{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,15]],"date-time":"2026-06-15T20:48:58Z","timestamp":1781556538754,"version":"3.54.5"},"reference-count":50,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"3","license":[{"start":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T00:00:00Z","timestamp":1777593600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T00:00:00Z","timestamp":1777593600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T00:00:00Z","timestamp":1777593600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001711","name":"Swiss National Science Foundation","doi-asserted-by":"publisher","award":["2000-1-240094"],"award-info":[{"award-number":["2000-1-240094"]}],"id":[{"id":"10.13039\/501100001711","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Signal Process. Mag."],"published-print":{"date-parts":[[2026,5]]},"DOI":"10.1109\/msp.2026.3682808","type":"journal-article","created":{"date-parts":[[2026,6,12]],"date-time":"2026-06-12T19:44:54Z","timestamp":1781293494000},"page":"115-129","source":"Crossref","is-referenced-by-count":0,"title":["Sequence Modeling Architectures: Foundations [Special Issue on the Mathematics of Deep Learning]"],"prefix":"10.1109","volume":"43","author":[{"ORCID":"https:\/\/orcid.org\/0009-0004-0010-4472","authenticated-orcid":false,"given":"Felix","family":"Sarnthein","sequence":"first","affiliation":[{"name":"Max Planck Institute for Intelligent Systems, ELLIS Institute Tübingen, Tübingen, Germany"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Arshia","family":"Afzal","sequence":"additional","affiliation":[{"name":"Ecole polytechnique fédérale de Lausanne (EPFL), Lausanne, Switzerland"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Razvan","family":"Pascanu","sequence":"additional","affiliation":[{"name":"Google DeepMind, London, U.K."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1914-0367","authenticated-orcid":false,"given":"Antonio","family":"Orvieto","sequence":"additional","affiliation":[{"name":"ELLIS Institute Tübingen, Tübingen, Germany"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Caglar","family":"Gulcehre","sequence":"additional","affiliation":[{"name":"Ecole polytechnique fédérale de Lausanne (EPFL), Lausanne, Switzerland"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2752-574X","authenticated-orcid":false,"given":"Jonas","family":"Geiping","sequence":"additional","affiliation":[{"name":"ELLIS Institute Tübingen, Tübingen, Germany"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5644-0086","authenticated-orcid":false,"given":"Francesco","family":"Tonin","sequence":"additional","affiliation":[{"name":"Ecole polytechnique fédérale de Lausanne (EPFL), Lausanne, Switzerland"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Albert","family":"Gu","sequence":"additional","affiliation":[{"name":"Machine Learning Department, Carnegie Mellon University, Pittsburgh, PA, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5004-201X","authenticated-orcid":false,"given":"Volkan","family":"Cevher","sequence":"additional","affiliation":[{"name":"Ecole polytechnique fédérale de Lausanne (EPFL), Lausanne, Switzerland"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1207\/s15516709cog1402_1"},{"key":"ref2","article-title":"How to construct deep recurrent neural networks","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Pascanu","year":"2014"},{"key":"ref3","author":"Ba","year":"2016","journal-title":"Layer normalization"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr.2016.90"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.3115\/v1\/D14-1181"},{"key":"ref6","first-page":"28,043","article-title":"Hyena hierarchy: Towards larger convolutional language models","volume-title":"Proc. 40th Int. Conf. Mach. Learn.","author":"Poli","year":"2023"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1016\/S0166-4115(97)80111-2"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1162\/neco.1997.9.8.1735"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.3115\/v1\/D14-1179"},{"key":"ref10","first-page":"3800","article-title":"Improving the gating mechanism of recurrent neural networks","volume-title":"Proc. 37th Int. Conf. Mach. Learn. (ICML)","author":"Gu","year":"2020"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.48550\/ARXIV.1706.03762"},{"key":"ref12","article-title":"Neural machine translation by jointly learning to align and translate","author":"Bahdanau","year":"2014"},{"key":"ref13","article-title":"Neural turing machines","author":"Graves","year":"2014"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1016\/j.neucom.2023.127063"},{"key":"ref15","first-page":"606","article-title":"Efficiently scaling transformer inference","volume-title":"Proc. Mach. Learn. Syst.","author":"Pope","year":"2023"},{"key":"ref16","first-page":"5156","article-title":"Transformers are RNNs: Fast autoregressive transformers with linear attention","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Katharopoulos","year":"2020"},{"key":"ref17","article-title":"Hopfield networks is all you need","author":"Ramsauer","year":"2020"},{"key":"ref18","article-title":"Random feature attention","author":"Peng","year":"2021"},{"key":"ref19","article-title":"Retentive network: A successor to transformer for large language models","author":"Sun","year":"2023"},{"key":"ref20","first-page":"56,501","article-title":"Gated