{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,23]],"date-time":"2026-04-23T18:45:50Z","timestamp":1776969950333,"version":"3.51.4"},"reference-count":93,"publisher":"Association for Computing Machinery (ACM)","issue":"1","content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["SIGMOD Rec."],"published-print":{"date-parts":[[2026,4,23]]},"abstract":"<jats:p>Modern machine learning (ML) systems leverage multiple backends, including CPUs, GPUs, and distributed execution platforms like Apache Spark or Ray. Depending on workload and cluster characteristics, these systems typically compile an ML pipeline into hybrid plans of in-memory CPU, GPU, and distributed operations. Prior work found that exploratory data science processes exhibit a high degree of redundancy, and accordingly applied tailor-made techniques for reusing intermediates in specific backend scenarios. However, achieving efficient holistic reuse in multibackend data systems remains a challenge due to its tight coupling with other aspects such as memory management, data exchange, and operator scheduling. In this paper, we introduce MEMPHIS, a principled framework for holistic, application-agnostic, multi-backend reuse and memory management. MEMPHIS's core component is a hierarchical lineage-based reuse cache, which acts as a unified abstraction and manages the reuse, recycling, exchange, and cache eviction across different backends. To address challenges of different backends such as lazy evaluation, asynchronous execution, memory allocation overheads, small available memory, and different interconnect bandwidths, we devise a suite of cache management policies. Moreover, we extend an optimizing ML system compiler by special operators and rewrites for asynchronous data exchange, workload-aware speculative cache management, and related operator ordering for concurrent execution. Our experiments across diverse ML tasks and pipelines show improvements of up to 9.6x compared to state-of-the-art ML systems.<\/jats:p>","DOI":"10.1145\/3810900.3810916","type":"journal-article","created":{"date-parts":[[2026,4,23]],"date-time":"2026-04-23T18:16:38Z","timestamp":1776968198000},"page":"85-95","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Lineage-based Reuse and Memory Management for Multi-backend ML Systems"],"prefix":"10.1145","volume":"55","author":[{"given":"Arnab","family":"Phani","sequence":"first","affiliation":[{"name":"TU Berlin &amp; BIFOLD, Berlin, Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Matthias","family":"Boehm","sequence":"additional","affiliation":[{"name":"TU Berlin &amp; BIFOLD, Berlin, Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"320","published-online":{"date-parts":[[2026,4,23]]},"reference":[{"key":"e_1_2_1_1_1","first-page":"265","volume-title":"TensorFlow: A System for Large-Scale Machine Learning. In OSDI","author":"Mart\u00b4n","year":"2016","unstructured":"Mart\u00b4n Abadi et al. TensorFlow: A System for Large-Scale Machine Learning. In OSDI, pages 265-283, 2016."},{"issue":"12","key":"e_1_2_1_2_1","first-page":"1849","article-title":"Mind the Gap","volume":"10","author":"Aberger Christopher R.","year":"2017","unstructured":"Christopher R. Aberger, Andrew Lamb, Kunle Olukotun, and Christopher R\u00b4e. Mind the Gap: Bridging Multi-Domain Query Workloads with EmptyHeaded. PVLDB, 10(12):1849-1852, 2017.","journal-title":"Bridging Multi-Domain Query Workloads with EmptyHeaded. PVLDB"},{"issue":"11","key":"e_1_2_1_3_1","first-page":"1414","article-title":"RHEEM: enabling cross-platform data processing - may the big data be with you! -","volume":"11","author":"Divy Agrawal","year":"2018","unstructured":"Divy Agrawal et al. RHEEM: enabling cross-platform data processing - may the big data be with you! -. PVLDB, 11(11):1414-1427, 2018.","journal-title":"PVLDB"},{"key":"e_1_2_1_4_1","first-page":"1840","volume-title":"SIGMOD","author":"Al-Sayeh Hani","year":"2022","unstructured":"Hani Al-Sayeh, Bunjamin Memishi, Muhammad Attahir Jibril, Marcus Paradies, and Kai-Uwe Sattler. Juggler: Autonomous cost optimization and performance prediction of big data applications. In SIGMOD, pages 1840-1854, 2022."