Distribution Of Label Research Articles

FedFixer: Mitigating Heterogeneous Label Noise in Federated Learning

Federated Learning (FL) heavily depends on label quality for its performance. However, the label distribution among individual clients is always both noisy and heterogeneous. The high loss incurred by client-specific samples in heterogeneous label noise poses challenges for distinguishing between client-specific and noisy label samples, impacting the effectiveness of existing label noise learning approaches. To tackle this issue, we propose FedFixer, where the personalized model is introduced to cooperate with the global model to effectively select clean client-specific samples. In the dual models, updating the personalized model solely at a local level can lead to overfitting on noisy data due to limited samples, consequently affecting both the local and global models’ performance. To mitigate overfitting, we address this concern from two perspectives. Firstly, we employ a confidence regularizer to alleviate the impact of unconfident predictions caused by label noise. Secondly, a distance regularizer is implemented to constrain the disparity between the personalized and global models. We validate the effectiveness of FedFixer through extensive experiments on benchmark datasets. The results demonstrate that FedFixer can perform well in filtering noisy label samples on different clients, especially in highly heterogeneous label noise scenarios.

Point Transformer with Federated Learning for Predicting Breast Cancer HER2 Status from Hematoxylin and Eosin-Stained Whole Slide Images

Directly predicting human epidermal growth factor receptor 2 (HER2) status from widely available hematoxylin and eosin (HE)-stained whole slide images (WSIs) can reduce technical costs and expedite treatment selection. Accurately predicting HER2 requires large collections of multi-site WSIs. Federated learning enables collaborative training of these WSIs without gigabyte-size WSIs transportation and data privacy concerns. However, federated learning encounters challenges in addressing label imbalance in multi-site WSIs from the real world. Moreover, existing WSI classification methods cannot simultaneously exploit local context information and long-range dependencies in the site-end feature representation of federated learning. To address these issues, we present a point transformer with federated learning for multi-site HER2 status prediction from HE-stained WSIs. Our approach incorporates two novel designs. We propose a dynamic label distribution strategy and an auxiliary classifier, which helps to establish a well-initialized model and mitigate label distribution variations across sites. Additionally, we propose a farthest cosine sampling based on cosine distance. It can sample the most distinctive features and capture the long-range dependencies. Extensive experiments and analysis show that our method achieves state-of-the-art performance at four sites with a total of 2687 WSIs. Furthermore, we demonstrate that our model can generalize to two unseen sites with 229 WSIs. Code is available at: https://github.com/boyden/PointTransformerFL

Exploiting Label Skews in Federated Learning with Model Concatenation

Federated Learning (FL) has emerged as a promising solution to perform deep learning on different data owners without exchanging raw data. However, non-IID data has been a key challenge in FL, which could significantly degrade the accuracy of the final model. Among different non-IID types, label skews have been challenging and common in image classification and other tasks. Instead of averaging the local models in most previous studies, we propose FedConcat, a simple and effective approach that concatenates these local models as the base of the global model to effectively aggregate the local knowledge. To reduce the size of the global model, we adopt the clustering technique to group the clients by their label distributions and collaboratively train a model inside each cluster. We theoretically analyze the advantage of concatenation over averaging by analyzing the information bottleneck of deep neural networks. Experimental results demonstrate that FedConcat achieves significantly higher accuracy than previous state-of-the-art FL methods in various heterogeneous label skew distribution settings and meanwhile has lower communication costs. Our code is publicly available at https://github.com/sjtudyq/FedConcat.

