Development of a Deep Learning Model for Early Alzheimer’s Disease Detection from Structural MRIs and External Validation on an Independent Cohort

S, Liu; Masurkar, Arjun V.; Rusinek, Henry; Chen, Jingyun; Zhang, B; Zhu, Weiliang; Fernandez‐Granda, Carlos; Razavian, Narges

doi:10.1101/2021.05.28.21257318

Cited by 3 publications

(5 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Eric et al [34] applied a label correction model to distinguish noisy and clean labels. Liu [17] first investigated the early-learning phenomenon to counteract the influence of noisy labels. Huang et al [35] constructed a Gaussian mixture model to handle noisy labels while training twin contrastive learning models for robust representations.…”

Section: Noisy Learningmentioning

confidence: 99%

MLTU: Mixup Long-Tail Unsupervised Zero-Shot Image Classification on Vision-Language Models

Jia,

Ye,

Mei

et al. 2024

Preprint

View full text Add to dashboard Cite

Vision-language models, such as Contrastive Language-Image Pretraining (CLIP), have demonstrated powerful capabilities in image classification under zero-shot settings. However, current Zero-Shot Learning (ZSL) relies on manually tagged samples of known classes through supervised learning, resulting in a waste of labor costs and limitations on foreseeable classes in real-world applications. To address these challenges, we propose the Mixup Long-Tail Unsupervised (MLTU) approach for open-world ZSL problems. The proposed approach employed a novel long-tail mixup loss that integrated class-based re-weighting assignments with a given mixup factor for each mixed visual embedding. To mitigate the adverse impact over time, we adopted a noisy learning strategy to filter out samples that generated incorrect labels. We reproduced the unsupervised results of existing state-of-the-art long-tail and noisy learning approaches. Experimental results demonstrate that MLTU achieves significant improvements in classification compared to these proven existing approaches on public datasets. Moreover, it serves as a plug-and-play solution for amending previous assignments and enhancing unsupervised performance. MLTU enables the automatic classification and correction of incorrect predictions caused by the projection bias of CLIP.

show abstract

Section: Noisy Learningmentioning

confidence: 99%

MLTU: Mixup Long-Tail Unsupervised Zero-Shot Image Classification on Vision-Language Models

Jia,

Ye,

Mei

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…However, the identified neighbours or their annotations may be incorrect due to the noisy labels used in training. Moreover, towards the end of training, the network is more prone to overfitting to incorrect examples than in early epochs [11,3]. With the aim of obtaining more robust scores and inspired by the temporal ensembling strategy used in [8,11], we propose to compute a weighted average of the scores obtained for the previous epoch and the current k-NN scores.…”

Section: Proposed Approachmentioning

confidence: 99%

“…Moreover, towards the end of training, the network is more prone to overfitting to incorrect examples than in early epochs [11,3]. With the aim of obtaining more robust scores and inspired by the temporal ensembling strategy used in [8,11], we propose to compute a weighted average of the scores obtained for the previous epoch and the current k-NN scores. Unlike other k-NN sample selection approaches [4,5,6], we compute scores by taking into account the model's evolution during training.…”

Section: Proposed Approachmentioning

confidence: 99%

“…This strategy aims to improve the scores computations by considering both past and current scores. Our approach is different from the method introduced in [11], which applies temporal ensembling to obtain network predictions used for computing a new robust regularizer [11]. In contrast, we introduce a temporal ensembling procedure for estimating scores that measure how likely the label of a sample is to be correct.…”

Section: Proposed Approachmentioning

confidence: 99%

See 1 more Smart Citation

Temporal Ensembling-based Deep k-Nearest Neighbours for Learning with Noisy Labels

Albu

2023

ESANN 2023 Proceesdings

View full text Add to dashboard Cite

Label noise can significantly affect the generalization of deep neural networks. Nevertheless, it is omnipresent in real world applications. This paper introduces an approach for identifying the samples from a dataset which are likely to have correct annotations. The proposed method computes the agreement of a sample with its nearest neighbours retrieved from the feature space provided by a neural network. We introduce a temporal ensembling strategy which takes into account the agreement scores obtained by a sample during previous training epochs. The superiority of our approach over several baselines is shown on image classification datasets.

show abstract

“…Recently, various methods have been proposed to relax this strong clean subset assumption by taking advantage of the characteristic that noisy labels are more difficult-tolearn than clean samples. For example, such difficult-to-learn samples are guided by regularizers (Liu et al 2020;Cao et al 2019Cao et al , 2020, giving lower weights (Wang et al 2019;Zhang and Sabuncu 2018), cleansing out (Mirzasoleiman, Cao, and Leskovec 2020;Wu et al 2020;Pleiss et al 2020;Han et al 2018;Yu et al 2019), or utilizing a semi-supervised learning algorithm by considering them as unlabeled samples Li, Socher, and Hoi 2019).…”

Section: Introductionmentioning

confidence: 99%

Denoising after Entropy-Based Debiasing a Robust Training Method for Dataset Bias with Noisy Labels

Ahn

Yun

2023

AAAI

View full text Add to dashboard Cite

Improperly constructed datasets can result in inaccurate inferences. For instance, models trained on biased datasets perform poorly in terms of generalization (i.e., dataset bias). Recent debiasing techniques have successfully achieved generalization performance by underestimating easy-to-learn samples (i.e., bias-aligned samples) and highlighting difficult-to-learn samples (i.e., bias-conflicting samples). However, these techniques may fail owing to noisy labels, because the trained model recognizes noisy labels as difficult-to-learn and thus highlights them. In this study, we find that earlier approaches that used the provided labels to quantify difficulty could be affected by the small proportion of noisy labels. Furthermore, we find that running denoising algorithms before debiasing is ineffective because denoising algorithms reduce the impact of difficult-to-learn samples, including valuable bias-conflicting samples. Therefore, we propose an approach called denoising after entropy-based debiasing, i.e., DENEB, which has three main stages. (1) The prejudice model is trained by emphasizing (bias-aligned, clean) samples, which are selected using a Gaussian Mixture Model. (2) Using the per-sample entropy from the output of the prejudice model, the sampling probability of each sample that is proportional to the entropy is computed. (3) The final model is trained using existing denoising algorithms with the mini-batches constructed by following the computed sampling probability. Compared to existing debiasing and denoising algorithms, our method achieves better debiasing performance on multiple benchmarks.

show abstract

Development of a Deep Learning Model for Early Alzheimer’s Disease Detection from Structural MRIs and External Validation on an Independent Cohort

Cited by 3 publications

References 41 publications

MLTU: Mixup Long-Tail Unsupervised Zero-Shot Image Classification on Vision-Language Models

MLTU: Mixup Long-Tail Unsupervised Zero-Shot Image Classification on Vision-Language Models

Temporal Ensembling-based Deep k-Nearest Neighbours for Learning with Noisy Labels

Denoising after Entropy-Based Debiasing a Robust Training Method for Dataset Bias with Noisy Labels

Contact Info

Product

Resources

About