{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,20]],"date-time":"2026-05-20T16:02:26Z","timestamp":1779292946302,"version":"3.51.4"},"reference-count":85,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,10,19]],"date-time":"2025-10-19T00:00:00Z","timestamp":1760832000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,10,19]],"date-time":"2025-10-19T00:00:00Z","timestamp":1760832000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/100000181","name":"Air Force Office of Scientific Research","doi-asserted-by":"publisher","award":["FA2386-23-1-4044"],"award-info":[{"award-number":["FA2386-23-1-4044"]}],"id":[{"id":"10.13039\/100000181","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,10,19]]},"DOI":"10.1109\/iccv51701.2025.00235","type":"proceedings-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T19:45:49Z","timestamp":1777491949000},"page":"2440-2450","source":"Crossref","is-referenced-by-count":1,"title":["Beyond Losses Reweighting: Empowering Multi-Task Learning via the Generalization Perspective"],"prefix":"10.1109","author":[{"given":"Hoang","family":"Phan","sequence":"first","affiliation":[{"name":"New York University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Lam","family":"Tran","sequence":"additional","affiliation":[{"name":"Qualcomm AI Research, Qualcomm AI Research is an initiative of Qualcomm Technologies, Inc."}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Quyen","family":"Tran","sequence":"additional","affiliation":[{"name":"Qualcomm AI Research, Qualcomm AI Research is an initiative of Qualcomm Technologies, Inc."}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Ngoc","family":"Tran","sequence":"additional","affiliation":[{"name":"Vanderbilt University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Tuan","family":"Truong","sequence":"additional","affiliation":[{"name":"Qualcomm AI Research, Qualcomm AI Research is an initiative of Qualcomm Technologies, Inc."}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Qi","family":"Lei","sequence":"additional","affiliation":[{"name":"New York University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Nhat","family":"Ho","sequence":"additional","affiliation":[{"name":"University of Texas at Austin"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Dinh","family":"Phung","sequence":"additional","affiliation":[{"name":"Qualcomm AI Research, Qualcomm AI Research is an initiative of Qualcomm Technologies, Inc."}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Trung","family":"Le","sequence":"additional","affiliation":[{"name":"Monash University"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Sharp-maml: Sharpness-aware model-agnostic meta learning","author":"Abbas","year":"2022","journal-title":"arXiv preprint arXiv"},{"issue":"236","key":"ref2","first-page":"1","article-title":"On the properties of variational approximations of gibbs posteriors","volume":"17","author":"Alquier","year":"2016","journal-title":"Journal of Machine Learning Research"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2016.2644615"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.508"},{"key":"ref5","article-title":"Fair resource allocation in multi-task learning","volume-title":"Forty-first International Conference on Machine Learning","author":"Ban"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1175\/1520-0493(1950)078<0001:VOFEIT>2.0.CO;2"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-20050-2_38"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1023\/A:1007379606734"},{"key":"ref9","first-page":"22405","article-title":"Swad: Domain generalization by seeking flat minima","volume":"34","author":"Cha","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1088\/1742-5468\/ab39d9"},{"key":"ref11","article-title":"When vision transformers outperform resnets without pre-training or strong data augmentations","author":"Chen","year":"2021","journal-title":"arXiv preprint arXiv"},{"key":"ref12","first-page":"794","article-title":"Gradnorm: Gradient normalization for adaptive loss balancing in deep multitask networks","volume-title":"International conference on machine learning","author":"Chen"},{"key":"ref13","first-page":"2039","article-title":"Just pick a sign: Optimizing deep multitask models with gradient sign dropout","volume":"33","author":"Chen","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.350"},{"key":"ref15","first-page":"18710","article-title":"Flattening sharpness for dynamic gradient projection memory benefits continual learning","volume":"34","author":"Deng","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1016\/j.crma.2012.03.014"},{"key":"ref17","first-page":"1019","article-title":"Sharp minima can generalize for deep nets","volume-title":"International Conference on Machine learning","author":"Dinh"},{"key":"ref18","article-title":"Efficient sharpness-aware minimization for improved