},{"key":"e_1_2_1_5_1","unstructured":"Alex Hall. Caching & reuse of subresults across queries 2024. URL: https:\/\/www.firebolt.io\/blog\/cach ing-reuse-of-subresults-across-queries."},{"key":"e_1_2_1_6_1","first-page":"47","volume-title":"SIGMOD","author":"Alexandrov Alexander","year":"2015","unstructured":"Alexander Alexandrov, Andreas Kunft, Asterios Katsifodimos, Felix Sch\u00a8uler, Lauritz Thamsen, Odej Kao, Tobias Herb, and Volker Markl. Implicit parallelism through deep language embedding. In SIGMOD, pages 47-61, 2015."},{"key":"e_1_2_1_7_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01424-7"},{"key":"e_1_2_1_8_1","volume-title":"Diana Marculescu, Yuejie Chi, and Carole-Jean Wu","author":"Paul Barham","year":"2022","unstructured":"Paul Barham et al. Pathways: Asynchronous distributed dataflow for ML. In Diana Marculescu, Yuejie Chi, and Carole-Jean Wu, editors, MLSys, 2022."},{"key":"e_1_2_1_9_1","doi-asserted-by":"publisher","DOI":"10.14778\/3574245.3574260"},{"key":"e_1_2_1_10_1","doi-asserted-by":"publisher","DOI":"10.1145\/3448016.3457549"},{"key":"e_1_2_1_11_1","first-page":"1387","volume-title":"KDD","author":"Denis","year":"2017","unstructured":"Denis Baylor et al. TFX: A tensorflow-based production-scale machine learning platform. In KDD, pages 1387-1395, 2017."},{"issue":"3","key":"e_1_2_1_12_1","doi-asserted-by":"crossref","first-page":"30","DOI":"10.1145\/3631504.3631510","article-title":"A unified data analytics framework","volume":"52","author":"Beedkar Kaustubh","year":"2023","unstructured":"Kaustubh Beedkar, Bertty Contreras-Rojas, Haralampos Gavriilidis, Zoi Kaoudi, Volker Markl, Rodrigo Pardo-Meza, and Jorge-Arnulfo Quian\u00b4e-Ruiz. Apache wayang: A unified data analytics framework. SIGMOD Rec., 52(3):30-35, 2023.","journal-title":"SIGMOD Rec."},{"key":"e_1_2_1_13_1","doi-asserted-by":"publisher","DOI":"10.1137\/141000671"},{"key":"e_1_2_1_14_1","doi-asserted-by":"publisher","DOI":"10.14778\/3007263.3007279"},{"key":"e_1_2_1_15_1","volume-title":"CIDR","author":"Matthias Boehm","year":"2020","unstructured":"Matthias Boehm et al. SystemDS: A Declarative Machine Learning System for the End-to-End Data Science Lifecycle. In CIDR, 2020. URL: http: \/\/cidrdb.org\/cidr2020\/papers\/p22-boehm-cidr20.pdf."},{"key":"e_1_2_1_16_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-01869-5"},{"key":"e_1_2_1_17_1","volume-title":"Speeding up neural network training with data echoing","author":"Choi Dami","year":"2020","unstructured":"Dami Choi and George Dahl. Speeding up neural network training with data echoing, 2020. URL: https:\/\/ai.googleblog.com\/2020\/05\/speeding-up-neura l-network-training.html."},{"key":"e_1_2_1_18_1","volume-title":"Memory management of pytorch","author":"Contributors PyTorch","year":"2023","unstructured":"PyTorch Contributors. Memory management of pytorch, 2023. URL: https:\/\/pytorch.org\/docs\/stable\/ notes\/cuda.html#memory-management."},{"key":"e_1_2_1_19_1","first-page":"613","volume-title":"NSDI","author":"Crankshaw Daniel","year":"2017","unstructured":"Daniel Crankshaw, Xin Wang, Giulio Zhou, Michael J. Franklin, Joseph E. Gonzalez, and Ion Stoica. Clipper: A low-latency online prediction serving system. In NSDI, pages 613-627, 2017."