{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:DTJIJLJEQ55OPIOGFEV3T6PCOK","short_pith_number":"pith:DTJIJLJE","schema_version":"1.0","canonical_sha256":"1cd284ad24877ae7a1c6292bb9f9e272bf2cc6baa8438b6048612256b45b4455","source":{"kind":"arxiv","id":"2507.14736","version":1},"attestation_state":"computed","paper":{"title":"Balancing Expressivity and Robustness: Constrained Rational Activations for Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alex Lewandowski, Clare Lyle, Mateusz Ostaszewski, Micha{\\l} Bortkiewicz, Rafa{\\l} Surdej","submitted_at":"2025-07-19T19:53:08Z","abstract_excerpt":"Trainable activation functions, whose parameters are optimized alongside network weights, offer increased expressivity compared to fixed activation functions. Specifically, trainable activation functions defined as ratios of polynomials (rational functions) have been proposed to enhance plasticity in reinforcement learning. However, their impact on training stability remains unclear. In this work, we study trainable rational activations in both reinforcement and continual learning settings. We find that while their flexibility enhances adaptability, it can also introduce instability, leading t"},"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":"2507.14736","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-19T19:53:08Z","cross_cats_sorted":[],"title_canon_sha256":"7d5a9b041a2d98b462872013525367d6d67e192dec0212df74d851d0d18366cb","abstract_canon_sha256":"17c350a00042082bf0fbbe34b5f2d29c00499779e7bbb165f93ab5da5900c48a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:40:12.624198Z","signature_b64":"KNFIuTuH7oYjSMcrugYB9/khilEsb9Ogq2hCUMgLKm/ch1qm878ILteqhaBJtBEIkxdfOO6YLDQncl3Az7NVCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1cd284ad24877ae7a1c6292bb9f9e272bf2cc6baa8438b6048612256b45b4455","last_reissued_at":"2026-07-05T11:40:12.623673Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:40:12.623673Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Balancing Expressivity and Robustness: Constrained Rational Activations for Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alex Lewandowski, Clare Lyle, Mateusz Ostaszewski, Micha{\\l} Bortkiewicz, Rafa{\\l} Surdej","submitted_at":"2025-07-19T19:53:08Z","abstract_excerpt":"Trainable activation functions, whose parameters are optimized alongside network weights, offer increased expressivity compared to fixed activation functions. Specifically, trainable activation functions defined as ratios of polynomials (rational functions) have been proposed to enhance plasticity in reinforcement learning. However, their impact on training stability remains unclear. In this work, we study trainable rational activations in both reinforcement and continual learning settings. We find that while their flexibility enhances adaptability, it can also introduce instability, leading t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.14736","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2507.14736/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"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":"2507.14736","created_at":"2026-07-05T11:40:12.623732+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.14736v1","created_at":"2026-07-05T11:40:12.623732+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.14736","created_at":"2026-07-05T11:40:12.623732+00:00"},{"alias_kind":"pith_short_12","alias_value":"DTJIJLJEQ55O","created_at":"2026-07-05T11:40:12.623732+00:00"},{"alias_kind":"pith_short_16","alias_value":"DTJIJLJEQ55OPIOG","created_at":"2026-07-05T11:40:12.623732+00:00"},{"alias_kind":"pith_short_8","alias_value":"DTJIJLJE","created_at":"2026-07-05T11:40:12.623732+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2509.22562","citing_title":"Activation Function Design Sustains Plasticity in Continual Learning","ref_index":24,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DTJIJLJEQ55OPIOGFEV3T6PCOK","json":"https://pith.science/pith/DTJIJLJEQ55OPIOGFEV3T6PCOK.json","graph_json":"https://pith.science/api/pith-number/DTJIJLJEQ55OPIOGFEV3T6PCOK/graph.json","events_json":"https://pith.science/api/pith-number/DTJIJLJEQ55OPIOGFEV3T6PCOK/events.json","paper":"https://pith.science/paper/DTJIJLJE"},"agent_actions":{"view_html":"https://pith.science/pith/DTJIJLJEQ55OPIOGFEV3T6PCOK","download_json":"https://pith.science/pith/DTJIJLJEQ55OPIOGFEV3T6PCOK.json","view_paper":"https://pith.science/paper/DTJIJLJE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.14736&json=true","fetch_graph":"https://pith.science/api/pith-number/DTJIJLJEQ55OPIOGFEV3T6PCOK/graph.json","fetch_events":"https://pith.science/api/pith-number/DTJIJLJEQ55OPIOGFEV3T6PCOK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DTJIJLJEQ55OPIOGFEV3T6PCOK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DTJIJLJEQ55OPIOGFEV3T6PCOK/action/storage_attestation","attest_author":"https://pith.science/pith/DTJIJLJEQ55OPIOGFEV3T6PCOK/action/author_attestation","sign_citation":"https://pith.science/pith/DTJIJLJEQ55OPIOGFEV3T6PCOK/action/citation_signature","submit_replication":"https://pith.science/pith/DTJIJLJEQ55OPIOGFEV3T6PCOK/action/replication_record"}},"created_at":"2026-07-05T11:40:12.623732+00:00","updated_at":"2026-07-05T11:40:12.623732+00:00"}