{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RV26YPLDFN5GV55JSXWLBJZEV2","short_pith_number":"pith:RV26YPLD","schema_version":"1.0","canonical_sha256":"8d75ec3d632b7a6af7a995ecb0a724ae94e78a3d8d6c60dd46444027553f141d","source":{"kind":"arxiv","id":"2503.06183","version":2},"attestation_state":"computed","paper":{"title":"Lightweight Software Kernels and Hardware Extensions for Efficient Sparse Deep Neural Networks on Microcontrollers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.DC","cs.PF"],"primary_cat":"cs.LG","authors_text":"Alessio Burrello, Daniele Jahier Pagliari, Francesco Conti, Francesco Daghero, Luca Benini, Massimo Poncino","submitted_at":"2025-03-08T11:59:12Z","abstract_excerpt":"The acceleration of pruned Deep Neural Networks (DNNs) on edge devices such as Microcontrollers (MCUs) is a challenging task, given the tight area- and power-constraints of these devices. In this work, we propose a three-fold contribution to address this problem. First, we design a set of optimized software kernels for N:M pruned layers, targeting ultra-low-power, multicore RISC-V MCUs, which are up to 2.1x and 3.4x faster than their dense counterparts at 1:8 and 1:16 sparsity, respectively. Then, we implement a lightweight Instruction-Set Architecture (ISA) extension to accelerate the indirec"},"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":"2503.06183","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-03-08T11:59:12Z","cross_cats_sorted":["cs.AI","cs.DC","cs.PF"],"title_canon_sha256":"45701a345acb2f98a907a9e329ec327af19dc438454373b0075861e912150de2","abstract_canon_sha256":"cf22e43f05c2406a2410415db4079153147435f38137c880339978d653075dd0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:34:17.332213Z","signature_b64":"8DdcwL3OdHP1xEpOTQfeN/OKAbHlwv6WRBTZJ2PedVDmIPwdEBHu3iNXHVaxxt6SOszdR6e19DBbb6EXTrQRDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8d75ec3d632b7a6af7a995ecb0a724ae94e78a3d8d6c60dd46444027553f141d","last_reissued_at":"2026-07-05T10:34:17.331336Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:34:17.331336Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Lightweight Software Kernels and Hardware Extensions for Efficient Sparse Deep Neural Networks on Microcontrollers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.DC","cs.PF"],"primary_cat":"cs.LG","authors_text":"Alessio Burrello, Daniele Jahier Pagliari, Francesco Conti, Francesco Daghero, Luca Benini, Massimo Poncino","submitted_at":"2025-03-08T11:59:12Z","abstract_excerpt":"The acceleration of pruned Deep Neural Networks (DNNs) on edge devices such as Microcontrollers (MCUs) is a challenging task, given the tight area- and power-constraints of these devices. In this work, we propose a three-fold contribution to address this problem. First, we design a set of optimized software kernels for N:M pruned layers, targeting ultra-low-power, multicore RISC-V MCUs, which are up to 2.1x and 3.4x faster than their dense counterparts at 1:8 and 1:16 sparsity, respectively. Then, we implement a lightweight Instruction-Set Architecture (ISA) extension to accelerate the indirec"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.06183","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2503.06183/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":"2503.06183","created_at":"2026-07-05T10:34:17.331458+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.06183v2","created_at":"2026-07-05T10:34:17.331458+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.06183","created_at":"2026-07-05T10:34:17.331458+00:00"},{"alias_kind":"pith_short_12","alias_value":"RV26YPLDFN5G","created_at":"2026-07-05T10:34:17.331458+00:00"},{"alias_kind":"pith_short_16","alias_value":"RV26YPLDFN5GV55J","created_at":"2026-07-05T10:34:17.331458+00:00"},{"alias_kind":"pith_short_8","alias_value":"RV26YPLD","created_at":"2026-07-05T10:34:17.331458+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RV26YPLDFN5GV55JSXWLBJZEV2","json":"https://pith.science/pith/RV26YPLDFN5GV55JSXWLBJZEV2.json","graph_json":"https://pith.science/api/pith-number/RV26YPLDFN5GV55JSXWLBJZEV2/graph.json","events_json":"https://pith.science/api/pith-number/RV26YPLDFN5GV55JSXWLBJZEV2/events.json","paper":"https://pith.science/paper/RV26YPLD"},"agent_actions":{"view_html":"https://pith.science/pith/RV26YPLDFN5GV55JSXWLBJZEV2","download_json":"https://pith.science/pith/RV26YPLDFN5GV55JSXWLBJZEV2.json","view_paper":"https://pith.science/paper/RV26YPLD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.06183&json=true","fetch_graph":"https://pith.science/api/pith-number/RV26YPLDFN5GV55JSXWLBJZEV2/graph.json","fetch_events":"https://pith.science/api/pith-number/RV26YPLDFN5GV55JSXWLBJZEV2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RV26YPLDFN5GV55JSXWLBJZEV2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RV26YPLDFN5GV55JSXWLBJZEV2/action/storage_attestation","attest_author":"https://pith.science/pith/RV26YPLDFN5GV55JSXWLBJZEV2/action/author_attestation","sign_citation":"https://pith.science/pith/RV26YPLDFN5GV55JSXWLBJZEV2/action/citation_signature","submit_replication":"https://pith.science/pith/RV26YPLDFN5GV55JSXWLBJZEV2/action/replication_record"}},"created_at":"2026-07-05T10:34:17.331458+00:00","updated_at":"2026-07-05T10:34:17.331458+00:00"}