},{"key":"e_1_2_1_20_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00020"},{"key":"e_1_2_1_21_1","volume-title":"CIDR","author":"Patrick","year":"2022","unstructured":"Patrick Damme et al. DAPHNE: an open and extensible system infrastructure for integrated data analysis pipelines. In CIDR, 2022."},{"key":"e_1_2_1_22_1","first-page":"1","volume-title":"Amazon sagemaker autopilot: a white box automl solution at scale","author":"Piali Das","year":"2020","unstructured":"Piali Das et al. Amazon sagemaker autopilot: a white box automl solution at scale. In Sebastian Schelter, Steven Whang, and Julia Stoyanovich, editors, DEEM@SIGMOD, pages 2:1-2:7, 2020."},{"key":"e_1_2_1_23_1","doi-asserted-by":"publisher","DOI":"10.1145\/3318464.3389715"},{"key":"e_1_2_1_24_1","volume-title":"The Unified Apache Beam Model","author":"Developers Apache Beam","year":"2024","unstructured":"Apache Beam Developers. The Unified Apache Beam Model, 2024. URL: https:\/\/beam.apache.org\/."},{"key":"e_1_2_1_25_1","doi-asserted-by":"publisher","DOI":"10.1145\/3483840"},{"key":"e_1_2_1_26_1","doi-asserted-by":"publisher","DOI":"10.1145\/3437801.3441593"},{"key":"e_1_2_1_27_1","volume-title":"Advances in Neural Information Processing Systems","author":"Haoyang Fang","year":"2025","unstructured":"Haoyang Fang et al. Mlzero: A multi-agent system for end-to-end machine learning automation. In Advances in Neural Information Processing Systems, 2025. URL: https:\/\/neurips.cc\/virtual\/2025\/poster\/119504."},{"key":"e_1_2_1_28_1","doi-asserted-by":"crossref","first-page":"113","DOI":"10.1007\/978-3-030-05318-5_6","volume-title":"Automated Machine Learning - Methods","author":"Feurer Matthias","year":"2019","unstructured":"Matthias Feurer, Aaron Klein, Katharina Eggensperger, Jost Tobias Springenberg, Manuel Blum, and Frank Hutter. Auto-sklearn: Efficient and robust automated machine learning. In Automated Machine Learning - Methods, Systems, Challenges, pages 113-134. 2019."},{"key":"e_1_2_1_29_1","first-page":"2794","volume-title":"ICDE","author":"Gavriilidis Haralampos","year":"2023","unstructured":"Haralampos Gavriilidis, Kaustubh Beedkar, Jorge-Arnulfo Quian\u00b4e-Ruiz, and Volker Markl. In-situ cross-database query processing. In ICDE, pages 2794-2807, 2023."},{"key":"e_1_2_1_30_1","first-page":"452","volume-title":"IEEE BigData","author":"Giannakouris Victor","year":"2016","unstructured":"Victor Giannakouris, Nikolaos Papailiou, Dimitrios Tsoumakos, and Nectarios Koziris. Musqle: Distributed SQL query execution over multiple engine environments. In IEEE BigData, pages 452-461, 2016."},{"key":"e_1_2_1_31_1","first-page":"1","volume-title":"EuroSys","author":"Gog Ionel","year":"2015","unstructured":"Ionel Gog, Malte Schwarzkopf, Natacha Crooks, Matthew P. Grosvenor, Allen Clement, and Steven Hand. Musketeer: all for one, one for all in data processing systems. In EuroSys, pages 2:1-2:16, 2015."},{"key":"e_1_2_1_32_1","doi-asserted-by":"publisher","DOI":"10.1145\/3589273"},{"key":"e_1_2_1_33_1","first-page":"689","volume-title":"ATC","author":"Graur Dan","year":"2022","unstructured":"Dan Graur, Damien Aymon, Dan Kluser, Tanguy Albrici, Chandramohan A. Thekkath, and Ana Klimovic. Cachew: Machine learning input data processing as a service. In ATC, pages 689-706, 2022."},{"key":"e_1_2_1_34_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"e_1_2_1_35_1","doi-asserted-by":"publisher","DOI":"10.1145\/3373376.3378465"},{"key":"e_1_2_1_36_1","doi-asserted-by":"publisher","DOI":"10.1145\/1559845.1559879"},{"key":"e_1_2_1_37_1","volume-title":"MLSys","author":"Paras","year":"2020","unstructured":"Paras Jain et al. Checkmate: Breaking the memory wall with optimal tensor rematerialization. In MLSys, 2020."