{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,8]],"date-time":"2026-06-08T20:05:39Z","timestamp":1780949139988,"version":"3.54.1"},"reference-count":62,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"7","license":[{"start":{"date-parts":[[2026,7,1]],"date-time":"2026-07-01T00:00:00Z","timestamp":1782864000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2026,7,1]],"date-time":"2026-07-01T00:00:00Z","timestamp":1782864000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,7,1]],"date-time":"2026-07-01T00:00:00Z","timestamp":1782864000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62576364"],"award-info":[{"award-number":["62576364"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Shenzhen Basic Research","award":["JCYJ20241202124430041"],"award-info":[{"award-number":["JCYJ20241202124430041"]}]},{"name":"CCF-Tencent Rhino-Bird Open Research Fund","award":["CCF-Tencent RAGR20250114"],"award-info":[{"award-number":["CCF-Tencent RAGR20250114"]}]},{"name":"CCF-Tencent Rhino-Bird Open Research Fund","award":["JR2025TEG002"],"award-info":[{"award-number":["JR2025TEG002"]}]},{"name":"NTU RSR"},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62225113"],"award-info":[{"award-number":["62225113"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["U23A20318"],"award-info":[{"award-number":["U23A20318"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62276195"],"award-info":[{"award-number":["62276195"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Science and Technology Major Project of Hubei Province","award":["2025BCB026"],"award-info":[{"award-number":["2025BCB026"]}]},{"name":"Science and Technology Major Project of Hubei Province","award":["2024BAB046"],"award-info":[{"award-number":["2024BAB046"]}]},{"DOI":"10.13039\/501100012556","name":"Foundation for Innovative Research Groups of Hubei Province","doi-asserted-by":"publisher","award":["2024AFA017"],"award-info":[{"award-number":["2024AFA017"]}],"id":[{"id":"10.13039\/501100012556","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Pattern Anal. Mach. Intell."],"published-print":{"date-parts":[[2026,7]]},"DOI":"10.1109\/tpami.2026.3665753","type":"journal-article","created":{"date-parts":[[2026,2,23]],"date-time":"2026-02-23T20:47:39Z","timestamp":1771879659000},"page":"7375-7386","source":"Crossref","is-referenced-by-count":1,"title":["Aligning Few-Step Diffusion Models With Dense Reward Difference Learning"],"prefix":"10.1109","volume":"48","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-1728-2588","authenticated-orcid":false,"given":"Ziyi","family":"Zhang","sequence":"first","affiliation":[{"name":"School of Computer Science, National Engineering Research Center for Multimedia Software, Hubei Key Laboratory of Multimedia and Network Communication Engineering and Institute of Artificial Intelligence, Wuhan University, Wuhan, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5659-3464","authenticated-orcid":false,"given":"Li","family":"Shen","sequence":"additional","affiliation":[{"name":"School of Cyber Science and Technology, Shenzhen Campus of Sun Yat-sen University, Shenzhen, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sen","family":"Zhang","sequence":"additional","affiliation":[{"name":"TikTok, ByteDance, Sydney, NSW, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1754-1837","authenticated-orcid":false,"given":"Deheng","family":"Ye","sequence":"additional","affiliation":[{"name":"Tencent Inc., Shenzhen, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2296-6370","authenticated-orcid":false,"given":"Yong","family":"Luo","sequence":"additional","affiliation":[{"name":"School of Computer Science, National Engineering Research Center for Multimedia Software, Hubei Key Laboratory of Multimedia and Network Communication Engineering and Institute of Artificial Intelligence, Wuhan University, Wuhan, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4933-0073","authenticated-orcid":false,"given":"Miaojing","family":"Shi","sequence":"additional","affiliation":[{"name":"College of Electronic and Information Engineering, Tongji University, Shanghai, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6386-6096","authenticated-orcid":false,"given":"Dongjing","family":"Shan","sequence":"additional","affiliation":[{"name":"School of Medical Information and Engineering, Southwest Medical University, Luzhou, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0059-8458","authenticated-orcid":false,"given":"Bo","family":"Du","sequence":"additional","affiliation":[{"name":"School of Computer Science, National Engineering Research Center for Multimedia Software, Hubei Key Laboratory of Multimedia and Network Communication Engineering and Institute of Artificial Intelligence, Wuhan University, Wuhan, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7225-5449","authenticated-orcid":false,"given":"Dacheng","family":"Tao","sequence":"additional","affiliation":[{"name":"College of Computing and Data Science and the Generative AI Lab, Nanyang Technological University, Singapore"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-73016-0_6"},{"key":"ref2","article-title":"Training diffusion models with reinforcement learning","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Black"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.52202\/075280-1594"},{"key":"ref4","article-title":"Progressive distillation for fast sampling of diffusion models","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Salimans"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.01374"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.52202\/075280-3344"},{"key":"ref7","first-page":"32211","article-title":"Consistency models","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Song"},{"key":"ref8","article-title":"Latent consistency models: Synthesizing high-resolution images with few-step inference","author":"Luo","year":"2023"},{"key":"ref9","article-title":"Consistency trajectory models: Learning probability flow ODE trajectory of diffusion","volume-title":"Proc. 12th Int. Conf. Learn. Representations","author":"Kim"},{"key":"ref10","article-title":"InstaFlow: One step is enough for high-quality diffusion-based text-to-image generation","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Liu"},{"key":"ref11","article-title":"SDXL-Lightning: Progressive adversarial diffusion distillation","author":"Lin","year":"2024"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.00632"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.52202\/079017-1505"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.00746"},{"key":"ref15","first-page":"2256","article-title":"Deep