{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2020:EERPHK5U2HKNEJG4SZ5QMZV3XZ","short_pith_number":"pith:EERPHK5U","canonical_record":{"source":{"id":"2006.10637","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-18T16:06:18Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"f9be48c8fd6765a1c2be668df5c09ae24b8210db46f0d83a6085b0772dec9bbd","abstract_canon_sha256":"bde6367331ef4af9851a56e8da3df391cb761f43f5829d4a091e6604519bc4b8"},"schema_version":"1.0"},"canonical_sha256":"2122f3abb4d1d4d224dc967b0666bbbe604964abc44ea8ac4e61548d32858289","source":{"kind":"arxiv","id":"2006.10637","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2006.10637","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"arxiv_version","alias_value":"2006.10637v3","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.10637","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"pith_short_12","alias_value":"EERPHK5U2HKN","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"EERPHK5U2HKNEJG4","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"EERPHK5U","created_at":"2026-05-18T12:33:33Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2020:EERPHK5U2HKNEJG4SZ5QMZV3XZ","target":"record","payload":{"canonical_record":{"source":{"id":"2006.10637","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-18T16:06:18Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"f9be48c8fd6765a1c2be668df5c09ae24b8210db46f0d83a6085b0772dec9bbd","abstract_canon_sha256":"bde6367331ef4af9851a56e8da3df391cb761f43f5829d4a091e6604519bc4b8"},"schema_version":"1.0"},"canonical_sha256":"2122f3abb4d1d4d224dc967b0666bbbe604964abc44ea8ac4e61548d32858289","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:13.566561Z","signature_b64":"aGAIl2uLV0arzoZBpA6CLfoUBbAw6WivcRxxrm/4LspBn2g1O9crGHfEUPxE5mec79xhGlaqk4/C//ZIyHmWDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2122f3abb4d1d4d224dc967b0666bbbe604964abc44ea8ac4e61548d32858289","last_reissued_at":"2026-05-17T23:38:13.565876Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:13.565876Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2006.10637","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:13Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"DdaM8hnEsYDnw+9k2wQdwCamTCZiIB1Zsu5/KaTGhgoobfVdzRFZSIPJq2w89onpJQfocg6omEFbJXs5bpiNCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-01T04:38:56.230962Z"},"content_sha256":"b1c1e46da7dbf0779351f5a56069f9b3aa8429bb70c70fd151ea506b6c946038","schema_version":"1.0","event_id":"sha256:b1c1e46da7dbf0779351f5a56069f9b3aa8429bb70c70fd151ea506b6c946038"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2020:EERPHK5U2HKNEJG4SZ5QMZV3XZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Temporal Graph Networks for Deep Learning on Dynamic Graphs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Temporal Graph Networks combine memory modules with graph operators to outperform prior dynamic graph methods while using less computation.","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Ben Chamberlain, Davide Eynard, Emanuele Rossi, Fabrizio Frasca, Federico Monti, Michael Bronstein","submitted_at":"2020-06-18T16:06:18Z","abstract_excerpt":"Graph Neural Networks (GNNs) have recently become increasingly popular due to their ability to learn complex systems of relations or interactions arising in a broad spectrum of problems ranging from biology and particle physics to social networks and recommendation systems. Despite the plethora of different models for deep learning on graphs, few approaches have been proposed thus far for dealing with graphs that present some sort of dynamic nature (e.g. evolving features or connectivity over time). In this paper, we present Temporal Graph Networks (TGNs), a generic, efficient framework for de"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Thanks to a novel combination of memory modules and graph-based operators, TGNs are able to significantly outperform previous approaches being at the same time more computationally efficient.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That dynamic graphs can be represented effectively as sequences of timed events and that the introduced memory modules plus graph operators will capture the necessary temporal dependencies without introducing prohibitive computational or modeling limitations.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Temporal Graph Networks combine memory modules and graph operators to learn on dynamic graphs as timed event sequences, outperforming prior methods on transductive and inductive tasks while unifying earlier models as special cases.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Temporal Graph Networks combine memory modules with graph operators to outperform prior dynamic graph methods while using less computation.