training of neural networks","volume-title":"International Conference on Learning Representations","author":"Du"},{"key":"ref19","article-title":"Computing nonvacuous generalization bounds for deep (stochastic) neural networks with many more parameters than training data","volume-title":"UAI","author":"Dziugaite","year":"2017"},{"key":"ref20","article-title":"Sharpness-aware minimization for efficiently improving generalization","volume-title":"International Conference on Learning Representations","author":"Foret"},{"key":"ref21","article-title":"Emergent properties of the local geometry of neural loss landscapes","author":"Fort","year":"2019","journal-title":"arXiv preprint arXiv"},{"key":"ref22","first-page":"1321","article-title":"On calibration of modern neural networks","volume-title":"International conference on machine learning","author":"Guo"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01270-0_17"},{"key":"ref24","article-title":"Asymmetric valleys: Beyond sharp and flat local minima","author":"He","year":"2019","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref26","first-page":"529","article-title":"Simplifying neural nets by discovering flat minima","volume-title":"NIPS","author":"Hochreiter","year":"1994"},{"key":"ref27","first-page":"876","article-title":"Averaging weights leads to wider optima and better generalization","volume-title":"UAI","author":"Izmailov","year":"2018"},{"key":"ref28","article-title":"Rotograd: Gradient homogenization in multitask learning","volume-title":"International Conference on Learning Representations","author":"Javaloy"},{"key":"ref29","article-title":"Fantastic generalization measures and where to find them","author":"Jiang","year":"2020","journal-title":"ICLR. OpenReview.net"},{"key":"ref30","article-title":"A fair comparison of two popular flat minima optimizers: Stochastic weight averaging vs. sharpness-aware minimization","author":"Kaddour","year":"2022","journal-title":"arXiv preprint arXiv"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00781"},{"key":"ref32","article-title":"On largebatch training for deep learning: Generalization gap and sharp minima","author":"Keskar","year":"2017","journal-title":"ICLR. OpenReview.net"},{"key":"ref33","article-title":"On largebatch training for deep learning: Generalization gap and sharp minima","volume-title":"5th International Conference on Learning Representations, ICLR 2017","author":"Keskar"},{"key":"ref34","article-title":"Adam: A method for stochastic optimization","author":"Kingma","year":"2014","journal-title":"arXiv preprint arXiv"},{"key":"ref35","volume-title":"Asam: Adaptive sharpness-aware minimization for scale-invariant learning of deep neural networks","author":"Kwon","year":"2021"},{"key":"ref36","article-title":"Simple and scalable predictive uncertainty estimation using deep ensembles","author":"Lakshminarayanan","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1214\/aos\/1015957395"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/5.726791"},{"key":"ref39","article-title":"Visualizing the loss landscape of neural nets","author":"Li","year":"2018","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.52202\/068431-0725"},{"key":"ref41","article-title":"Reasonable effectiveness of random weighting: A litmus test for multi-task learning","author":"Lin","year":"2021","journal-title":"arXiv preprint arXiv"},{"key":"ref42","article-title":"Pareto multi-task learning","author":"Lin","year":"2019","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref43","article-title":"Controllable pareto multi-task learning","author":"Lin","year":"2020","journal-title":"arXiv preprint arXiv"},{"key":"ref44","first-page":"18878","article-title":"Conflict-averse gradient descent for multi-task learning","volume":"34","author":"Liu","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.52202\/075280-2500"},{"key":"ref46","article-title":"Towards impartial multi-task learning","volume-title":"International Conference on Learning Representations","author":"Liu"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00197"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P19-1441"},{"key":"ref49","first-page":"14721","article-title":"Profiling pareto front with multi-objective stein variational gradient descent","volume":"34","author":"Liu","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01204"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2015.425"},{"key":"ref52","first-page":"11","volume":"15","author":"Liu","year":"2018","journal-title":"Large-scale celebfaces attributes (celeba) dataset"},{"key":"ref53","article-title":"Understanding the generalization benefit of normalization layers: Sharpness reduction","author":"Lyu","year":"2022","journal-title":"arXiv preprint arXiv"},{"key":"ref54","first-page":"6597","article-title":"Multi-task learning with user preferences: Gradient descent with controlled