Research map / Label-efficient and robust learning / Domain adaptation and generalization
Reducing Distributional Uncertainty by Mutual Information Maximisation and Transferable Feature Learning
Jian Gao, Yang Hua, Guosheng Hu, Chi Wang, Neil M. Robertson
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add MI maximisation (MIM) to transferable feature learning (ablation) | TFL · Office-Home single-to-single UDA (ResNet-50) | mean accuracy (%): 63.4 -> 64.7 | Improved |
| add transferable shape feature learning (TFL) to MI maximisation (ablation) | MIM · Office-Home single-to-single UDA (ResNet-50) | mean accuracy (%): 63.1 -> 64.7 | Improved |
| add MI maximisation (MIM) to transferable feature learning (ablation) | TFL · DomainNet single-to-single (15 difficult tasks) | mean accuracy (%): 10.8 -> 12.6 | Improved |
| add MI maximisation (MIM) to transferable feature learning (ablation) | TFL · DomainNet single-to-single (all 30 tasks) | all mean accuracy (%): 27.3 -> 28.1 | Improved |
| add MEC loss | MIMTFL · Office-Home single-to-single UDA (ResNet-50) | mean accuracy (%): 64.7 -> 67.2 | Improved |
| train with target data (UDA) instead of source only | MIMTFL src · Office-Home single-to-single UDA (ResNet-50) | mean accuracy (%): 63.1 -> 64.7 | Improved |
| use MIMTFL instead of source-only training | Source only · DomainNet single-to-single (15 difficult tasks) | mean accuracy (%): 10.8 -> 12.6 | Improved |
| use ADDA instead of source-only training | Source only · DomainNet single-to-single (15 difficult tasks) | mean accuracy (%): 10.8 -> 9.7 | Hurt |
| use MCD instead of source-only training | Source only · DomainNet single-to-single (15 difficult tasks) | mean accuracy (%): 10.8 -> 9.2 | Hurt |
| use SE instead of source-only training | Source only · Office-Home multi-to-multi UDA | mean accuracy (%): 70.2 -> 65.6 | Hurt |
| use MIMTFL+MEC instead of source-only training | Source only · Office-Home APR->C multi-to-single | accuracy (%): 58.5 -> 64.3 | Improved |
| use shape-learning MIMTFL instead of jigsaw-puzzle JiGen | JiGen · Office-Home multi-to-single DG (ResNet-18) | mean accuracy (%): 61.2 -> 63.0 | Improved |
Closest papers by meaning
- Uncertainty Modeling for Out-of-Distribution Generalization (ICLR 2022 · similarity 0.85)
- A Bit More Bayesian: Domain-Invariant Learning with Uncertainty (ICML 2021 · similarity 0.85)
- Structure-Aware Feature Fusion for Unsupervised Domain Adaptation (AAAI 2020 · similarity 0.85)
- Domain Adaptation with Conditional Transferable Components (ICML 2016 · similarity 0.84)
- Robust Domain Generalization under Divergent Marginal and Conditional Distributions (NeurIPS 2026 · similarity 0.83)
- Probability-Polarized Optimal Transport for Unsupervised Domain Adaptation (AAAI 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
