Filters








12,816 Hits in 10.6 sec

Adaptive Risk Minimization: Learning to Adapt to Domain Shift [article]

Marvin Zhang, Henrik Marklund, Nikita Dhawan, Abhishek Gupta, Sergey Levine, Chelsea Finn
2021 arXiv   pre-print
Our primary contribution is to introduce the framework of adaptive risk minimization (ARM), in which models are directly optimized for effective adaptation to shift by learning to adapt on the training  ...  Compared to prior methods for robustness, invariance, and adaptation, ARM methods provide performance gains of 1-4% test accuracy on a number of image classification problems exhibiting domain shift.  ...  Adaptive Risk Minimization In this section, we formally describe the ARM framework, which defines an objective for training adaptive models to tackle domain shift.  ... 
arXiv:2007.02931v4 fatcat:2gjsygbev5cmhdryhrwijvfeju

Adaptation policy and community discourse: risk, vulnerability, and just transformation

David Schlosberg, Lisette B. Collins, Simon Niemeyer
2017 Environmental Politics  
Despite a discursive disconnect between governmental focus on a risk or resilience-based approach and a community concern with the vulnerability of basic needs and capabilities of everyday life, deliberative  ...  articulated by local governments, environmental groups, and local residents engaged in adaptation planning in Australia.  ...  This paper, along with the others in this symposium, was originally presented at a workshop on Just Adaptation hosted and funded by the Sydney Environment Institute.  ... 
doi:10.1080/09644016.2017.1287628 fatcat:rlr5d3qcurezngsczcll44ncpa

Discrepancy Minimization in Domain Generalization with Generative Nearest Neighbors [article]

Prashant Pandey, Mrigank Raman, Sumanth Varambally, Prathosh AP
2020 arXiv   pre-print
Features extracted from this source domain are learned using a generative model whose latent space is used as a sampler to retrieve the nearest neighbors for the target data points.  ...  Domain generalization (DG) deals with the problem of domain shift where a machine learning model trained on multiple-source domains fail to generalize well on a target domain with different statistics.  ...  Existing methods tackle the DG problem by learning domain invariant feature representations using adversarial methods (Li et al., 2018b,c) or meta-learning approaches where held-out source domains are  ... 
arXiv:2007.14284v1 fatcat:y5twxlfenvc35ikeu5eexrtbzi

Comparative Risk Assessment to Inform Adaptation Priorities for the Natural Environment: Observations from the First UK Climate Change Risk Assessment

Iain Brown
2015 Climate  
This suggests that goals for risk assessment need to be more clearly explicated and assumptions on tolerable risk declared as a primer for further dialogue on expectations for managed outcomes.  ...  Risk assessment can potentially provide an objective framework to synthesise and prioritise climate change risks to inform adaptation policy.  ...  Acknowledgments The CCRA Advisory Group are thanked for providing very useful and informative comments on the use of evidence to inform policy requirements.  ... 
doi:10.3390/cli3040937 fatcat:zfa3jmggu5f7bmm5la36man374

Understanding and unlocking transformative learning as a method for enabling behaviour change for adaptation and resilience to disaster threats

Justin Sharpe
2016 International Journal of Disaster Risk Reduction  
Cite this article as: Justin Sharpe, Understanding and unlocking transformative learning as a method for enabling behaviour change for adaptation and resilience to disaster threats, International Journal  ...  of Disaster Risk Reduction, http://dx.  ...  This includes social contexts and pathways for learning. Therefore new ways of approaching learning are required to help break-out of established ways of thinking and tackling problems.  ... 
doi:10.1016/j.ijdrr.2016.04.014 fatcat:znnf3idgqneq5ob5yr74j4r3qq

Risk, responsibility and reconfiguration: Penal adaptation and misadaptation

F. McNeill, N. Burns, S. Halliday, N. Hutton, C. Tata
2009 Punishment & Society  
Drawing on Bourdieu, we suggest that this may be best understood not as a counter-example to accounts of penal transformation but as evidence of an incompleteness in their analyses which reflects the '  ...  In so doing, we problematize and explore what we term the 'governmentality gap'; meaning, a lacuna in the existing penological scholarship which concerns the contingent relationships between changing governmental  ...  to support a variety of penal strategies -including 'correctional treatment' as a form of risk minimization.  ... 
doi:10.1177/1462474509341153 fatcat:74bs773xkreajjynfqcutoufaq

Uncertainty Minimization for Personalized Federated Semi-Supervised Learning [article]

Yanhang Shi, Siguang Chen, Haijun Zhang
2022 arXiv   pre-print
To tackle this problem, 1) we propose a novel personalized semi-supervised learning paradigm which allows partial-labeled or unlabeled clients to seek labeling assistance from data-related clients (helper  ...  Since federated learning (FL) has been introduced as a decentralized learning technique with privacy preservation, statistical heterogeneity of distributed data stays the main obstacle to achieve robust  ...  In [25] , Fallah et al. utilized the model-agnostic meta-learning (MAML) framework to find an initial global model, which can be easily adapt to clients' personal datasets by performing a few steps of  ... 
arXiv:2205.02438v1 fatcat:lsx2xo6tbbenjdn4prq7cs4cpy

What Is Minimally Cooperative Behavior? [chapter]

Kirk Ludwig
2020 From Conventionalism to Social Authenticity  
In principle, all phenomena dealing with sociality are covered as long as they are approached from a philosophical point of view, broadly understood.  ...  Covers a new and rapidly developing field that has become one of the key topics of the international philosophical world.  ...  Acknowledgments This research has been supported by a grant on Minimal Cooperation, led by Acknowledgements Thanks to Lambros Malafouris and Deb Tollefsen for helpful discussions on these topics.  ... 
doi:10.1007/978-3-030-29783-1_2 fatcat:kilqd6dgpvepvmr6lxmhlyaatq

