Hierarchical Block Aggregation Network for Long-Tailed Visual Recognition

Pang, Shanmin; Wang, Weiye; Zhang, Renzhong; Wu, Hao

doi:10.2139/ssrn.4364285

Cited by 4 publications

(5 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1) Ensemble Learning: Ensemble learning has been a commonly used strategy that averages (sometimes weighted) the output of multiple models trained with different data proportions [217], data distributions [218], model architectures [219] and so on. In healthcare applications, missing data/input variables and even the whole sensor modalities are common, which are difficult to handle in developing appropriate machine learning models.…”

Section: F Learn With Multiple Modelsmentioning

confidence: 99%

Beyond Supervised Learning for Pervasive Healthcare

Deligianni

Han

et al. 2024

IEEE Rev. Biomed. Eng.

View full text Add to dashboard Cite

The integration of machine/deep learning and sensing technologies is transforming healthcare and medical practice. However, inherent limitations in healthcare data, namely scarcity, quality, and heterogeneity, hinder the effectiveness of supervised learning techniques which are mainly based on pure statistical fitting between data and labels. In this paper, we first identify the challenges present in machine learning for pervasive healthcare and we then review the current trends beyond fully supervised learning that are developed to address these three issues. Rooted in the inherent drawbacks of empirical risk minimization that underpins pure fully supervised learning, this survey summarizes seven key lines of learning strategies, to promote the generalization performance for real-world deployment. In addition, we point out several directions that are emerging and promising in this area, to develop data-efficient, scalable, and trustworthy computational models, and to leverage multi-modality and multi-source sensing informatics, for pervasive healthcare.

show abstract

Section: F Learn With Multiple Modelsmentioning

confidence: 99%

Beyond Supervised Learning for Pervasive Healthcare

Deligianni

Han

et al. 2024

IEEE Rev. Biomed. Eng.

View full text Add to dashboard Cite

show abstract

“…Much research has been done on the class-imbalance problem [15][16][17][18][19][20]40], and different solutions have been proposed to solve this problem including under-sampling and over-sampling [41,42], reconciliation of loss function [15,17,43,44], and learning paradigms such as self-supervised learning [16,45], transfer learning [18], ensemble learning [46,47], metalearning [48], and metric learning [49]. All these methods have been used in the scenario of a single domain and use the data splits for all participants from the same domain, while we extend the data heterogeneity problem to multi-domain and imbalance classes in FL environment.…”

Section: Related Work 21 Class-imbalance and Label Distributionmentioning

confidence: 99%

Exploiting data diversity in multi-domain federated learning

Ahmad Madni,

Muhammad Umer,

Luca Foresti

2024

Mach. Learn.: Sci. Technol.

View full text Add to dashboard Cite

Federated Learning (FL) is an evolving machine learning technique that allows collaborative model training without sharing the original data among participants. In real-world scenarios, data residing at multiple clients are often heterogeneous in terms of different resolutions, magnifications, scanners, or imaging protocols, and thus challenging for global FL model convergence in collaborative training. Most of the existing FL methods consider data heterogeneity within one domain by assuming same data variation in each client site. In this paper, we consider data heterogeneity in FL with different domains of heterogeneous data by raising the problems of domain-shift, class-imbalance, and missing data. We propose a method, MDFL (Multi-Domain Federated Learning) as a solution to heterogeneous training data from multiple domains by training robust Transformer model. We use two loss functions, one for correctly predicting class labels and other for encouraging similarity and dissimilarity over latent features, to optimize the global FL model. We perform various experiments using different convolution-based networks and non-convolutional Transformer architectures on multi-domain datasets. We evaluate the proposed approach on benchmark datasets and compare with the existing FL methods. Our results show the superiority of the proposed approach which performs better in term of robust FL global model than the exiting methods.

show abstract

“…ISSN 2616-5775 Vol. [10] 43.7 64.3 37.1 8.2 LDAM-DRW [1] 49.8 60.4 46.9 30.7 Decouple-cRT [4] 47.3 58.8 44.0 26.1 Balanced Softmax [9] 51.4 62.2 48.8 29.8 LADE [16] 51.9 62.3 49.3 31.2 DisAlign [17] 52.2 60.8 50.4 34.7 Seesaw [21] 50.4 67.1 45.2 21.4 MARC [3] 52.…”

Section: Experiments On Imagenet-ltmentioning

confidence: 99%

“…Specifically, prior to any calibration efforts, it was observed that classes featuring a higher volume of images exhibited larger margins and logits. Conversely, classes with fewer images displayed reduced margins and logits [3] . Moreover, we found that uncalibrated margins and logits will substantially detract from image classification performance.…”

Section: Introductionmentioning

confidence: 96%

A Mixup-based Margin Aware and Calibration Model for Imbalanced in Soil Classification

2024

AJCIS

View full text Add to dashboard Cite

In the process of soil image classification, the issue of class imbalance occurs, which leads to a decline in the generalization performance of the classifier due to the lack of data from minority classes. We investigated the effectiveness of Mixup through margin statistical analysis and successfully improved the deep imbalanced classification with uneven margins. Additionally, we investigated the relationship between margins and logits, and empirically discovered that uncalibrated margins exhibit a positive correlation with logits. Based on this revelation, we propose a Mixup-based margin-aware and calibration model to address the challenge of handling imbalanced soil image classification data. We conducted experiments using the Soil dataset and additionally tested the generalization capabilities of our method on the CIFAR10-LT, CIFAR100-LT, and ImageNet-LT datasets. The experimental results indicate that our approach achieved impressive results.

show abstract

Hierarchical Block Aggregation Network for Long-Tailed Visual Recognition

Cited by 4 publications

References 0 publications

Beyond Supervised Learning for Pervasive Healthcare

Beyond Supervised Learning for Pervasive Healthcare

Exploiting data diversity in multi-domain federated learning

A Mixup-based Margin Aware and Calibration Model for Imbalanced in Soil Classification

Contact Info

Product

Resources

About