Historical inference based on semi-supervised learning

Lee, Dong-Gi; Lee, Sangkuk; Kim, Myung-Jun; Shin, Hyunjung

doi:10.1016/j.eswa.2018.03.059

Cited by 12 publications

(9 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…generate unbiased and accurate pseudo-labels, such as resampling (Lee, Shin, and Kim 2021;Guo and Li 2022), re-weighting (Lai et al 2022), transfer learning (Fan et al 2022), and logit adjustment (Wang et al 2022a). However, these works usually assume a similar class distribution between labeled and unlabeled set and show inferior performances when such assumption is violated.…”

Section: Long-tailed Learningmentioning

confidence: 99%

Three Heads Are Better than One: Complementary Experts for Long-Tailed Semi-supervised Learning

Ma,

Elezi,

Deng

et al. 2024

AAAI

View full text Add to dashboard Cite

We address the challenging problem of Long-Tailed Semi-Supervised Learning (LTSSL) where labeled data exhibit imbalanced class distribution and unlabeled data follow an unknown distribution. Unlike in balanced SSL, the generated pseudo-labels are skewed towards head classes, intensifying the training bias. Such a phenomenon is even amplified as more unlabeled data will be mislabeled as head classes when the class distribution of labeled and unlabeled datasets are mismatched. To solve this problem, we propose a novel method named ComPlementary Experts (CPE). Specifically, we train multiple experts to model various class distributions, each of them yielding high-quality pseudo-labels within one form of class distribution. Besides, we introduce Classwise Batch Normalization for CPE to avoid performance degradation caused by feature distribution mismatch between head and non-head classes. CPE achieves state-of-the-art performances on CIFAR-10-LT, CIFAR-100-LT, and STL-10-LT dataset benchmarks. For instance, on CIFAR-10-LT, CPE improves test accuracy by over >2.22% compared to baselines. Code is available at https://github.com/machengcheng2016/CPE-LTSSL.

show abstract

Section: Long-tailed Learningmentioning

confidence: 99%

Three Heads Are Better than One: Complementary Experts for Long-Tailed Semi-supervised Learning

Ma,

Elezi,

Deng

et al. 2024

AAAI

View full text Add to dashboard Cite

show abstract

“…The weights of similarity between nodes are calculated by Gaussian functions. They also suggest additional measure that any two connected nodes should not have a "high similarity difference" [44].…”

Section: Semi Supervised Learning In Record Linkagementioning

confidence: 99%

“…Similarly to Lee et al [44] they had access to a genealogical network consisting of over 100,000 individuals. This was constructed by an single genealogist over a long period of time.…”

Section: Semi Supervised Learning In Record Linkagementioning

confidence: 99%

“…Since the data that was used as 'ground truth' in the work by Malmi et al [25] and Lee et al [44] is based upon human generated data there may be a mistakes or inaccuracies. Malmi et al accept the fact that this human error will then affect the performance of the algorithms and the conclusions drawn from any results.…”

Section: Semi Supervised Learning In Record Linkagementioning

confidence: 99%

“…An example of using natural groupings can be seen in the work done by Lee et al where they use the natural grouping created by blood relations to infer political opinions [44]. Had they not done this their task would have been almost impossible.…”

Section: What Is Missing?mentioning

confidence: 99%

See 2 more Smart Citations

Record linkage on Māori land data to find and connect the missing

Chan¹

View full text Add to dashboard Cite

<p>We consider probabilistic modelling for accomplishing record linkage across two large scale publicly available data sources: New Zealand Births Deaths and Marriages (BDM), and Māori Land Online (MLO). We undertake this project in the context of te ao Māori, integrating mātauranga Māori principles into the work. We present several methods for record linkage and several novel ways to reject false linkages.</p>

show abstract

Adaptive graph regularized nonnegative matrix factorization for data representation

Zhang

et al. 2019

Appl Intell

View full text Add to dashboard Cite

Historical inference based on semi-supervised learning

Cited by 12 publications

References 22 publications

Three Heads Are Better than One: Complementary Experts for Long-Tailed Semi-supervised Learning

Three Heads Are Better than One: Complementary Experts for Long-Tailed Semi-supervised Learning

Record linkage on Māori land data to find and connect the missing

Adaptive graph regularized nonnegative matrix factorization for data representation

Contact Info

Product

Resources

About