ascent in pareto optimization","volume-title":"International Conference on Machine learning","author":"Mahapatra"},{"key":"ref55","article-title":"Exact pareto optimal search for multi-task learning: Touring the pareto front","author":"Mahapatra","year":"2021","journal-title":"arXiv preprint arXiv"},{"key":"ref56","article-title":"Predictive uncertainty estimation via prior networks","author":"Malinin","year":"2018","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00195"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1145\/307400.307435"},{"key":"ref59","article-title":"SAM as an optimal relaxation of bayes","volume-title":"The Eleventh International Conference on Learning Representations","author":"M\u00f6llenhoff"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v29i1.9602"},{"key":"ref61","article-title":"Learning the pareto front with hypernetworks","volume-title":"International Conference on Learning Representations","author":"Navon"},{"key":"ref62","article-title":"Multitask learning as a bargaining game","author":"Navon","year":"2022","journal-title":"arXiv preprint arXiv"},{"key":"ref63","article-title":"Exploring generalization in deep learning","author":"Neyshabur","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref64","doi-asserted-by":"publisher","DOI":"10.52202\/075280-1343"},{"key":"ref65","article-title":"Can you trust your model\u2019s uncertainty? evaluating predictive uncertainty under dataset shift","author":"Ovadia","year":"2019","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref66","article-title":"Regularizing neural networks by penalizing confident output distributions","volume-title":"ICLR (Workshop). OpenReview.net","author":"Pereyra"},{"key":"ref67","first-page":"18420","article-title":"Relative flatness and generalization","author":"Petzka","year":"2021","journal-title":"NeurIPS"},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1645"},{"key":"ref69","article-title":"Generalized federated learning via sharpness aware minimization","author":"Qu","year":"2022","journal-title":"arXiv preprint arXiv"},{"key":"ref70","article-title":"An overview of multi-task learning in deep neural networks","author":"Ruder","year":"2017","journal-title":"arXiv preprint arXiv"},{"key":"ref71","article-title":"Multi-task learning as multi-objective optimization","author":"Sener","year":"2018","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref72","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-642-33715-4_54"},{"key":"ref73","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.00367"},{"key":"ref74","article-title":"Gradient vaccine: Investigating and improving multitask optimization in massively multilingual models","volume-title":"International Conference on Learning Representations","author":"Wang"},{"key":"ref75","first-page":"2958","article-title":"Adversarial weight perturbation helps robust generalization","volume":"33","author":"Wu","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref76","article-title":"Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms","author":"Xiao","year":"2017","journal-title":"arXiv preprint arXiv"},{"key":"ref77","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.00518"},{"key":"ref78","first-page":"5824","article-title":"Gradient surgery for multitask learning","volume":"33","author":"Yu","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref79","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.00479"},{"key":"ref80","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-10599-4_7"},{"key":"ref81","article-title":"Penalizing gradient norm for efficiently improving generalization in deep learning","author":"Zhao","year":"2022","journal-title":"arXiv preprint arXiv"},{"key":"ref82","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR46437.2021.00806"},{"key":"ref83","article-title":"Sharpness-aware minimization efficiently selects flatter minima late in training","author":"Zhou","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref84","doi-asserted-by":"publisher","DOI":"10.1145\/3701551.3703503"},{"key":"ref85","volume-title":"Surrogate gap minimization improves","author":"Zhuang"}],"event":{"name":"2025 IEEE\/CVF International Conference on Computer Vision (ICCV)","location":"Honolulu, HI, USA","start":{"date-parts":[[2025,10,19]]},"end":{"date-parts":[[2025,10,25]]}},"container-title":["2025 IEEE\/CVF International Conference on Computer Vision (ICCV)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11443115\/11443287\/11443352.pdf?arnumber=11443352","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T04:43:21Z","timestamp":1777610601000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11443352\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,19]]},"references-count":85,"URL":"https:\/\/doi.org\/10.1109\/iccv51701.2025.00235","relation":{},"subject":[],"published":{"date-parts":[[2025,10,19]]}}}