},{"key":"e_1_2_1_38_1","doi-asserted-by":"publisher","DOI":"10.1145\/3503222.3507778"},{"key":"e_1_2_1_39_1","first-page":"221","volume-title":"ICDE","author":"Kontaxakis Antonios","year":"2024","unstructured":"Antonios Kontaxakis, Dimitris Sacharidis, Alkis Simitsis, Alberto Abell\u00b4o, and Sergi Nadal. HYPPO: Using Equivalences to Optimize Pipelines in Exploratory Machine Learning. In ICDE, pages 221-234, 2024."},{"key":"e_1_2_1_40_1","first-page":"1106","volume-title":"NeurIPS","author":"Krizhevsky Alex","year":"2012","unstructured":"Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. Imagenet classification with deep convolutional neural networks. In NeurIPS, pages 1106-1114, 2012."},{"key":"e_1_2_1_41_1","first-page":"19","volume-title":"OSDI","author":"Lai Fan","year":"2021","unstructured":"Fan Lai, Xiangfeng Zhu, Harsha V. Madhyastha, and Mosharaf Chowdhury. Oort: Efficient federated learning via guided participant selection. In OSDI, pages 19-35, 2021."},{"key":"e_1_2_1_42_1","volume-title":"CIDR","author":"Landgraf Lukas","year":"2022","unstructured":"Lukas Landgraf, Wolfgang Lehner, Florian Wolf, and Alexander Boehm. Memory efficient scheduling of query pipeline execution. In CIDR, 2022."},{"key":"e_1_2_1_43_1","first-page":"611","volume-title":"OSDI","author":"Lee Yunseong","year":"2018","unstructured":"Yunseong Lee, Alberto Scolari, Byung-Gon Chun, Marco Domenico Santambrogio, Markus Weimer, and Matteo Interlandi. PRETZEL: opening the black box of machine learning prediction serving systems. In OSDI, pages 611-626, 2018."},{"key":"e_1_2_1_44_1","volume-title":"MLSys","author":"Li Liam","year":"2020","unstructured":"Liam Li, Kevin G. Jamieson, Afshin Rostamizadeh, Ekaterina Gonina, Jonathan Ben-tzur, Moritz Hardt, Benjamin Recht, and Ameet Talwalkar. A system for massively parallel hyperparameter tuning. In MLSys, 2020."},{"key":"e_1_2_1_45_1","first-page":"185","article-title":"A novel bandit-based approach to hyperparameter optimization","volume":"18","author":"Li Lisha","year":"2017","unstructured":"Lisha Li, Kevin G. Jamieson, Giulia DeSalvo, Afshin Rostamizadeh, and Ameet Talwalkar. Hyperband: A novel bandit-based approach to hyperparameter optimization. J. Mach. Learn. Res., 18:185:1-185:52, 2017. URL: http:\/\/jmlr.org\/papers\/v18\/16-558.html.","journal-title":"J. Mach. Learn. Res."},{"key":"e_1_2_1_46_1","doi-asserted-by":"publisher","DOI":"10.1145\/3318464.3386126"},{"key":"e_1_2_1_47_1","first-page":"681","volume-title":"WWW","author":"Liu Chao","year":"2010","unstructured":"Chao Liu, Hung-chih Yang, Jinliang Fan, Li-wei He, and Yi-Min Wang. Distributed nonnegative matrix factorization for web-scale dyadic data analysis on mapreduce. In WWW, pages 681-690, 2010."},{"key":"e_1_2_1_48_1","first-page":"4765","volume-title":"NIPS","author":"Scott","year":"2017","unstructured":"Scott M. Lundberg and Su-In Lee. A unified approach to interpreting model predictions. In NIPS, pages 4765-4774, 2017."},{"key":"e_1_2_1_49_1","first-page":"58","volume-title":"(ICML","volume":"64","author":"Mendoza Hector","year":"2016","unstructured":"Hector Mendoza, Aaron Klein, Matthias Feurer, Jost Tobias Springenberg, and Frank Hutter. Towards automatically-tuned neural networks. In (ICML, volume 64, pages 58-65, 2016."},{"key":"e_1_2_1_50_1","volume-title":"ML System@NeurIPS Workshop","author":"Meng Chen","year":"2017","unstructured":"Chen Meng, Minmin Sun, Jun Yang, Minghui Qiu, and Yang Gu. Training deeper models by gpu memory optimization on tensorflow. In ML System@NeurIPS Workshop, 2017."