Robust Visual Recognition with Class-Imbalanced Open-World Noisy Data

Learning from open-world noisy data, where both closed-set and open-set noise co-exist in the dataset, is a realistic but underexplored setting. Only recently, several efforts have been initialized to tackle this problem. However, these works assume the classes are balanced when dealing with open-world noisy data. This assumption often violates the nature of real-world large-scale datasets, where the label distributions are generally long-tailed, i.e. class-imbalanced. In this paper, we study the problem of robust visual recognition with class-imbalanced open-world noisy data. We propose a probabilistic graphical model-based approach: iMRF to achieve label noise correction that is robust to class imbalance via an efficient iterative inference of a Markov Random Field (MRF) in each training mini-batch. Furthermore, we design an agreement-based thresholding strategy to adaptively collect clean samples from all classes that includes corrected closed-set noisy samples while rejecting open-set noisy samples. We also introduce a noise-aware balanced cross-entropy loss to explicitly eliminate the bias caused by class-imbalanced data. Extensive experiments on several benchmark datasets including synthetic and real-world noisy datasets demonstrate the superior performance robustness of our method over existing methods. Our code is available at https://github.com/Na-Z/LIOND.

Transportable Representations for Domain Generalization

One key assumption in machine learning literature is that the testing and training data come from the same distribution, which is often violated in practice. The anchors that allow generalizations to take place are causal, and provenient in terms of the stability and modularity of the mechanisms underlying the system of variables. Building on the theory of causal transportability, we define the notion of ``transportable representations", and show that these representations are suitable candidates for the domain generalization task. Specifically, considering that the graphical assumptions about the underlying system are provided, the transportable representations can be characterized accordingly, and the distribution of label conditioned on the representation can be computed in terms of the source distributions. Finally, we relax the assumption of having access to the underlying graph by proving a graphical-invariance duality theorem, which delineates certain probabilistic invariances present in the source data as a sound and complete criterion for generalizable classification. Our findings provide a unifying theoretical basis for several existing approaches to the domain generalization problem.

NodeMixup: Tackling Under-Reaching for Graph Neural Networks

Graph Neural Networks (GNNs) have become mainstream methods for solving the semi-supervised node classification problem. However, due to the uneven location distribution of labeled nodes in the graph, labeled nodes are only accessible to a small portion of unlabeled nodes, leading to the under-reaching issue. In this study, we firstly reveal under-reaching by conducting an empirical investigation on various well-known graphs. Then, we demonstrate that under-reaching results in unsatisfactory distribution alignment between labeled and unlabeled nodes through systematic experimental analysis, significantly degrading GNNs' performance. To tackle under-reaching for GNNs, we propose an architecture-agnostic method dubbed NodeMixup. The fundamental idea is to (1) increase the reachability of labeled nodes by labeled-unlabeled pairs mixup, (2) leverage graph structures via fusing the neighbor connections of intra-class node pairs to improve performance gains of mixup, and (3) use neighbor label distribution similarity incorporating node degrees to determine sampling weights for node mixup. Extensive experiments demonstrate the efficacy of NodeMixup in assisting GNNs in handling under-reaching. The source code is available at https://github.com/WeigangLu/NodeMixup.

Harnessing the Power of Beta Scoring in Deep Active Learning for Multi-Label Text Classification

Within the scope of natural language processing, the domain of multi-label text classification is uniquely challenging due to its expansive and uneven label distribution. The complexity deepens due to the demand for an extensive set of annotated data for training an advanced deep learning model, especially in specialized fields where the labeling task can be labor-intensive and often requires domain-specific knowledge. Addressing these challenges, our study introduces a novel deep active learning strategy, capitalizing on the Beta family of proper scoring rules within the Expected Loss Reduction framework. It computes the expected increase in scores using the Beta Scoring Rules, which are then transformed into sample vector representations. These vector representations guide the diverse selection of informative sample, directly linking this process to the model's expected proper score. Comprehensive evaluations across both synthetic and real datasets reveal our method's capability to often outperform established acquisition techniques in multi-label text classification, presenting encouraging outcomes across various architectural and dataset scenarios.