Giant Proliferating Pilar Tumor of the Scalp: A Minimal Risk Approach

Christopher Rassekh, Kelly Malloy, Tom Thomas, Jason Brandt, Xiaowei Xu, Ara Chalian
2012 Journal of Neurological Surgery. Part B: Skull Base: an interdisciplinary approach  
The Occipital Transtentorial Approach for Superior Vermian and Superomedial Cerebellar Arteriovenous Malformations: Advantages, Limitations, and Options Nancy McLaughlin (presenter), Neil A.  ...  Giant Proliferating Pilar Tumor of the Scalp: A Minimal Risk Approach Christopher H. Rassekh (presenter), Kelly M. Malloy, Tom Thomas, Jason Brandt, Xiaowei Xu, Ara A.  ...  Finally, the reader will learn a systematic approach to characterizing complex lesions.  ... 
doi:10.1055/s-0032-1312280 fatcat:kksrjga6drfb7ojybu3lcsm55e

Software Replica of Minimal Living Processes

Hugues Bersini
2010 Origins of life and evolution of the biosphere  
Minimal life begins at the intersection of a series of processes which need to be isolated, differentiated and duplicated as such in computers.  ...  Only software developments and running make possible to understand the way these processes are intimately interconnected in order for life to appear at the crossroad.  ...  Acknowledgements G.B. thanks the Belgian Fonds de la Recherche Scientifique-FNRS for a post-doctoral fellowship.  ... 
doi:10.1007/s11084-010-9190-5 pmid:20204519 fatcat:m46uvtkpnbh7bhjzmwctk4q43e

Class-Incremental Domain Adaptation [article]

Jogendra Nath Kundu, Rahul Mysore Venkatesh, Naveen Venkat, Ambareesh Revanur, R. Venkatesh Babu
2020 arXiv   pre-print
Existing DA methods tackle domain-shift but are unsuitable for learning novel target-domain classes.  ...  We introduce a practical Domain Adaptation (DA) paradigm called Class-Incremental Domain Adaptation (CIDA).  ...  B) During CIDA, we learn a targetspecific feature extractor ft (to minimize domain-shift) and classifier gt (to learn C t ).  ... 
arXiv:2008.01389v1 fatcat:ukx4f6ohbzgyvo3vo2rvplxjje

Emerging Trends in Federated Learning: From Model Fusion to Federated X Learning [article]

Shaoxiong Ji and Teemu Saravirta and Shirui Pan and Guodong Long and Anwar Walid
2021 arXiv   pre-print
Following the emerging trends, we also discuss federated learning in the intersection with other learning paradigms, termed as federated x learning, where x includes multitask learning, meta-learning,  ...  We conduct a focused survey of federated learning in conjunction with other learning algorithms.  ...  [25] studied the federated unsupervised domain adaptation that aligns the shifted domains under federated setting with a couple of learning paradigms.  ... 
arXiv:2102.12920v2 fatcat:5fcwfhxibbedbcbuzrfyqdedky

Few-shot Adaptive Faster R-CNN [article]

Tao Wang, Xiaopeng Zhang, Li Yuan, Jiashi Feng
2019 arXiv   pre-print
To mitigate the detection performance drop caused by domain shift, we aim to develop a novel few-shot adaptation approach that requires only a few target domain images with limited bounding box annotations  ...  Third, the model suffers from over-adaptation (similar to overfitting when training with a few data example) and instability risk that may lead to degraded detection performance in the target domain.  ...  Later, [19] proposed to transfer the base class feature to a new class; a recent work [10] proposed a meta learning based approach which achieves state-of-the-art.  ... 
arXiv:1903.09372v1 fatcat:3mxc75hstbextk2fesqx7hib7q

MetaPred: Meta-Learning for Clinical Risk Prediction with Limited Patient Electronic Health Records [article]

Xi Sheryl Zhang, Fengyi Tang, Hiroko Dodge, Jiayu Zhou, Fei Wang
2019 arXiv   pre-print
In this paper, we propose MetaPred, a meta-learning for clinical risk prediction from longitudinal patient EHRs.  ...  In particular, in order to predict the target risk where there are limited data samples, we train a meta-learner from a set of related risk prediction tasks which learns how a good predictor is learned  ...  Specifically, model-agnostic meta-learning [15] aims to learn a good parameter initialization for the fast adaptation of testing tasks.  ... 
arXiv:1905.03218v1 fatcat:nkxrocx56vfrdj3hxos7qay2xq

Generalizing to Unseen Domains: A Survey on Domain Generalization [article]

Jindong Wang, Cuiling Lan, Chang Liu, Yidong Ouyang, Tao Qin, Wang Lu, Yiqiang Chen, Wenjun Zeng, Philip S. Yu
2022 arXiv   pre-print
We categorize recent algorithms into three classes: data manipulation, representation learning, and learning strategy, and present several popular algorithms in detail for each category.  ...  Domain generalization deals with a challenging setting where one or several different but related domain(s) are given, and the goal is to learn a model that can generalize to an unseen test domain.  ...  There are many generalization-related research topics such as domain adaptation, meta-learning, transfer learning, covariate shift, and so on.  ... 
arXiv:2103.03097v6 fatcat:gztglwp4nrdwbnddjr7t3j3ize
« Previous Showing results 1 — 15 out of 12,816 results