{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,11]],"date-time":"2026-07-11T06:12:30Z","timestamp":1783750350455,"version":"3.55.0"},"reference-count":59,"publisher":"Association for Computing Machinery (ACM)","issue":"4","license":[{"start":{"date-parts":[[2027,1,26]],"date-time":"2027-01-26T00:00:00Z","timestamp":1800921600000},"content-version":"vor","delay-in-days":312,"URL":"http:\/\/www.acm.org\/publications\/policies\/copyright_policy#Background"}],"funder":[{"DOI":"10.13039\/100000185","name":"DARPA","doi-asserted-by":"crossref","award":["AMP, N6600120C4020; FIRE, P000050426"],"award-info":[{"award-number":["AMP, N6600120C4020; FIRE, P000050426"]}],"id":[{"id":"10.13039\/100000185","id-type":"DOI","asserted-by":"crossref"}]},{"name":"NSF","award":["FDT-Biotech, 2436801"],"award-info":[{"award-number":["FDT-Biotech, 2436801"]}]},{"DOI":"10.13039\/100007028","name":"Helmsley Charitable Trust","doi-asserted-by":"crossref","award":["2-SRA-2017-503-M-B"],"award-info":[{"award-number":["2-SRA-2017-503-M-B"]}],"id":[{"id":"10.13039\/100007028","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Des. Autom. Electron. Syst."],"published-print":{"date-parts":[[2026,7,31]]},"abstract":"<jats:p>Understanding the physical laws that govern real-world data is fundamental to achieving safe and explainable operation of mission-critical autonomous systems (MCAS)\u2014a cornerstone of physical AI. Model Recovery (MR) serves as a key mechanism for inferring governing equations directly from data. However, deploying MR within MCAS must contend with stringent latency, computational, and power constraints, making edge-AI acceleration essential. Field-Programmable Gate Arrays (FPGAs) provide an attractive hardware substrate due to their reconfigurability and real-time processing capability. Yet, existing MR techniques often rely on computationally intensive nonlinear optimization or the numerical solution of multiple ordinary differential equations (ODEs) embedded within neural architectures, leading to high compute and memory overheads.<\/jats:p>\n                  <jats:p>To address these challenges, this article introduces MERINDA (Model Recovery in Dynamic Architecture), an FPGA-accelerated MR framework that integrates Gated Recurrent Units (GRUs) with an invertible mapping parameterized as a linear combination of nonlinear basis functions. We theoretically establish that MERINDA is functionally equivalent to Neural Ordinary Differential Equation (NODE)-based MR architectures, while eliminating the need for repeated ODE integration during training and inference. Empirical evaluations on benchmark datasets against Extracting sparse Model from ImpLicit dYnamics (EMILY), Sparse Identification of Nonlinear Dynamics (SINDY), and Physics-Informed Neural Networks with Sparse Regression (PINN+SR) demonstrate that MERINDA achieves comparable accuracy with substantial gains in processing speed, energy efficiency, and DRAM utilization.<\/jats:p>\n                  <jats:p>We further analyze the energy\u2013memory tradeoff in MERINDA, showing that optimizing one resource directly impacts the other under fixed accuracy constraints. Using mixed-integer programming, we derive resource-aware optimal hyperparameters and construct the Pareto front that compares FPGA edge deployment against Mobile GPU (M-GPU) and server-class Graphics Processing Units (GPU) platforms. Our results highlight MERINDA\u2019s energy efficiency, reduced training time, and smaller memory footprint, reinforcing its viability for deployment in resource-constrained autonomous systems.