De-biased Attention Supervision for Text Classification with Causality

In text classification models, while the unsupervised attention mechanism can enhance performance, it often produces attention distributions that are puzzling to humans, such as assigning high weight to seemingly insignificant conjunctions. Recently, numerous studies have explored Attention Supervision (AS) to guide the model toward more interpretable attention distributions. However, such AS can impact classification performance, especially in specialized domains. In this paper, we address this issue from a causality perspective. Firstly, we leverage the causal graph to reveal two biases in the AS: 1) Bias caused by the label distribution of the dataset. 2) Bias caused by the words' different occurrence ranges that some words can occur across labels while others only occur in a particular label. We then propose a novel De-biased Attention Supervision (DAS) method to eliminate these biases with causal techniques. Specifically, we adopt backdoor adjustment on the label-caused bias and reduce the word-caused bias by subtracting the direct causal effect of the word. Through extensive experiments on two professional text classification datasets (e.g., medicine and law), we demonstrate that our method achieves improved classification accuracy along with more coherent attention distributions.

Communication Efficiency and Non-Independent and Identically Distributed Data Challenge in Federated Learning: A Systematic Mapping Study

Federated learning has emerged as a promising approach for collaborative model training across distributed devices. Federated learning faces challenges such as Non-Independent and Identically Distributed (non-IID) data and communication challenges. This study aims to provide in-depth knowledge in the federated learning environment by identifying the most used techniques for overcoming non-IID data challenges and techniques that provide communication-efficient solutions in federated learning. The study highlights the most used non-IID data types, learning models, and datasets in federated learning. A systematic mapping study was performed using six digital libraries, and 193 studies were identified and analyzed after the inclusion and exclusion criteria were applied. We identified that enhancing the aggregation method and clustering are the most widely used techniques for non-IID data problems (used in 18% and 16% of the selected studies), and a quantization technique was the most common technique in studies that provide communication-efficient solutions in federated learning (used in 27% and 15% of the selected studies). Additionally, our work shows that label distribution skew is the most used case to simulate a non-IID environment, specifically, the quantity label imbalance. The supervised learning model CNN model is the most commonly used learning model, and the image datasets MNIST and Cifar-10 are the most widely used datasets when evaluating the proposed approaches. Furthermore, we believe the research community needs to consider the client’s limited resources and the importance of their updates when addressing non-IID and communication challenges to prevent the loss of valuable and unique information. The outcome of this systematic study will benefit federated learning users, researchers, and providers.

PalScDiff: A diffusion-based framework with progressive augmentation learning and semantic consistency for breast ultrasound tumor segmentation

Background: Breast cancer diagnosis relies on accurate lesion segmentation in medical images. Automated computer-aided diagnosis reduces clinician workload and improves efficiency, but existing image segmentation methods face challenges in model performance and generalization. Objective: This study aims to develop a generative framework using a denoising diffusion model for efficient and accurate breast cancer lesion segmentation in medical images. Methods: We design a novel generative framework, PalScDiff, that leverages a denoising diffusion probabilistic model to reconstruct the label distribution for medical images, thereby enabling the sampling of diverse, plausible segmentation outcomes. Specifically, with the condition of the corresponding image, PalScDiff learns to estimate the masses region probability through denoising step by step. Furthermore, we design a Progressive Augmentation Learning strategy to incrementally handle segmentation challenges of irregular and blurred tumors. Moreover, multi-round sampling is employed to achieve robust breast mass segmentation. Results: Our experimental results show that PalScDiff outperforms established models such as U-Net and transformer-based alternatives, achieving an accuracy of 95.15%, precision of 79.74%, Dice coefficient of 77.61%, and Intersection over Union (IOU) of 81.51% . Conclusion: The proposed model demonstrates promising capabilities for accurate and efficient computer-aided segmentation of breast cancer.

