]> 2026-09-08T16:34:23+09:00 Lesson:a neural network accelerator for mobile application processors 72ce8850 3000 ko 작거나 불규칙한 layer가 포함된 error-tolerant mobile media neural network에 해당한다. Inter-neuron과 intra-neuron parallelism을 모두 활용할 수 있는 mobile neural-network accelerator를 설계한다. S3ResearchAgent medium Publication scope: international. Lab publication metadata: 1. Doo Young Kim, Jin Min Kim, Hakbeom Jang, Jinkyu Jeong, and Jae W. Lee, "A Neural Network Accelerator for Mobile Application Processors," IEEE Transactions on Consumer Electronics, vol. 61, no. 4, pp. 555-563, Nov. 2015 Verification level: official_abstract. Sources checked: https://doi.org/10.1109/TCE.2015.7389812 https://yonsei.elsevierpure.com/en/publications/a-neural-network-accelerator-for-mobile-application-processors/ 2026-07-16T15:06:32Z 2461238.129537 Publication record 1: Doo Young Kim, Jin Min Kim, Hakbeom Jang, Jinkyu Jeong, and Jae W. Lee, "A Neural Network Accelerator for Mobile Application Processors," IEEE Transactions on Consumer Electronics, vol. 61, no. 4, pp. 555-563, Nov. 2015 공개 초록상 5개 application만 평가됐으며 model 종류, accuracy loss, silicon/process 조건과 area overhead는 검증하지 못했다. Technical interpretation: Layer 폭이 작아 coarse parallelism이 부족하면 한 neuron 내부 reduction까지 분할해 PE utilization을 유지할 수 있다. A Neural Network Accelerator for Mobile Application Processors 5개 application에서 general-purpose CPU 대비 평균 126% speedup, inter-neuron-only state-of-the-art accelerator 대비 23% speedup, 해당 accelerator 대비 energy 22% 절감을 보고한다. Neuron 수가 적은 layer에서 inter-neuron parallelism만 사용하는 accelerator의 PE underutilization을 줄일 수 있는가? lab Layer 폭이 작아 coarse parallelism이 부족하면 한 neuron 내부 reduction까지 분할해 PE utilization을 유지할 수 있다. Draft 2026-07-18T14:58:15Z 2461240.1237847 2026-07-18T14:58:15Z 2461240.1237847 a neural network accelerator for mobile application processors 72ce8850 3000 ev_e4aa49c5d478491a 2026-07-16T15:06:32Z 2461238.129537 S3ResearchAgent Doo Young Kim, Jin Min Kim, Hakbeom Jang, Jinkyu Jeong, and Jae W. Lee, "A Neural Network Accelerator for Mobile Application Processors," IEEE Transactions on Consumer Electronics, vol. 61, no. 4, pp. 555-563, Nov. 2015 paper Source: Yonsei University Computer Systems Laboratory publication list supplied by the user. Manifestation 1 of 1. not_recorded a neural network accelerator for mobile application processors 72ce8850# 11b1d0c372948f6f113ca767288bba5b 3000 canonical-paper-v2-72ce8850 2026-07-18T05:11:23Z 2461239.7162384 S3ResearchAgent Doo Young Kim, Jin Min Kim, Hakbeom Jang, Jinkyu Jeong, and Jae W. Lee, "A Neural Network Accelerator for Mobile Application Processors" IEEE Transactions on Consumer Electronics, vol. 61, no. 4, pp. 555-563, Nov. 2015 paper 정본 Lesson 보강 근거. 검토 원본: Lesson:technical_review_a_neural_network_accelerator_for_mobile_application_processors_d4383cb5. 확인 범위: official_abstract. 확인한 자료: https://doi.org/10.1109/TCE.2015.7389812 ; https://yonsei.elsevierpure.com/en/publications/a-neural-network-accelerator-for-mobile-application-processors/. 질문, 방법, 평가, 해석, 재사용 교훈, 적용 범위와 한계를 같은 Lesson 본문에 통합했습니다. official_abstract a neural network accelerator for mobile application processors 72ce8850# cf359c418d711b68bf66225161e921f3 3000 2026-07-18T14:58:15Z 2461240.1237847 S3ResearchAgent verify_c4bb2406d7d79095a787 metadata_only context 수집 manifest의 실패 원장만 보존되어 원문 주장을 검증하지 못함. inconclusive 689990da723c183d73e5102f6b5a7c8972497dfa210f09dd79bb20472fb857d5 R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=689990da723c183d73e5102f6b5a7c8972497dfa210f09dd79bb20472fb857d5 보존 파일 manifest.json ev_e4aa49c5d478491a 78ad2ad593af2a2a23cde7fc31cfb7c4c09bec26982b7c03db62d18521bf1c8f a neural network accelerator for mobile application processors 72ce8850# 60aa11c9ee1a78d94838fe39cce26cbb 3000 2026-07-18T14:58:15Z 2461240.1237847 S3ResearchAgent verify_cf346518b18d38910f4f metadata_only context 수집 manifest의 실패 원장만 보존되어 원문 주장을 검증하지 못함. inconclusive 689990da723c183d73e5102f6b5a7c8972497dfa210f09dd79bb20472fb857d5 R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=689990da723c183d73e5102f6b5a7c8972497dfa210f09dd79bb20472fb857d5 보존 파일 manifest.json canonical-paper-v2-72ce8850 28d83f275ddc7163c56dcb9c243f20aeee8d08153bde00bd77009ef2feba3de9 a neural network accelerator for mobile application processors 72ce8850# 1b850b3117a8e1263f27a7cd38392bf1 Lesson:technical review a neural network accelerator for mobile application processors d4383cb5 3000 ko technical review a neural network accelerator for mobile application processors d4383cb5 Supersedes 102 ko Supersedes