<\/jats:p>","DOI":"10.1145\/3793545","type":"journal-article","created":{"date-parts":[[2026,1,26]],"date-time":"2026-01-26T12:01:06Z","timestamp":1769428866000},"page":"1-36","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["MERINDA: Model Recovery in FPGA-Based Dynamic Architecture for Edge and Physical AI"],"prefix":"10.1145","volume":"31","author":[{"ORCID":"https:\/\/orcid.org\/0009-0001-2639-7283","authenticated-orcid":false,"given":"Bin","family":"Xu","sequence":"first","affiliation":[{"name":"Arizona State University Ira A Fulton Schools of Engineering","place":["Tempe, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6529-1644","authenticated-orcid":false,"given":"Ayan","family":"Banerjee","sequence":"additional","affiliation":[{"name":"Arizona State University","place":["Tempe, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6108-5584","authenticated-orcid":false,"given":"Sandeep","family":"Gupta","sequence":"additional","affiliation":[{"name":"Computer Science and Engineering, Arizona State University","place":["Tempe, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2026,3,20]]},"reference":[{"key":"e_1_3_1_2_2","doi-asserted-by":"publisher","DOI":"10.1145\/3715323"},{"key":"e_1_3_1_3_2","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2022.3144977"},{"key":"e_1_3_1_4_2","first-page":"1","volume-title":"Proceedings of the ECAI Workshop on Machine Learning Meets Differential Eqns: From Theory to Applications","author":"Banerjee Ayan","year":"2024","unstructured":"Ayan Banerjee, Sandeep Gupta, C. Coelho, B. Zimmering, M. F. P. Costa, L. L. Ferr\u00e1s, and O. Niggemann. 2024. EMILY: Extracting sparse model from implicit dynamics. In Proceedings of the ECAI Workshop on Machine Learning Meets Differential Eqns: From Theory to Applications. 1\u201311."},{"key":"e_1_3_1_5_2","doi-asserted-by":"publisher","DOI":"10.3233\/FAIA240556"},{"key":"e_1_3_1_6_2","first-page":"21325","article-title":"Neural flows: Efficient alternative to neural ODEs","volume":"34","author":"Bilo\u0161 Marin","year":"2021","unstructured":"Marin Bilo\u0161, Johanna Sommer, Syama Sundar Rangapuram, Tim Januschowski, and Stephan G\u00fcnnemann. 2021. Neural flows: Efficient alternative to neural ODEs. Advances in Neural Information Processing Systems 34 (2021), 21325\u201321337.","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_3_1_7_2","doi-asserted-by":"publisher","DOI":"10.1145\/343477.343502"},{"key":"e_1_3_1_8_2","doi-asserted-by":"publisher","DOI":"10.1145\/3543622.3573044"},{"key":"e_1_3_1_9_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICAIIS49377.2020.9194894"},{"key":"e_1_3_1_10_2","unstructured":"Ricky T. Q. Chen Yulia Rubanova Jesse Bettencourt and David K. Duvenaud. 2018. Neural ordinary differential equations. Advances in Neural Information Processing Systems 31 (2018) 6572\u20136583."},{"key":"e_1_3_1_11_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-021-26434-1"},{"key":"e_1_3_1_12_2","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2021.3124125"},{"key":"e_1_3_1_13_2","unstructured":"Jungwook Choi Zhuo Wang Swagath Venkataramani Pierce I-Jen Chuang Vijayalakshmi Srinivasan and Kailash Gopalakrishnan. 2018. Pact: Parameterized clipping activation for quantized neural networks. arXiv preprint arXiv:1805.06085 (2018)."},{"key":"e_1_3_1_14_2","first-page":"93","volume-title":"Proceedings of the 2018 IEEE 26th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM)","author":"Cong Jason","year":"2018","unstructured":"Jason Cong, Zhenman Fang, Michael Lo, Hanrui Wang, Jingxian Xu, and Shaochong Zhang. 2018. Understanding performance differences of FPGAs and GPUs. In Proceedings of the 2018 IEEE 26th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM). IEEE, 93\u201396."},{"key":"e_1_3_1_15_2","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2017.2705434"},{"key":"e_1_3_1_16_2","doi-asserted-by":"crossref","unstructured":"Achraf El Bouazzaoui Abdelkader Hadjoudja Omar Mouhib and Nazha Cherkaoui. 