},{"key":"e_1_2_1_51_1","first-page":"34","volume":"17","author":"Xiangrui Meng","year":"2016","unstructured":"Xiangrui Meng et al. MLlib: Machine Learning in Apache Spark. JMLR, 17:34:1-34:7, 2016.","journal-title":"MLlib: Machine Learning in Apache Spark. JMLR"},{"key":"e_1_2_1_52_1","volume-title":"NeurIPS 2023","author":"Mince Fraser","year":"2023","unstructured":"Fraser Mince, Dzung Dinh, Jonas Kgomo, Neil Thompson, and Sara Hooker. The grand illusion: The myth of software portability and implications for ML progress. In NeurIPS 2023, 2023."},{"key":"e_1_2_1_53_1","first-page":"2430","volume-title":"ICML","author":"Azalia","year":"2017","unstructured":"Azalia Mirhoseini et al. Device placement optimization with reinforcement learning. In Doina Precup and Yee Whye Teh, editors, ICML, pages 2430-2439, 2017."},{"key":"e_1_2_1_54_1","first-page":"32","volume-title":"ProvenanceWeek@SIGMOD","author":"Mohammed Haneen","year":"2025","unstructured":"Haneen Mohammed and Eugene Wu. Lineage capture trade-offs: A case study in duckdb. In ProvenanceWeek@SIGMOD, pages 32-36, 2025."},{"key":"e_1_2_1_55_1","doi-asserted-by":"publisher","DOI":"10.14778\/3446095.3446100"},{"issue":"4","key":"e_1_2_1_56_1","first-page":"877","article-title":"an architecture for database kernel composition","volume":"17","author":"Mohr-Daurat Hubert","year":"2023","unstructured":"Hubert Mohr-Daurat, Xuan Sun, and Holger Pirk. BOSS - an architecture for database kernel composition. PVLDB, 17(4):877-890, 2023.","journal-title":"PVLDB"},{"key":"e_1_2_1_57_1","volume-title":"MLSys","author":"Moldovan Dan","year":"2019","unstructured":"Dan Moldovan, James M. Decker, Fei Wang, Andrew A. Johnson, Brian K. Lee, Zachary Nado, D. Sculley, Tiark Rompf, and Alexander B. Wiltschko. Autograph: Imperative-style coding with graph-based performance. In MLSys, 2019. URL: https:\/\/proceedings.mlsys.org\/book\/272.pdf."},{"key":"e_1_2_1_58_1","first-page":"561","volume-title":"OSDI","author":"Philipp","year":"2018","unstructured":"Philipp Moritz et al. Ray: A distributed framework for emerging AI applications. In OSDI, pages 561-577, 2018."},{"key":"e_1_2_1_59_1","doi-asserted-by":"publisher","DOI":"10.14778\/3476311.3476374"},{"key":"e_1_2_1_60_1","first-page":"1685","volume-title":"SIGMOD","author":"Nakandala Supun","year":"2020","unstructured":"Supun Nakandala and Arun Kumar. Vista: Optimized system for declarative feature transfer from deep cnns at scale. In SIGMOD, pages 1685-1700, 2020."},{"key":"e_1_2_1_61_1","volume-title":"NeurIPS","author":"Nam Jaehyun","year":"2025","unstructured":"Jaehyun Nam, Jinsung Yoon, Jiefeng Chen, Jinwoo Shin, Sercan \u00a8 O. Arik, and Tomas Pfister. MLE-STAR: machine learning engineering agent via search and targeted refinement. In NeurIPS, 2025."},{"key":"e_1_2_1_62_1","first-page":"1","volume-title":"SOSP","author":"Narayanan Deepak","year":"2019","unstructured":"Deepak Narayanan, Aaron Harlap, Amar Phanishayee, Vivek Seshadri, Nikhil R. Devanur, Gregory R. Ganger, Phillip B. Gibbons, and Matei Zaharia. Pipedream: generalized pipeline parallelism for DNN training. In SOSP, pages 1-15, 2019."},{"key":"e_1_2_1_63_1","first-page":"7294","volume-title":"ICML","author":"Nguyen Cuong V.","year":"2020","unstructured":"Cuong V. Nguyen, Tal Hassner, Matthias W. Seeger, and C\u00b4edric Archambeau. LEEP: A new measure to evaluate transferability of learned representations. In ICML, pages 7294-7305, 2020."},{"key":"e_1_2_1_64_1","first-page":"8024","volume-title":"NeurIPS","author":"Adam","year":"2019","unstructured":"Adam Paszke et al. Pytorch: An imperative style, high-performance deep learning library. In NeurIPS, pages 8024-8035, 2019."