{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:EHZB3QCK4IGE3ZRIYOFND7CMDZ","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":"682945cf8d4f1107465c0cee4ee65a2ef4f4499e4a34a7a445544088db922129","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-05T12:19:30Z","title_canon_sha256":"14361b957f80377928a6e5dcdd8791b0ec0edb9f8447bffddd270b6042673fc2"},"schema_version":"1.0","source":{"id":"2605.03680","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.03680","created_at":"2026-06-26T00:15:25Z"},{"alias_kind":"arxiv_version","alias_value":"2605.03680v1","created_at":"2026-06-26T00:15:25Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.03680","created_at":"2026-06-26T00:15:25Z"},{"alias_kind":"pith_short_12","alias_value":"EHZB3QCK4IGE","created_at":"2026-06-26T00:15:25Z"},{"alias_kind":"pith_short_16","alias_value":"EHZB3QCK4IGE3ZRI","created_at":"2026-06-26T00:15:25Z"},{"alias_kind":"pith_short_8","alias_value":"EHZB3QCK","created_at":"2026-06-26T00:15:25Z"}],"graph_snapshots":[{"event_id":"sha256:ad01773961673ced88d8f460c3a0305fef6032a38093a9c0f39c84dd1e13fbf5","target":"graph","created_at":"2026-06-26T00:15:25Z","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":3,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"The 1.96M-parameter student recovers 99.8% of the teacher's restoration quality via high-alpha knowledge distillation (alpha = 0.9), achieving a 21.2x parameter reduction while closing the PSNR gap from 1.63 dB to only 0.05 dB."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That restricting the student to NPU-native primitives (3x3 convolutions, ReLU, nearest-neighbor upsampling) combined with progressive context expansion up to 1024x1024 crops will preserve generalization on real-world noisy images without significant quality loss outside the specific Mobile AI 2026 benchmarks."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"A 1.96M-parameter LiteDenoiseNet student model achieves 37.58 dB PSNR on full-resolution real image denoising benchmarks while running in 34-46 ms on mobile NPUs by leveraging NPU-compatible primitives and high-alpha knowledge distillation."}],"snapshot_sha256":"f36e721f99bb769c0246475716a945e5ef32fc83a607ec832d3da8448df10d3b"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-20T13:36:25.249369Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_title_agreement","ran_at":"2026-05-20T00:31:21.545513Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T15:06:14.581799Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.03680/integrity.json","findings":[],"snapshot_sha256":"942c313fa52c2682407f1e9bb4b983f04ed10f0bc90f074dd604a4e60c26fbb0","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"While deep-learning-based image restoration has achieved unprecedented fidelity, deployment on mobile Neural Processing Units (NPUs) remains bottlenecked by operator incompatibility and memory-access overhead. We propose an NPU-aware hardware-algorithm co-design approach for real-world image denoising on mobile NPUs. Our approach employs a high-capacity teacher to supervise a lightweight student network specifically designed to leverage the tiled-memory architectures of modern mobile SoCs. By prioritizing NPU-native primitives -- standard 3x3 convolutions, ReLU activations, and nearest-neighbo","authors_text":"Asad Ahmed, Dmitry Ignatov, Faraz Kayani, Radu Timofte, Sarmad Kayani","cross_cats":["cs.LG"],"headline":"A 1.96M-parameter LiteDenoiseNet student model achieves 37.58 dB PSNR on full-resolution real image denoising benchmarks while running in 34-46 ms on mobile NPUs by leveraging NPU-compatible primitives and high-alpha knowledge distillation.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-05T12:19:30Z","title":"Real Image Denoising with Knowledge Distillation for High-Performance Mobile NPUs"},"references":{"count":30,"internal_anchors":2,"resolved_work":30,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Abdelrahman Abdelhamed, Stephen Lin, and Michael S. Brown. A high-quality denoising dataset for smartphone cameras. InCVPR, pages 1692–1700, 2018. 2","work_id":"4b8cf503-8324-4766-98e7-9286ae010ac7","year":2018},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Abdelrahman Abdelhamed, Radu Timofte, and Michael S. Brown. NTIRE 2019 challenge on real image denoising: Methods and results. InCVPRW, pages 0–0, 2019. 2","work_id":"bb8e0d6a-6bcb-4011-8c92-9e88bed61f0d","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Abdelrahman Abdelhamed, Radu Timofte, and Michael S. Brown. NTIRE 2020 challenge on real image denoising: Dataset, methods and results. InCVPRW, pages 496–497,","work_id":"677e60f8-8340-493b-b3ba-da8a234fff49","year":2020},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Real image denoising with feature attention","work_id":"a438b169-e5e8-41d8-8571-b778c71c47c7","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Hinet: Half instance normalization network for image restoration","work_id":"b0e6b4c1-eb9b-4179-be1d-20b04f665bbc","year":2021}],"snapshot_sha256":"066c1d1e7c61f6c2d4706ed64f9f34412797e21a56352a39d42c53207cd078d6"},"source":{"id":"2605.03680","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-07T17:55:39.329059Z","id":"986ad60d-41d1-4416-8a0a-62f5ebf607a9","model_set":{"reader":"grok-4.3"},"one_line_summary":"A 1.96M-parameter LiteDenoiseNet student model achieves 37.58 dB PSNR on full-resolution real image denoising benchmarks while running in 34-46 ms on mobile NPUs by leveraging NPU-compatible primitives and high-alpha knowledge distillation.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"","strongest_claim":"The 1.96M-parameter student recovers 99.8% of the teacher's restoration quality via high-alpha knowledge distillation (alpha = 0.9), achieving a 21.2x parameter reduction while closing the PSNR gap from 1.63 dB to only 0.05 dB.","weakest_assumption":"That restricting the student to NPU-native primitives (3x3 convolutions, ReLU, nearest-neighbor upsampling) combined with progressive context expansion up to 1024x1024 crops will preserve generalization on real-world noisy images without significant quality loss outside the specific Mobile AI 2026 benchmarks."}},"verdict_id":"986ad60d-41d1-4416-8a0a-62f5ebf607a9"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c478b67390174c432c60ec049e39e3f131012472638239b8f3f5064bbcb6816a","target":"record","created_at":"2026-06-26T00:15:25Z","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":"682945cf8d4f1107465c0cee4ee65a2ef4f4499e4a34a7a445544088db922129","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-05T12:19:30Z","title_canon_sha256":"14361b957f80377928a6e5dcdd8791b0ec0edb9f8447bffddd270b6042673fc2"},"schema_version":"1.0","source":{"id":"2605.03680","kind":"arxiv","version":1}},"canonical_sha256":"21f21dc04ae20c4de628c38ad1fc4c1e4a70dd268889dcc74786ba971bdbaf4f","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"21f21dc04ae20c4de628c38ad1fc4c1e4a70dd268889dcc74786ba971bdbaf4f","first_computed_at":"2026-06-26T00:15:25.742466Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-26T00:15:25.742466Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"PtJsYI+sgaaTpdVti6VOxqCaCzNx34l6XUCn3HZY0gLwlQJCaSrGfA58aQTQB5oIEDYSoWUUXtW2T4dC0bEACw==","signature_status":"signed_v1","signed_at":"2026-06-26T00:15:25.742943Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.03680","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c478b67390174c432c60ec049e39e3f131012472638239b8f3f5064bbcb6816a","sha256:ad01773961673ced88d8f460c3a0305fef6032a38093a9c0f39c84dd1e13fbf5"],"state_sha256":"b16c4a47ebdf938afcda7fb656fa5c16344f65188d4333fc110702eb710ba568"}