2024. FPGA-based ML adaptive accelerator: A partial reconfiguration approach for optimized ML accelerator utilization. Array 21 3 (2024) 100337.","DOI":"10.1016\/j.array.2024.100337"},{"key":"e_1_3_1_17_2","doi-asserted-by":"publisher","DOI":"10.1109\/AICAS48895.2020.9074001"},{"key":"e_1_3_1_18_2","doi-asserted-by":"publisher","DOI":"10.1201\/9781003162810-13"},{"key":"e_1_3_1_19_2","unstructured":"Tom Gunter Zirui Wang Chong Wang Ruoming Pang Andy Narayanan Aonan Zhang Bowen Zhang Chen Chen Chung-Cheng Chiu David Qiu et\u00a0al. 2024. Apple intelligence foundation language models. arXiv preprint arXiv:2407.21075 (2024)."},{"key":"e_1_3_1_20_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.neucom.2017.03.044"},{"key":"e_1_3_1_21_2","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v35i9.16936"},{"key":"e_1_3_1_22_2","doi-asserted-by":"publisher","DOI":"10.1145\/3447818.3460371"},{"key":"e_1_3_1_23_2","doi-asserted-by":"publisher","DOI":"10.1039\/B902479B"},{"issue":"241","key":"e_1_3_1_24_2","first-page":"1","article-title":"Sparsity in deep learning: Pruning and growth for efficient inference and training in neural networks","volume":"22","author":"Hoefler Torsten","year":"2021","unstructured":"Torsten Hoefler, Dan Alistarh, Tal Ben-Nun, Nikoli Dryden, and Alexandra Peste. 2021. Sparsity in deep learning: Pruning and growth for efficient inference and training in neural networks. Journal of Machine Learning Research 22, 241 (2021), 1\u2013124.","journal-title":"Journal of Machine Learning Research"},{"key":"e_1_3_1_25_2","doi-asserted-by":"publisher","DOI":"10.1007\/s00521-022-07783-4"},{"key":"e_1_3_1_26_2","doi-asserted-by":"publisher","DOI":"10.1145\/3447548.3467139"},{"key":"e_1_3_1_27_2","doi-asserted-by":"publisher","DOI":"10.1098\/rspa.2018.0335"},{"key":"e_1_3_1_28_2","volume-title":"Measuring and Navigating the Gap between FPGAs and ASICs","author":"Kuon Ian","year":"2007","unstructured":"Ian Kuon. 2007. Measuring and Navigating the Gap between FPGAs and ASICs. Ph. D. Dissertation."},{"key":"e_1_3_1_29_2","doi-asserted-by":"publisher","DOI":"10.1109\/TWC.2019.2946140"},{"key":"e_1_3_1_30_2","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA.2019.00028"},{"key":"e_1_3_1_31_2","volume-title":"Proceedings of the International Conference on Learning Representations","author":"Liu Shiwei","year":"2022","unstructured":"Shiwei Liu, Tianlong Chen, Xiaohan Chen, Li Shen, Decebal Constantin Mocanu, Zhangyang Wang, and Mykola Pechenizkiy. 2022. The unreasonable effectiveness of random pruning: Return of the most naive baseline for sparse training. In Proceedings of the International Conference on Learning Representations. Retrieved from https:\/\/openreview.net\/forum?id=VBZJ_3tz-t"},{"key":"e_1_3_1_32_2","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-023-01824-8"},{"key":"e_1_3_1_33_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.procs.2018.10.155"},{"key":"e_1_3_1_34_2","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2018.2815603"},{"key":"e_1_3_1_35_2","first-page":"71","volume-title":"Proceedings of the CEUR Workshop.","author":"Marling Cindy","year":"2020","unstructured":"Cindy Marling and Razvan Bunescu. 2020. The OhioT1DM dataset for blood glucose level prediction: Update 2020. In Proceedings of the CEUR Workshop. NIH Public Access, 71."},{"key":"e_1_3_1_36_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-35713-9"},{"key":"e_1_3_1_37_2","doi-asserted-by":"publisher","DOI":"10.5555\/993491"},{"key":"e_1_3_1_38_2","volume-title":"Proceedings of the 38th Annual Conference on Neural Information Processing Systems","author":"Petersen Felix","year":"2024","unstructured":"Felix Petersen, Hilde Kuehne, Christian Borgelt, Julian Welzel, and Stefano Ermon. 