},{"key":"e_1_2_1_65_1","first-page":"2825","article-title":"Scikit-learn: Machine learning in python","volume":"12","author":"Fabian Pedregosa","year":"2011","unstructured":"Fabian Pedregosa et al. Scikit-learn: Machine learning in python. JMLR, 12:2825-2830, 2011.","journal-title":"JMLR"},{"key":"e_1_2_1_66_1","doi-asserted-by":"publisher","DOI":"10.1145\/3373376.3378505"},{"key":"e_1_2_1_67_1","first-page":"1","volume-title":"ICPP","author":"Perez Tiago B. G.","year":"2018","unstructured":"Tiago B. G. Perez, Xiaobo Zhou, and Dazhao Cheng. Reference-distance eviction and prefetching for cache management in spark. In ICPP, pages 88:1-88:10, 2018."},{"key":"e_1_2_1_68_1","unstructured":"Arnab Phani. Fine-grained reuse and feature transformations in machine learning systems. PhD thesis Technical University of Berlin Germany 2025. URL: https:\/\/nbn-resolving.org\/urn:nbn:de:101: 1-2504300159545.425811098571."},{"key":"e_1_2_1_69_1","doi-asserted-by":"publisher","DOI":"10.48786\/EDBT.2025.21"},{"issue":"11","key":"e_1_2_1_70_1","first-page":"2929","article-title":"parallelization strategies for feature transformations in machine learning workloads","volume":"15","author":"Phani Arnab","year":"2022","unstructured":"Arnab Phani, Lukas Erlbacher, and Matthias Boehm. UPLIFT: parallelization strategies for feature transformations in machine learning workloads. PVLDB, 15(11):2929-2938, 2022.","journal-title":"PVLDB"},{"key":"e_1_2_1_71_1","doi-asserted-by":"publisher","DOI":"10.1145\/3448016.3452788"},{"key":"e_1_2_1_72_1","doi-asserted-by":"publisher","DOI":"10.5441\/002\/EDBT.2021.07"},{"issue":"2","key":"e_1_2_1_73_1","first-page":"304","article-title":"An efficient, flexible search engine for transfer learning","volume":"16","author":"Renggli C\u00b4edric","year":"2022","unstructured":"C\u00b4edric Renggli, Xiaozhe Yao, Luka Kolar, Luka Rimanic, Ana Klimovic, and Ce Zhang. Shift: An efficient, flexible search engine for transfer learning. PVLDB, 16(2):304-316, 2022.","journal-title":"PVLDB"},{"key":"e_1_2_1_74_1","first-page":"130","volume-title":"SciPy","author":"Rocklin Matthew","year":"2015","unstructured":"Matthew Rocklin. Dask: Parallel computation with blocked algorithms and task scheduling. In SciPy, pages 130 - 136, 2015."},{"key":"e_1_2_1_75_1","doi-asserted-by":"publisher","DOI":"10.1145\/3448016.3457323"},{"issue":"12","key":"e_1_2_1_76_1","first-page":"1781","volume":"11","author":"Schelter Sebastian","year":"2018","unstructured":"Sebastian Schelter, Dustin Lange, Philipp Schmidt, Meltem Celikel, Felix Bie\u00dfmann, and Andreas Grafberger. Automating Large-Scale Data Quality Verification. PVLDB, 11(12):1781-1794, 2018.","journal-title":"Automating Large-Scale Data Quality Verification. PVLDB"},{"key":"e_1_2_1_77_1","first-page":"10435","volume-title":"NeurIPS","author":"Noam","year":"2018","unstructured":"Noam Shazeer et al. Mesh-tensorflow: Deep learning for supercomputers. In NeurIPS, pages 10435-10444, 2018."},{"key":"e_1_2_1_78_1","volume-title":"Proc. ACM Manag. Data","author":"Siddiqi Shafaq","year":"2024","unstructured":"Shafaq Siddiqi, Roman Kern, and Matthias Boehm. Saga: A scalable framework for optimizing data cleaning pipelines for machine learning applications. Proc. ACM Manag. Data, 2024."},{"key":"e_1_2_1_79_1","volume-title":"ACM TODS","author":"Siddiqi Shafaq","year":"2025","unstructured":"Shafaq Siddiqi, Arnab Phani, Roman Kern, and Matthias Boehm. Saga++: A scalable framework for optimizing data cleaning pipelines for machine learning applications. ACM TODS, 2025."