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"295c6474c8ae3fa707f1e5f4a8c80d6dbae4248ac758354a0f65b76df355a700"},"source":{"id":"2006.10637","kind":"arxiv","version":3},"verdict":{"id":"8f045f95-5ed4-4d56-bd3b-e560a308685f","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T16:58:39.051759Z","strongest_claim":"Thanks to a novel combination of memory modules and graph-based operators, TGNs are able to significantly outperform previous approaches being at the same time more computationally efficient.","one_line_summary":"Temporal Graph Networks combine memory modules and graph operators to learn on dynamic graphs as timed event sequences, outperforming prior methods on transductive and inductive tasks while unifying earlier models as special cases.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That dynamic graphs can be represented effectively as sequences of timed events and that the introduced memory modules plus graph operators will capture the necessary temporal dependencies without introducing prohibitive computational or modeling limitations.","pith_extraction_headline":"Temporal Graph Networks combine memory modules with graph operators to outperform prior dynamic graph methods while using less computation."},"references":{"count":141,"sample":[{"doi":"","year":null,"title":"Interaction networks for learning about objects, relations and physics , author=. NIPS , pages=","work_id":"c59f27bb-f5a6-446c-9328-19edeb6150b0","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"IEEE Signal Process","work_id":"306ddf58-1118-4b4b-a6c6-a8145303741b","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"and Leskovec, Jure , title =","work_id":"184a1d06-f093-4ef6-9342-c96d0bc36233","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"and Ying, Rex and Leskovec, Jure , title =","work_id":"03ac5d41-e101-49d6-b53d-c11264414030","ref_index":6,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Journal of Machine Learning Research , year =","work_id":"94c538c8-bc48-49e9-94e2-c2102ead1925","ref_index":7,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":141,"snapshot_sha256":"d07d001b8bc31459e2f31088f81fc4d7b5695154eaf8821e2ae6f64217f0a653","internal_anchors":10},"formal_canon":{"evidence_count":3,"snapshot_sha256":"4c83cc1b1eaf3307075b14696c2816aecced5e249f43d2d63ffc25ba3377a90e"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"8f045f95-5ed4-4d56-bd3b-e560a308685f"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:13Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"SJatprx7Q4MPtTgfLsbetF6fqyTsBvfKivQFXryLkQzfrjisTzZ2/TzMSRrGs4a+VP/zZjcfCmb+kL6366IkDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-01T04:38:56.231485Z"},"content_sha256":"905d81328dbfb98c427960c5bb48d3b2f055a2cfecc1e51319f336a271a8c958","schema_version":"1.0","event_id":"sha256:905d81328dbfb98c427960c5bb48d3b2f055a2cfecc1e51319f336a271a8c958"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/EERPHK5U2HKNEJG4SZ5QMZV3XZ/bundle.json","state_url":"https://pith.science/pith/EERPHK5U2HKNEJG4SZ5QMZV3XZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/EERPHK5U2HKNEJG4SZ5QMZV3XZ/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-01T04:38:56Z","links":{"resolver":"https://pith.science/pith/EERPHK5U2HKNEJG4SZ5QMZV3XZ","bundle":"https://pith.science/pith/EERPHK5U2HKNEJG4SZ5QMZV3XZ/bundle.json","state":"https://pith.science/pith/EERPHK5U2HKNEJG4SZ5QMZV3XZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/EERPHK5U2HKNEJG4SZ5QMZV3XZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2020:EERPHK5U2HKNEJG4SZ5QMZV3XZ","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"bde6367331ef4af9851a56e8da3df391cb761f43f5829d4a091e6604519bc4b8","cross_cats_sorted":["stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-18T16:06:18Z","title_canon_sha256":"f9be48c8fd6765a1c2be668df5c09ae24b8210db46f0d83a6085b0772dec9bbd"},"schema_version":"1.0","source":{"id":"2006.10637","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2006.10637","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"arxiv_version","alias_value":"2006.10637v3","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.10637","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"pith_short_12","alias_value":"EERPHK5U2HKN","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"EERPHK5U2HKNEJG4","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"EERPHK5U","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:905d81328dbfb98c427960c5bb48d3b2f055a2cfecc1e51319f336a271a8c958","target":"graph","created_at":"2026-05-17T23:38:13Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Thanks to a novel combination of memory modules and graph-based operators, TGNs are able to significantly outperform previous approaches being at the same time more computationally efficient."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That dynamic graphs can be represented effectively as sequences of timed events and that the introduced memory modules plus graph operators will capture the necessary temporal dependencies without introducing prohibitive computational or modeling limitations."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Temporal Graph Networks combine memory modules and graph operators to learn on dynamic graphs as timed event sequences, outperforming prior methods on transductive and inductive tasks while unifying earlier models as special cases."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Temporal Graph Networks combine memory modules with graph operators to outperform prior dynamic graph methods while using less computation."