pith.science/pith/CNS54ZTLBZPKQKUL4YCGISCJE3/action/citation_signature","submit_replication":"https://pith.science/pith/CNS54ZTLBZPKQKUL4YCGISCJE3/action/replication_record"}},"created_at":"2026-05-18T01:25:55.352952+00:00","updated_at":"2026-05-18T01:25:55.352952+00:00"}