Lune

ICLR2020Top-tier venue

Bayesian Meta Sampling for Fast Uncertainty Adaptation

Zhenyi Wang, Yang Zhao, Ping Yu, Ruiyi Zhang, Changyou Chen

Abstract

Meta learning has been making impressive progress for fast model adaptation. However, limited work has been done on learning fast uncertainty adaption for Bayesian modeling. In this paper, we propose to achieve the goal by placing meta learning on the space of probability measures, inducing the concept of meta sampling for fast uncertainty adaption. Specifically, we propose a Bayesian meta sampling framework consisting of two main components: a meta sampler and a sample adapter. The meta sampler is constructed by adopting a neural-inverse-autoregressive-flow (NIAF) structure, a variant of the recently proposed neural autoregressive flows, to efficiently generate meta samples to be adapted. The sample adapter then pushes meta samples to task-specific samples, based on a newly proposed and general Bayesian sampling technique, called optimal-transport Bayesian sampling. The combination of the two components allows a simple learning procedure for the meta sampler to be developed, which can be efficiently optimized via standard back-propagation. Extensive experimental results demonstrate the efficiency and effectiveness of the proposed framework, obtaining better sample quality and faster uncertainty adaption compared to related methods.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

5 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Task Groupings Regularization: Data-Free Meta-Learning with Heterogeneous Pre-trained Models cites Learning to Learn and Remember Super Long Multi-Domain Task Sequence
  • Task Groupings Regularization: Data-Free Meta-Learning with Heterogeneous Pre-trained Models cites Learning to Learn from APIs: Black-Box Data-Free Meta-Learning
  • Task Groupings Regularization: Data-Free Meta-Learning with Heterogeneous Pre-trained Models cites Architecture, Dataset and Model-Scale Agnostic Data-free Meta-Learning
  • Learning to Learn and Remember Super Long Multi-Domain Task Sequence cites Meta Learning on a Sequence of Imbalanced Domains with Difficulty Awareness
  • Learning to Learn from APIs: Black-Box Data-Free Meta-Learning cites Learning to Learn and Remember Super Long Multi-Domain Task Sequence
  • Learning to Learn from APIs: Black-Box Data-Free Meta-Learning cites Architecture, Dataset and Model-Scale Agnostic Data-free Meta-Learning
  • Learning to Learn from APIs: Black-Box Data-Free Meta-Learning cites Meta Learning on a Sequence of Imbalanced Domains with Difficulty Awareness
  • Architecture, Dataset and Model-Scale Agnostic Data-free Meta-Learning cites Learning to Learn and Remember Super Long Multi-Domain Task Sequence
  • Architecture, Dataset and Model-Scale Agnostic Data-free Meta-Learning cites Meta Learning on a Sequence of Imbalanced Domains with Difficulty Awareness
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers5

  1. Meta Learning on a Sequence of Imbalanced Domains with Difficulty AwarenessZhenyi Wang, Tiehang Duan, Le Fang, Qiuling Suo et al.ICCV 2021 · 21 citations
  2. Learning to Learn and Remember Super Long Multi-Domain Task SequenceZhenyi Wang, Li Shen, Tiehang Duan, Donglin Zhan et al.CVPR 2022 · 19 citations
  3. Learning to Learn from APIs: Black-Box Data-Free Meta-LearningZixuan Hu, Li Shen, Zhenyi Wang, Baoyuan Wu et al.ICML 2023 · 18 citations
  4. Task Groupings Regularization: Data-Free Meta-Learning with Heterogeneous Pre-trained ModelsYongxian Wei, Zixuan Hu, Li Shen, Zhenyi Wang et al.ICML 2024 · 11 citations
  5. Architecture, Dataset and Model-Scale Agnostic Data-free Meta-LearningZixuan Hu, Li Shen, Zhenyi Wang, Tongliang Liu et al.CVPR 2023

Builds on1

  1. Meta-Amortized Variational Inference and LearningMike Wu, Kristy Choi, Noah D. Goodman, Stefano ErmonAAAI 2020 · 39 citations

Related papers

  1. Annealing Flow Generative Models Towards Sampling High-Dimensional and Multi-Modal DistributionsDongze Wu, Yao XieICML 2025
  2. Gradient-EM Bayesian Meta-LearningYayi Zou, Xiaoqi LuNeurIPS 2020 · 16 citations
  3. A Meta-Learning Approach to Bayesian Causal DiscoveryAnish Dhir, Matthew Ashman, James Requeima, Mark van der WilkICLR 2025
  4. MALIBO: Meta-learning for Likelihood-free Bayesian OptimizationJiarong Pan, Stefan Falkner, Felix Berkenkamp, Joaquin VanschorenICML 2024 · 2 citations
  5. Bayesian Meta-Learning for the Few-Shot Setting via Deep KernelsMassimiliano Patacchiola, Jack Turner, Elliot J. Crowley, Michael F. P. O'Boyle et al.NeurIPS 2020 · 167 citations
Dusk over the sea between two cliffs drawn in fine vertical lines