}],"snapshot_sha256":"295c6474c8ae3fa707f1e5f4a8c80d6dbae4248ac758354a0f65b76df355a700"},"formal_canon":{"evidence_count":3,"snapshot_sha256":"4c83cc1b1eaf3307075b14696c2816aecced5e249f43d2d63ffc25ba3377a90e"},"paper":{"abstract_excerpt":"Graph Neural Networks (GNNs) have recently become increasingly popular due to their ability to learn complex systems of relations or interactions arising in a broad spectrum of problems ranging from biology and particle physics to social networks and recommendation systems. Despite the plethora of different models for deep learning on graphs, few approaches have been proposed thus far for dealing with graphs that present some sort of dynamic nature (e.g. evolving features or connectivity over time). In this paper, we present Temporal Graph Networks (TGNs), a generic, efficient framework for de","authors_text":"Ben Chamberlain, Davide Eynard, Emanuele Rossi, Fabrizio Frasca, Federico Monti, Michael Bronstein","cross_cats":["stat.ML"],"headline":"Temporal Graph Networks combine memory modules with graph operators to outperform prior dynamic graph methods while using less computation.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-18T16:06:18Z","title":"Temporal Graph Networks for Deep Learning on Dynamic Graphs"},"references":{"count":141,"internal_anchors":10,"resolved_work":141,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Interaction networks for learning about objects, relations and physics , author=. NIPS , pages=","work_id":"c59f27bb-f5a6-446c-9328-19edeb6150b0","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"IEEE Signal Process","work_id":"306ddf58-1118-4b4b-a6c6-a8145303741b","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"and Leskovec, Jure , title =","work_id":"184a1d06-f093-4ef6-9342-c96d0bc36233","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":6,"title":"and Ying, Rex and Leskovec, Jure , title =","work_id":"03ac5d41-e101-49d6-b53d-c11264414030","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":7,"title":"Journal of Machine Learning Research , year =","work_id":"94c538c8-bc48-49e9-94e2-c2102ead1925","year":null}],"snapshot_sha256":"d07d001b8bc31459e2f31088f81fc4d7b5695154eaf8821e2ae6f64217f0a653"},"source":{"id":"2006.10637","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-17T16:58:39.051759Z","id":"8f045f95-5ed4-4d56-bd3b-e560a308685f","model_set":{"reader":"grok-4.3"},"one_line_summary":"Temporal Graph Networks combine memory modules and graph operators to learn on dynamic graphs as timed event sequences, outperforming prior methods on transductive and inductive tasks while unifying earlier models as special cases.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Temporal Graph Networks combine memory modules with graph operators to outperform prior dynamic graph methods while using less computation.","strongest_claim":"Thanks to a novel combination of memory modules and graph-based operators, TGNs are able to significantly outperform previous approaches being at the same time more computationally efficient.","weakest_assumption":"That dynamic graphs can be represented effectively as sequences of timed events and that the introduced memory modules plus graph operators will capture the necessary temporal dependencies without introducing prohibitive computational or modeling limitations."}},"verdict_id":"8f045f95-5ed4-4d56-bd3b-e560a308685f"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:b1c1e46da7dbf0779351f5a56069f9b3aa8429bb70c70fd151ea506b6c946038","target":"record","created_at":"2026-05-17T23:38:13Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"bde6367331ef4af9851a56e8da3df391cb761f43f5829d4a091e6604519bc4b8","cross_cats_sorted":["stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-18T16:06:18Z","title_canon_sha256":"f9be48c8fd6765a1c2be668df5c09ae24b8210db46f0d83a6085b0772dec9bbd"},"schema_version":"1.0","source":{"id":"2006.10637","kind":"arxiv","version":3}},"canonical_sha256":"2122f3abb4d1d4d224dc967b0666bbbe604964abc44ea8ac4e61548d32858289","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2122f3abb4d1d4d224dc967b0666bbbe604964abc44ea8ac4e61548d32858289","first_computed_at":"2026-05-17T23:38:13.565876Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:13.565876Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"aGAIl2uLV0arzoZBpA6CLfoUBbAw6WivcRxxrm/4LspBn2g1O9crGHfEUPxE5mec79xhGlaqk4/C//ZIyHmWDQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:13.566561Z","signed_message":"canonical_sha256_bytes"},"source_id":"2006.10637","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b1c1e46da7dbf0779351f5a56069f9b3aa8429bb70c70fd151ea506b6c946038","sha256:905d81328dbfb98c427960c5bb48d3b2f055a2cfecc1e51319f336a271a8c958"],"state_sha256":"7ea78ede623ebdee6b7c7483cbb16292a6cbafb3a29c57a7e6507e095e2d3b4e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"SBhC5gd3qMedCmjpw+bDRotbsjaZQcQrPt7rSmAgpjnyeEsSuafso4f7ID45TwR61ooI5h9RQAOuPXr/xFHNCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-01T04:38:56.233832Z","bundle_sha256":"6f924b3cc85dcb594dcae2822edef838e9469f9ee1bc43743e01955b68034839"}}