Causality-Driven Intra-class Non-equilibrium Label-Specific Features Learning

In multi-label learning, label-specific feature learning can effectively avoid some ineffectual features that interfere with the classification performance of the model. However, most of the existing label-specific feature learning algorithms improve the performance of the model for classification by constraining the solution space through label correlation. The non-equilibrium of the label distribution not only leads to some spurious correlations mixed in with the calculated label correlations but also diminishes the performance of the classification model. Causal learning can improve the classification performance and robustness of the model by capturing real causal relationships from limited data. Based on this, this paper proposes a causality-driven intra-class non-equilibrium label-specific features learning, named CNSF. Firstly, the causal relationship between the labels is learned by the Peter-Clark algorithm. Secondly, the label density of all instances is calculated by the intra-class non-equilibrium method, which is used to relieve the non-equilibrium distribution of original labels. Then, the correlation of the density matrix is calculated using cosine similarity and combined with causality to construct the causal density correlation matrix, to solve the problem of spurious correlation mixed in the label correlation obtained by traditional methods. Finally, the causal density correlation matrix is used to induce label-specific feature learning. Compared with eight state-of-the-art multi-label algorithms on thirteen datasets, the experimental results prove the reasonability and effectiveness of the algorithms in this paper.

IoT data sharing technology based on blockchain and federated learning algorithms

To share data on Internet of Things devices more securely, accurately, and efficiently, this study designs a layered sharing architecture based on blockchain and federated learning. This architecture achieves efficient and secure Internet of Things data sharing through client node clustering and blockchain consensus processes. In addition, to address the issue of imbalanced distribution of data labels in system devices, a device clustering federated learning algorithm based on label similarity is designed to improve the accuracy and stability of the model. The experimental results showed that under independent synchronous data distribution and non independent synchronous data distribution, the research algorithm achieved a 95 % accuracy after 30 iterations, and the communication cost was relatively low. When testing algorithm stability under non independent synchronous data distribution, the more label categories there are, the higher the accuracy. When the label category M = 12, the accuracy could reach 96.0 %. In the medical sharing system of a certain hospital, the research system took about 42.9 % less time to extract information than the original system, and the accuracy could be maintained at over 98 %. This research method can effectively solve the problem of uneven distribution of device data labels, and improve the data transmission efficiency and accuracy of Internet of Things data sharing systems. Moreover, this method can also reduce the impact of malicious nodes on the global model, providing technical support for data transmission and security protection in other fields.

AG-Meta: Adaptive graph meta-learning via representation consistency over local subgraphs

Graph meta-learning has recently received significantly increased attention by virtue of its potential to extract common and transferable knowledge from learning different tasks on a graph. Existing methods for graph meta-learning usually leverage local subgraphs to transfer subgraph-specific information. However, they inherently face the challenge of imbalanced subgraphs due to inconsistent node density and different label distributions over local subgraphs. This paper proposes an adaptive graph meta-learning framework (AG-Meta) for learning the consistent and transferable representation of a graph in a way that can adapt to imbalanced subgraphs. Specifically, AG-Meta first learns the structural representation of subgraphs with various degrees using an Adaptive Graph Cascade Diffusion Network (AGCDN). AG-Meta then employs a prototype-consistency classifier to produce more accurate transferable inductive representations (also called prototypes) under few-shot settings with different label distributions of a subgraph. In the context of optimizing a model-agnostic meta-learner, a novel metric loss is finally introduced to achieve structural representation and prototype consistency. Extensive experiments are conducted to compare AG-Meta against baselines on five real-world networks, which validates that AG-Meta outperforms the state-of-the-art approaches.

Active Crowdsourcing for Multilabel Annotation.

Multilabel annotation is a critical step to generate training sets when learning classification models in various application domains, but asking domain experts to provide labels is usually time-consuming and expensive, which cannot meet the current requirement of the fast evolution of the models in the big data era. Although crowdsourcing provides a fast solution to acquire labels for multilabel learning, it faces the risk of high data acquisition cost and low label quality. This article proposes a novel one-coin label-dependent active crowdsourcing (OCLDAC) method to iteratively query noisy labels from crowd workers and learn multilabel classification models. In each iteration of active learning, integrated labels of instances are first inferred by a novel one-coin label-dependent model, which utilizes a mixture of multiple independent Bernoulli distributions to explore and exploit correlations among the labels to increase the accuracy of truth inference. Then, instances, labels, and workers are selected according to the novel strategies that incorporate the distribution of noisy labels, the prediction probability of learning models, label correlations, and the reliability of crowd workers. Simulations on eight multilabel datasets and evaluation on one real-world crowdsourcing dataset consistently show that the proposed OCLDAC significantly outperforms the state-of-the-art methods and their variants.