2024. Convolutional differentiable logic gate networks. In Proceedings of the 38th Annual Conference on Neural Information Processing Systems. Retrieved from https:\/\/openreview.net\/forum?id=4bKEFyUHT4"},{"key":"e_1_3_1_39_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.patcog.2020.107281"},{"key":"e_1_3_1_40_2","doi-asserted-by":"publisher","DOI":"10.1021\/acsphotonics.1c02016"},{"key":"e_1_3_1_41_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICFPT51103.2020.00012"},{"key":"e_1_3_1_42_2","doi-asserted-by":"publisher","DOI":"10.1109\/ASAP52443.2021.00025"},{"key":"e_1_3_1_43_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-46493-0_32"},{"key":"e_1_3_1_44_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.neunet.2022.07.023"},{"key":"e_1_3_1_45_2","doi-asserted-by":"publisher","DOI":"10.1109\/MPRV.2017.37"},{"key":"e_1_3_1_46_2","doi-asserted-by":"publisher","DOI":"10.1109\/JIOT.2016.2579198"},{"key":"e_1_3_1_47_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-023-36799-6"},{"key":"e_1_3_1_48_2","doi-asserted-by":"crossref","unstructured":"Subhrajit Sinha Sai P. Nandanoori and Enoch Yeung. 2020. Koopman Operator Methods for Global Phase Space Exploration of Equivariant Dynamical Systems. IFAC-PapersOnLine 53 2 (2020) 1150\u20131155.","DOI":"10.1016\/j.ifacol.2020.12.1322"},{"key":"e_1_3_1_49_2","doi-asserted-by":"publisher","DOI":"10.1145\/3120895.3120919"},{"key":"e_1_3_1_50_2","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2025.3541486"},{"key":"e_1_3_1_51_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.mbs.2018.12.012"},{"key":"e_1_3_1_52_2","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPSW52791.2021.00021"},{"key":"e_1_3_1_53_2","unstructured":"Bin Xu Ayan Banerjee and Sandeep K. S. Gupta. 2025. Enabling physical AI at the edge: Hardware-accelerated recovery of system dynamics. arXiv preprint arXiv:2512.23767 (2025)."},{"key":"e_1_3_1_54_2","unstructured":"Bin Xu Ayan Banerjee and Sandeep K. S. Gupta. 2025. Hardware acceleration for neural networks: A comprehensive survey. arXiv preprint arXiv:2512.23914 (2025)."},{"key":"e_1_3_1_55_2","doi-asserted-by":"crossref","unstructured":"B. Xu A. Banerjee and Sandeep K. S. Gupta. 2025. Fast online digital twinning on FPGA for mission critical applications. In Proceedings of the MILCOM 2025-2025 IEEE Military Communications Conference (MILCOM).","DOI":"10.1109\/MILCOM64451.2025.11310524"},{"key":"e_1_3_1_56_2","doi-asserted-by":"crossref","unstructured":"Bin Xu Ayan Banerjee and Sandeep K. S. Gupta. 2025. Model recovery at the edge under resource constraints for physical AI. In Proceedings of the 28th European Conference on Artificial Intelligence (ECAI\u201925). Frontiers in Artificial Intelligence and Applications. IOS Press. 413 (2025) 3904\u20133911.","DOI":"10.3233\/FAIA251275"},{"key":"e_1_3_1_57_2","doi-asserted-by":"publisher","DOI":"10.1109\/MWSCAS53549.2025.11244413"},{"key":"e_1_3_1_58_2","doi-asserted-by":"publisher","DOI":"10.1145\/3649329.3656497"},{"key":"e_1_3_1_59_2","doi-asserted-by":"publisher","DOI":"10.1145\/3289393"},{"key":"e_1_3_1_60_2","doi-asserted-by":"publisher","DOI":"10.1145\/279232.279236"}],"container-title":["ACM Transactions on Design Automation of Electronic Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3793545","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3793545","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,20]],"date-time":"2026-03-20T09:12:24Z","timestamp":1773997944000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3793545"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,20]]},"references-count":59,"journal-issue":{"issue":"4","published-print":{"date-parts":[[2026,7,31]]}},"alternative-id":["10.1145\/3793545"],"URL":"https:\/\/doi.org\/10.1145\/3793545","relation":{},"ISSN":["1084-4309","1557-7309"],"issn-type":[{"value":"1084-4309","type":"print"},{"value":"1557-7309","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,3,20]]},"assertion":[{"value":"2025-03-05","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-01-02","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-03-20","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}