},{"key":"e_1_2_1_80_1","volume-title":"ICLR","author":"Simonyan Karen","year":"2015","unstructured":"Karen Simonyan and Andrew Zisserman. Very deep convolutional networks for large-scale image recognition. In ICLR, 2015."},{"key":"e_1_2_1_81_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICDE.2017.109"},{"key":"e_1_2_1_82_1","first-page":"1395","volume-title":"ICCV","author":"Tran Anh Tuan","year":"2019","unstructured":"Anh Tuan Tran, Cuong V. Nguyen, and Tal Hassner. Transferability and hardness of supervised classification tasks. In ICCV, pages 1395-1405, 2019."},{"key":"e_1_2_1_83_1","doi-asserted-by":"publisher","DOI":"10.14778\/3579075.3579083"},{"key":"e_1_2_1_84_1","doi-asserted-by":"publisher","DOI":"10.1109\/MCSE.2011.37"},{"key":"e_1_2_1_85_1","first-page":"1285","volume-title":"SIGMOD","author":"Vartak Manasi","year":"2018","unstructured":"Manasi Vartak, Joana M. F. da Trindade, Samuel Madden, and Matei Zaharia. MISTIQUE: A system to store and query model intermediates for model diagnosis. In SIGMOD, pages 1285-1300, 2018."},{"key":"e_1_2_1_86_1","doi-asserted-by":"publisher","DOI":"10.14778\/3570690.3570692"},{"key":"e_1_2_1_87_1","doi-asserted-by":"publisher","DOI":"10.14778\/3297753.3297763"},{"key":"e_1_2_1_88_1","first-page":"383","volume-title":"IPDPS","author":"Xu Luna","year":"2016","unstructured":"Luna Xu, Min Li, Li Zhang, Ali Raza Butt, Yandong Wang, and Zane Zhenhua Hu. MEMTUNE: dynamic memory management for in-memory data analytic platforms. In IPDPS, pages 383-392, 2016."},{"key":"e_1_2_1_89_1","first-page":"15","volume-title":"Resilient Distributed Datasets: A Fault-Tolerant Abstraction for In-Memory Cluster Computing. In NSDI","author":"Matei","year":"2012","unstructured":"Matei Zaharia et al. Resilient Distributed Datasets: A Fault-Tolerant Abstraction for In-Memory Cluster Computing. In NSDI, pages 15-28, 2012."},{"key":"e_1_2_1_90_1","first-page":"265","volume-title":"SIGMOD","author":"Zhang Ce","year":"2014","unstructured":"Ce Zhang, Arun Kumar, and Christopher R\u00b4e. Materialization optimizations for feature selection workloads. In SIGMOD, pages 265-276, 2014."},{"issue":"12","key":"e_1_2_1_91_1","first-page":"3848","article-title":"Pytorch FSDP: experiences on scaling fully sharded data parallel","volume":"16","author":"Yanli Zhao","year":"2023","unstructured":"Yanli Zhao et al. Pytorch FSDP: experiences on scaling fully sharded data parallel. PVLDB, 16(12):3848-3860, 2023.","journal-title":"PVLDB"},{"key":"e_1_2_1_92_1","first-page":"559","volume-title":"Alpa: Automating Inter- and Intra-Operator Parallelism for Distributed Deep Learning. In OSDI","author":"Lianmin","year":"2022","unstructured":"Lianmin Zheng et al. Alpa: Automating Inter- and Intra-Operator Parallelism for Distributed Deep Learning. In OSDI, pages 559-578, 2022."},{"key":"e_1_2_1_93_1","doi-asserted-by":"publisher","DOI":"10.1145\/1247480.1247540"}],"container-title":["ACM SIGMOD Record"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3810900.3810916","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,23]],"date-time":"2026-04-23T18:16:55Z","timestamp":1776968215000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3810900.3810916"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,23]]},"references-count":93,"journal-issue":{"issue":"1","published-print":{"date-parts":[[2026,4,23]]}},"alternative-id":["10.1145\/3810900.3810916"],"URL":"https:\/\/doi.org\/10.1145\/3810900.3810916","relation":{},"ISSN":["0163-5808"],"issn-type":[{"value":"0163-5808","type":"print"}],"subject":[],"published":{"date-parts":[[2026,4,23]]},"assertion":[{"value":"2026-04-23","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}