Graphical Modeling for Multi-Source Domain Adaptation.

Multi-Source Domain Adaptation (MSDA) focuses on transferring the knowledge from multiple source domains to the target domain, which is a more practical and challenging problem compared to the conventional single-source domain adaptation. In this problem, it is essential to model multiple source domains and target domain jointly, and an effective domain combination scheme is also highly required. The graphical structure among different domains is useful to tackle these challenges, in which the interdependency among various instances/categories can be effectively modeled. In this work, we propose two types of graphical models, i.e. Conditional Random Field for MSDA (CRF-MSDA) and Markov Random Field for MSDA (MRF-MSDA), for cross-domain joint modeling and learnable domain combination. In a nutshell, given an observation set composed of a query sample and the semantic prototypes (i.e. representative category embeddings) on various domains, the CRF-MSDA model seeks to learn the joint distribution of labels conditioned on the observations. We attain this goal by constructing a relational graph over all observations and conducting local message passing on it. By comparison, MRF-MSDA aims to model the joint distribution of observations over different Markov networks via an energy-based formulation, and it can naturally perform label prediction by summing the joint likelihoods over several specific networks. Compared to the CRF-MSDA counterpart, the MRF-MSDA model is more expressive and possesses lower computational cost. We evaluate these two models on four standard benchmark data sets of MSDA with distinct domain shift and data complexity, and both models achieve superior performance over existing methods on all benchmarks. In addition, the analytical studies illustrate the effect of different model components and provide insights about how the cross-domain joint modeling performs.

Informative Training Data for Efficient Property Prediction in Metal-Organic Frameworks by Active Learning.

In recent data-driven approaches to material discovery, scenarios where target quantities are expensive to compute and measure are often overlooked. In such cases, it becomes imperative to construct a training set that includes the most diverse, representative, and informative samples. Here, a novel regression tree-based active learning algorithm is employed for such a purpose. It is applied to predict the band gap and adsorption properties of metal-organic frameworks (MOFs), a novel class of materials that results from the virtually infinite combinations of their building units. Simpler and low dimensional descriptors, such as those based on stoichiometric and geometric properties, are used to compute the feature space for this model owing to their ability to better represent MOFs in the low data regime. The partitions given by a regression tree constructed on the labeled part of the data set are used to select new samples to be added to the training set, thereby limiting its size while maximizing the prediction quality. Tests on the QMOF, hMOF, and dMOF data sets reveal that our method constructs small training data sets to learn regression models that predict the target properties more efficiently than existing active learning approaches, and with lower variance. Specifically, our active learning approach is highly beneficial when labels are unevenly distributed in the descriptor space and when the label distribution is imbalanced, which is often the case for real world data. The regions defined by the tree help in revealing patterns in the data, thereby offering a unique tool to efficiently analyze complex structure-property relationships in materials and accelerate materials discovery.

Multi-label feature selection based on fuzzy rough sets with metric learning and label enhancement

Multi-label feature selection based on fuzzy rough sets, as a key step of multi-label data preprocessing, has been widely concerned by scholars in recent years. Most of the existing multi-label feature selection algorithms directly treat labels as logical labels and use a single distance metric to describe similarity. However, the variability of label descriptions and the limitations of a single distance measure should not be overlooked. In this paper, we propose a fuzzy rough set model with metric learning and label enhancement. Specifically, we use a kernel membership label enhancement algorithm based on JS divergence to convert logical labels into numerical labels, which not only reflects the importance of different labels, but also takes into account the differences in label distribution. In addition, a multi-metric learning algorithm is proposed for multi-label learning, in which the metric distance function under the label space and feature space can be learned autonomously. Then, based on the proposed model, we propose a novel multi-label feature selection algorithm based on metric learning and fuzzy rough sets. On this basis, a fast multi-label feature selection algorithm is further designed to improve the computational efficiency. In the experiments, compared with other nine algorithms on real-world datasets, the results show the superiority of the proposed algorithm.