linear attention transformers with hardware-efficient training","volume-title":"Proc. 41st Int. Conf. Mach. Learn. (ICML)","author":"Yang","year":"2023"},{"key":"ref21","first-page":"9355","article-title":"Linear transformers are secretly fast weight programmers","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Schlag","year":"2021"},{"key":"ref22","first-page":"115,491","article-title":"Parallelizing linear transformers with the delta rule over sequence length","volume-title":"Proc. 38th Int. Conf. Neural Inf. Process. Syst. (NeurIPS)","author":"Yang","year":"2024"},{"key":"ref23","article-title":"Efficiently modeling long sequences with structured state spaces","volume-title":"Proc. Int. Conf. Learn. Representations (ICLR)","author":"Gu","year":"2022"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1115\/1.3662552"},{"key":"ref25","first-page":"22,982","article-title":"Diagonal state spaces are as effective as structured state spaces","volume-title":"Proc. 36th Int. Conf. Neural Inf. Process. Syst. (NeurIPS)","author":"Gupta","year":"2022"},{"key":"ref26","first-page":"26,670","article-title":"Resurrecting recurrent neural networks for long sequences","volume-title":"Proc. 40th Int. Conf. Mach. Learn. (ICML)","author":"Orvieto","year":"2023"},{"key":"ref27","volume-title":"Prefix sums and their applications","author":"Blelloch","year":"1990"},{"key":"ref28","article-title":"Simplified state space layers for sequence modeling","author":"Smith","year":"2022"},{"key":"ref29","volume-title":"Linear recurrences accessible to everyone","author":"Sarnthein","year":"2025"},{"key":"ref30","article-title":"HGRN2: Gated linear RNNs with state expansion","volume-title":"Proc. 1st Conf. Lang. Model. (COLM)","author":"Qin","year":"2024"},{"key":"ref31","article-title":"Mamba: Linear-time sequence modeling with selective state spaces","volume-title":"Proc. 1st Conf. Lang. Model. (COLM)","author":"Gu","year":"2024"},{"key":"ref32","first-page":"10,041","article-title":"Transformers are SSMs: Generalized models and efficient algorithms through structured state space duality","volume-title":"Proc. 41st Int. Conf. Mach. Learn.","author":"Dao","year":"2024"},{"key":"ref33","article-title":"Griffin: Mixing gated linear recurrences with local attention for efficient language models","author":"De","year":"2024"},{"key":"ref34","article-title":"Selective rotary position embedding","volume-title":"Proc. 14th Int. Conf. Learn. Representations","author":"Movahedi","year":"2026"},{"key":"ref35","article-title":"Hydra: Bidirectional state space models through generalized matrix mixers","author":"Hwang","year":"2024"},{"key":"ref36","article-title":"Linear attention for efficient bidirectional sequence modeling","author":"Afzal","year":"2025"},{"key":"ref37","article-title":"MiniMax-01: Scaling foundation models with lightning attention","author":"Li","year":"2025"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.02352"},{"key":"ref39","article-title":"The bitter lesson","author":"Sutton","year":"2019","journal-title":"Incomplete Ideas"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1145\/3467017"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1189"},{"key":"ref42","first-page":"7616","article-title":"Its raw audio generation with state-space models","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Goel","year":"2022"},{"key":"ref43","first-page":"2846","article-title":"S4ND: Modeling images and videos as multidimensional signals using state spaces","volume-title":"Proc. 36th Int. Conf. Neural Inf. Process. Syst. (NeurIPS)","author":"Nguyen","year":"2022"},{"key":"ref44","article-title":"A time series is worth 64 words: Long-term forecasting with transformers","author":"Nie","year":"2022"},{"key":"ref45","article-title":"An image is worth 16x16 words: Transformers for image recognition at scale","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Dosovitskiy","year":"2021"},{"key":"ref46","first-page":"62,429","article-title":"Vision mamba: Efficient visual representation learning with bidirectional state space model","volume-title":"Proc. 41st Int. Conf. Mach. Learn. (ICML)","author":"Zhu","year":"2024"},{"key":"ref47","article-title":"Vision-LSTM: xLSTM as generic vision backbone","author":"Alkin","year":"2024"},{"key":"ref48","article-title":"GPT-4 technical report","author":"Achiam","year":"2023"},{"key":"ref49","article-title":"LLaMA: Open and efficient foundation language models","author":"Touvron","year":"2023"},{"key":"ref50","first-page":"8748","article-title":"Learning transferable visual models from natural language supervision","volume-title":"Proc. Int. Conf. Mach. Learn. (ICML)","author":"Radford","year":"2021"}],"container-title":["IEEE Signal Processing Magazine"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/79\/11561017\/11561033.pdf?arnumber=11561033","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,15]],"date-time":"2026-06-15T19:50:07Z","timestamp":1781553007000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11561033\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5]]},"references-count":50,"journal-issue":{"issue":"3"},"URL":"https:\/\/doi.org\/10.1109\/msp.2026.3682808","relation":{},"ISSN":["1053-5888","1558-0792"],"issn-type":[{"value":"1053-5888","type":"print"},{"value":"1558-0792","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5]]}}}