unsupervised learning using nonequilibrium thermodynamics","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Sohl-Dickstein"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01042"},{"key":"ref17","article-title":"SDXL: Improving latent diffusion models for high-resolution image synthesis","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Podell"},{"key":"ref18","first-page":"12606","article-title":"Scaling rectified flow transformers for high-resolution image synthesis","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Esser"},{"key":"ref19","first-page":"25278","article-title":"LAION-5B: An open large-scale dataset for training next generation image-text models","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Schuhmann","year":"2022"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.52202\/075280-0700"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/iccv51070.2023.00200"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.01835"},{"key":"ref23","first-page":"9623","article-title":"Optimizing DDPM sampling with shortcut fine-tuning","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Fan"},{"key":"ref24","first-page":"79858","article-title":"DPOK: Reinforcement learning for fine-tuning text-to-image diffusion models","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Fan"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.00709"},{"key":"ref26","first-page":"52354","article-title":"Rebel: Reinforcement learning via regressing relative rewards","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Gao"},{"key":"ref27","first-page":"6840","article-title":"Denoising diffusion probabilistic models","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Ho"},{"key":"ref28","article-title":"Denoising diffusion implicit models","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Song"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1007\/BF00992696"},{"key":"ref30","article-title":"Proximal policy optimization algorithms","author":"Schulman","year":"2017"},{"key":"ref31","first-page":"53728","article-title":"Direct preference optimization: Your language model is secretly a reward model","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Rafailov"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1080\/02664768700000020"},{"key":"ref33","first-page":"725","article-title":"Kernel interpolation with continuous volume sampling","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Belhadji"},{"key":"ref34","article-title":"LoRA: Low-rank adaptation of large language models","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Hu"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.00854"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.00786"},{"key":"ref37","first-page":"24897","article-title":"Aligning diffusion models by optimizing human utility","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Li"},{"key":"ref38","first-page":"1656","article-title":"RL for consistency models: Reward guided text-to-image generation with fast inference","volume":"4","author":"Oertell","year":"2024","journal-title":"Reinforcement Learn. J."},{"key":"ref39","article-title":"Lykon\/dreamshaper-7","year":"2023","journal-title":"HuggingFace"},{"key":"ref40","article-title":"latent-consistency\/lcm-sdxl","year":"2023","journal-title":"HuggingFace"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51701.2025.01520"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51701.2025.01725"},{"key":"ref43","article-title":"Aligning text-to-image models using human feedback","author":"Lee","year":"2023"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.00200"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1145\/3796982"},{"key":"ref46","first-page":"60396","article-title":"Confronting reward overoptimization for diffusion models: A perspective of inductive and primacy biases","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Zhang"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-73036-8_1"},{"key":"ref48","first-page":"48892","article-title":"Feedback efficient online fine-tuning of diffusion models","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Uehara"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-72920-1_26"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2025.3590730"},{"key":"ref51","article-title":"Flow-GRPO: Training flow matching models via online RL","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Liu"},{"key":"ref52","article-title":"Directly fine-tuning diffusion models on differentiable rewards","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Clark"},{"key":"ref53","article-title":"Aligning text-to-image diffusion models with reward backpropagation","author":"Prabhudesai","year":"2023"},{"key":"ref54","article-title":"Video diffusion alignment via reward gradients","author":"Prabhudesai","year":"2024"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-73010-8_7"},{"key":"ref56","first-page":"55998","article-title":"A dense reward view on aligning text-to-image diffusion with preference","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Yang"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.52202\/079017-2334"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52734.2025.01232"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51701.2025.01710"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52734.2025.00269"},{"key":"ref61","article-title":"Reward guided latent consistency distillation","author":"Li","year":"2024","journal-title":"Trans. Mach. Learn. Res."},{"key":"ref62","first-page":"117340","article-title":"Hyper-SD: Trajectory segmented consistency model for efficient image synthesis","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Ren"}],"container-title":["IEEE Transactions on Pattern Analysis and Machine Intelligence"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/34\/11552636\/11408100.pdf?arnumber=11408100","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,8]],"date-time":"2026-06-08T19:52:21Z","timestamp":1780948341000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11408100\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7]]},"references-count":62,"journal-issue":{"issue":"7"},"URL":"https:\/\/doi.org\/10.1109\/tpami.2026.3665753","relation":{},"ISSN":["0162-8828","2160-9292","1939-3539"],"issn-type":[{"value":"0162-8828","type":"print"},{"value":"2160-9292","type":"electronic"},{"value":"1939-3539","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7]]}}}