{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2016:A23CZDWC7NDFAVXLZXZYZ2NDFB","short_pith_number":"pith:A23CZDWC","schema_version":"1.0","canonical_sha256":"06b62c8ec2fb465056ebcdf38ce9a3286269b00a01461d3afe2f230a69dfc9ae","source":{"kind":"arxiv","id":"1611.01491","version":6},"attestation_state":"computed","paper":{"title":"Understanding Deep Neural Networks with Rectified Linear Units","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cond-mat.dis-nn","cs.AI","cs.CC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Amitabh Basu, Anirbit Mukherjee, Poorya Mianjy, Raman Arora","submitted_at":"2016-11-04T18:54:50Z","abstract_excerpt":"In this paper we investigate the family of functions representable by deep neural networks (DNN) with rectified linear units (ReLU). We give an algorithm to train a ReLU DNN with one hidden layer to *global optimality* with runtime polynomial in the data size albeit exponential in the input dimension. Further, we improve on the known lower bounds on size (from exponential to super exponential) for approximating a ReLU deep net function by a shallower ReLU net. Our gap theorems hold for smoothly parametrized families of \"hard\" functions, contrary to countable, discrete families known in the lit"},"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":"1611.01491","kind":"arxiv","version":6},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2016-11-04T18:54:50Z","cross_cats_sorted":["cond-mat.dis-nn","cs.AI","cs.CC","stat.ML"],"title_canon_sha256":"264421d2c8543a1a88723cde82a59933db2bb982307cd3b787766f8a4828eec0","abstract_canon_sha256":"c2d8eef87890e2579534b210e2d12061a337be448c7795e170e70983ed99a21d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:22:21.158152Z","signature_b64":"42x7ST4wzpFU7Y8t8OnQFkdJWGkYSxiEAU1wdFsUuhu5sFj3zuN5gDtq9kIffjnN4nCM2YSgAbcOgjvxobHRAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"06b62c8ec2fb465056ebcdf38ce9a3286269b00a01461d3afe2f230a69dfc9ae","last_reissued_at":"2026-05-18T00:22:21.157503Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:22:21.157503Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding Deep Neural Networks with Rectified Linear Units","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cond-mat.dis-nn","cs.AI","cs.CC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Amitabh Basu, Anirbit Mukherjee, Poorya Mianjy, Raman Arora","submitted_at":"2016-11-04T18:54:50Z","abstract_excerpt":"In this paper we investigate the family of functions representable by deep neural networks (DNN) with rectified linear units (ReLU). We give an algorithm to train a ReLU DNN with one hidden layer to *global optimality* with runtime polynomial in the data size albeit exponential in the input dimension. Further, we improve on the known lower bounds on size (from exponential to super exponential) for approximating a ReLU deep net function by a shallower ReLU net. Our gap theorems hold for smoothly parametrized families of \"hard\" functions, contrary to countable, discrete families known in the lit"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1611.01491","kind":"arxiv","version":6},"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":"1611.01491","created_at":"2026-05-18T00:22:21.157585+00:00"},{"alias_kind":"arxiv_version","alias_value":"1611.01491v6","created_at":"2026-05-18T00:22:21.157585+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1611.01491","created_at":"2026-05-18T00:22:21.157585+00:00"},{"alias_kind":"pith_short_12","alias_value":"A23CZDWC7NDF","created_at":"2026-05-18T12:30:04.600751+00:00"},{"alias_kind":"pith_short_16","alias_value":"A23CZDWC7NDFAVXL","created_at":"2026-05-18T12:30:04.600751+00:00"},{"alias_kind":"pith_short_8","alias_value":"A23CZDWC","created_at":"2026-05-18T12:30:04.600751+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":14,"internal_anchor_count":10,"sample":[{"citing_arxiv_id":"2607.07778","citing_title":"A law of robustness for two-layer neural networks with arbitrary weights","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2606.27142","citing_title":"Estimation of High Dimensional Bounded Discrete Graphical Models via Regularized Generalized Score Matching","ref_index":248,"is_internal_anchor":true},{"citing_arxiv_id":"2606.10089","citing_title":"A Theory on Flow Matching with Neural Networks","ref_index":115,"is_internal_anchor":true},{"citing_arxiv_id":"2606.02228","citing_title":"Bayesian meta-learning for modeling Alzheimer's disease progression","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"1907.00560","citing_title":"On Symmetry and Initialization for Neural Networks","ref_index":4,"is_internal_anchor":true},{"citing_arxiv_id":"2212.08989","citing_title":"Deep learning applied to computational mechanics: A comprehensive review, state of the art, and the classics","ref_index":283,"is_internal_anchor":true},{"citing_arxiv_id":"2510.23912","citing_title":"Key and Value Weights Are Probably All You Need: On the Necessity of the Query, Key, Value weight Triplet in Self-Attention Transformers","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2510.26576","citing_title":"\"Show Me You Comply... Without Showing Me Anything\": Zero-Knowledge Software Auditing for AI-Enabled Systems","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14308","citing_title":"Reconstructing the Stripping History of the Sagittarius Stream with Neural Networks","ref_index":55,"is_internal_anchor":true},{"citing_arxiv_id":"2401.01335","citing_title":"Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models","ref_index":88,"is_internal_anchor":true},{"citing_arxiv_id":"2605.08877","citing_title":"Non-Uniqueness of Solutions in Neural Variational Methods","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08213","citing_title":"Low-Cost Stereo Vision for Robust 3D Positioning of Thin Radiata Pine Branches in Autonomous Drone Pruning","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08877","citing_title":"Non-Uniqueness of Solutions in Neural Variational Methods","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16480","citing_title":"Positioning radiata pine branches requiring pruning by drone stereo vision","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/A23CZDWC7NDFAVXLZXZYZ2NDFB","json":"https://pith.science/pith/A23CZDWC7NDFAVXLZXZYZ2NDFB.json","graph_json":"https://pith.science/api/pith-number/A23CZDWC7NDFAVXLZXZYZ2NDFB/graph.json","events_json":"https://pith.science/api/pith-number/A23CZDWC7NDFAVXLZXZYZ2NDFB/events.json","paper":"https://pith.science/paper/A23CZDWC"},"agent_actions":{"view_html":"https://pith.science/pith/A23CZDWC7NDFAVXLZXZYZ2NDFB","download_json":"https://pith.science/pith/A23CZDWC7NDFAVXLZXZYZ2NDFB.json","view_paper":"https://pith.science/paper/A23CZDWC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1611.01491&json=true","fetch_graph":"https://pith.science/api/pith-number/A23CZDWC7NDFAVXLZXZYZ2NDFB/graph.json","fetch_events":"https://pith.science/api/pith-number/A23CZDWC7NDFAVXLZXZYZ2NDFB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/A23CZDWC7NDFAVXLZXZYZ2NDFB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/A23CZDWC7NDFAVXLZXZYZ2NDFB/action/storage_attestation","attest_author":"https://pith.science/pith/A23CZDWC7NDFAVXLZXZYZ2NDFB/action/author_attestation","sign_citation":"https://pith.science/pith/A23CZDWC7NDFAVXLZXZYZ2NDFB/action/citation_signature","submit_replication":"https://pith.science/pith/A23CZDWC7NDFAVXLZXZYZ2NDFB/action/replication_record"}},"created_at":"2026-05-18T00:22:21.157585+00:00","updated_at":"2026-05-18T00:22:21.157585+00:00"}