Modelling long medical documents and code associations for explainable automatic ICD coding

Quick and accurate International Classification of Diseases (ICD) code assignment is vital for billing, reimbursement and medical research. Owing to the labour-intensive and error-prone nature of manual coding, automatic ICD coding using deep learning methods has flourished. However, this task is still challenging because of (1) the interpretability of coding, (2) lengthy clinical documents and (3) long-tail label distribution. In the current study, we propose a novel automatic ICD coding framework to address these issues. First, a biomedical-specific pre-trained language model, Clinical-Longformer, is used as an encoder, which generates meaningful representations of long clinical documents by injecting rich medical knowledge and capturing long-distance dependence among tokens. Second, a decoding architecture that combines the multi-synonym attention mechanism, hierarchical curriculum learning and distribution-balance loss is designed to perform ICD code prediction. The decoder improves the tail-end performance by fully capturing code associations in terms of semantics, structure and co-occurrence. In addition, the label-wise attention mechanism provides the interpretability of prediction. Experimental results on benchmark MIMIC-III datasets indicate that our model achieves higher F1 scores than previous state-of-the-art baselines. Our suggested model is expected to serve as an aid in improving the efficiency of manual ICD coding and to offer insights for other long text classification tasks with multiple label associations.

Unifying supervised learning and VAEs: coverage, systematics and goodness-of-fit in normalizing-flow based neural network models for astro-particle reconstructions

Neural-network based predictions of event properties in astro-particle physics are getting more and more common. However, in many cases the result is just utilized as a point prediction. Statistical uncertainties, coverage, systematic uncertainties or a goodness-of-fit measure are often not calculated. Here we describe a certain choice of training and network architecture that allows to incorporate all these properties into a single network model. We show that a KL-divergence objective of the joint distribution of data and labels allows to unify supervised learning and variational autoencoders (VAEs) under one umbrella of stochastic variational inference. The unification motivates an extended supervised learning scheme which allows to calculate a goodness-of-fit p-value for the neural network model. Conditional normalizing flows amortized with a neural network are crucial in this construction. We discuss how to calculate coverage probabilities without numerical integration for specific “base-ordered” contours that are unique to normalizing flows. Furthermore we show how systematic uncertainties can be included via effective marginalization during training. The proposed extended supervised training incorporates (1) coverage calculation, (2) systematics and (3) a goodness-of-fit measure in a single machine-learning model. There are in principle no constraints on the shape of the involved distributions, in fact the machinery works with complex multi-modal distributions defined on product spaces like Rn×Sm\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${\\mathbb {R}}^n \ imes {\\mathbb {S}}^m$$\\end{document}. The coverage calculation, however, requires care in its interpretation when the distributions are too degenerate. We see great potential for exploiting this per-event information in event selections or for fast astronomical alerts which require uncertainty guarantees.

ATBBC: Named entity recognition in emergency domains based on joint BERT-BILSTM-CRF adversarial training

With the rapid growth of Internet penetration, identifying emergency information from network news has become increasingly significant for emergency monitoring and early warning. Although deep learning models have been commonly used in Chinese Named Entity Recognition (NER), they require a significant amount of well-labeled training data, which is difficult to obtain for emergencies. In this paper, we propose an NER model that combines bidirectional encoder representations from Transformers (BERT), bidirectional long-short-term memory (BILSTM), and conditional random field (CRF) based on adversarial training (ATBBC) to address this issue. Firstly, we constructed an emergency dataset (ED) based on the classification and coding specifications of the national emergency platform system. Secondly, we utilized the BERT pre-training model with adversarial training to extract text features. Finally, BILSTM and CRF were used to predict the probability distribution of entity labels and decode the probability distribution into corresponding entity labels.Experiments on the ED show that our model achieves an F1-score of 85.39% on the test dataset, which proves the effectiveness of our model.