Noise Label Research Articles

Bearing fault diagnosis method based on improved meta-ResNet and sample weighting under noise label

Bearing fault diagnosis method based on improved meta-ResNet and sample weighting under noise label

Profiling effects of filtering noise labels on learning performance

Deciding whether a label in a dataset is noisy or not depends on factors such as filtering levels, data patterns, relabeling policies, and pre-defined regulations. By focusing on filtering acceptance levels (FAL), a decision support model (named DSM-ENL) is proposed and implemented on real datasets that are granulated, dealing with noise label detection and profiling effects of 50 FALs (51 %–100 %) on noise rate (NoR), classification accuracy (CA), and area under ROC curve (AUC). A case used to demonstrate the DSM-ENL is illustrated, followed by experiments on 27 unsupervised and 29 supervised granulated datasets using five classifiers with 70 %–30 % training and testing criteria. The findings revealed that: (1) the best CAs among classifiers vary with FALs; (2) NoR, CA, and AUC oscillate as FAL increases; (3) lower FAL boundaries (from 0.51 to 0.60) are more likely to lead to higher CAs; and (4) correlation of coefficient among dataset characteristics, FAL, NoR, CA, and AUC are not high compared to that of CA and NoR and that of CA and AUC. The research demonstrates the value of profiling the effects of FALs applied for noise-label datasets on learning performance.

Classification with noisy labels through tree-based models and semi-supervised learning: A case study of lithology identification

Lithology identification is a crucial task for reservoir characterization and evaluation. There exists an intricate non-linear response between formation lithology and logging data. However, it is difficult to avoid lithology mislabeling due to human error and interpretation coarsening, and label quality can seriously affect the effectiveness of supervised learning. The presence of noisy labels makes it essential to learn with noisy labels. Noise-filtering methods and noise-robust algorithms only concentrate on a singular aspect of data or algorithm. In this paper, hybrid noise label filtering and correction framework for lithology identification (HNFCL) is proposed. Isolation forest is utilized to detect suspicious data, as it is efficient and fast. Baseline classifiers are built by ensemble tree models. In particular, the labels of abnormal data are removed and Tri-training semi-supervised method is introduced to relabel these data, which minimizes the loss of valid training data. Comprehensive experiments of the HNFCL framework, noise filtering methods and deep neural network methods with optimized loss functions were carried out in the industrial application of logging lithology identification. HNFCL achieved average accuracy of 87.94% and 94.93% in two study wells. These results outperformed the noise filtering methods and showed no significant difference from the state-of-the-art method. The correction of noise by HNFCL will provide a prospect for lithology identification applications.

The EU Tyre Noise Label: The problem with measuring the noise level of only a few of all tyre variants

The STEER project, described in another Inter-Noise 2022 paper, has evaluated the performance of the noise label of the European tyre label. The major finding was that uncertainties of the tyre/road noise measurements were higher than should be accepted. One of the worst uncertainty sources was found to be the common prac-tice to measure only some tyres of all sizes or variants within a tyre line, to save money. Generally, only the nois-iest tyre(s) is/are measured individually and other tyres in that line get the same level, which means that many if not most tyres are labelled with too high noise levels. Then consumers cannot find the quietest tyres. STEER lacked resources to study this problem, but a Swedish project supported STEER in this respect, by making it pos-sible to measure noise of 53 tyres selected from tyre lines of three major tyre manufacturers. The results showed that even though tyres in each line were labelled with the same noise level, in practice they differed up to 6 dB in noise emission. To avoid this very serious source of uncertainty, a special simplified test is suggested in order to be able to label tyres correctly without too much extra effort.

Project STEER: Improving the EU Tyre Noise Label and its impact on European Roads

Tyre/road noise is the dominant component of overall vehicle noise at medium and high speeds and for cars even at low speeds. Consequently, road traffic noise can be reduced with the proliferation of quieter tyres. One way to achieve this is to give the tyre noise label greater attention among tyre and transportation consumers. Hence, the STEER project has evaluated the relevance and performance of the noise part of the European tyre label, looking at how it works in practice, analyzing its uncertainties and suggesting how it can be improved. Its main finding is that the uncertainties in the measurement of noise level for the label are too high to be acceptable. This paper focusses on the solutions offered by STEER for an improved tyre label. With four main improvements, the overall uncertainty of the current procedure can be halved. A few possible future strategies to increase the market share of quieter tyres have been analyzed and their effects quantified. If the tyre noise label is improved and the market share of quieter tyres can be increased as project STEER proposes, area-wide reductions of up to 3 dB in road traffic noise emissions compared to the present situation are possible.

Give noise a voice so it can be better understood.

The purpose of the European Outdoor Noise Directive (OND) 2000/14/EC is to provide to the buyer of a machine relevant noise information so an informed buying decision favoring low noise machinery will be made. This in return should incentivize manufacturers over time to produce quieter machinery. However, the results of this strategy have yet to meet the expectations. Multiple issues have been recognized and investigated over the last years. As a result two guides were published, one for manufacturers to provide correct noise information and one for buyers in their purchases. Efforts are made to resolve issues with measurement standards. The actual noise label applied on OND-machines is also one of the identified issues at time of purchase. And for market surveillance it provides no real help. So what could be done to improve the situation? What key parts in the process involving different stakeholders and procedures should be addressed? This paper proposes a technical solution to provide understandable noise information to buyers in an effective way by leveraging mobile communication technology and by taking in account insights about human buying behavior. The solution could also help making market surveillance more effective.

Attentive-Adaptive Network for Hyperspectral Images Classification With Noisy Labels

With the development of deep neural networks, hyperpsectral image (HSI) classification systems have achieved a significant improvement. These systems require numerous and accurate labeled hyperspectral data to be adequately trained. However, noisy labels are inherent in real-world hyperspectral systems, resulting in unreliable decisions. To handle noisy labels in hyperpsectral classification, an end-to-end attentive-adaptive network (AAN) is proposed for robust HSI classification training. The goal is to build a classifier with strong generalization capabilities that can be applied to both clean and noisy training sets without explicit noise label pre-treatment. Specifically, a spectral stem network with non-adjacent shortcut is exploited initially to re-distribute the sensitive layers for noisy labels to achieve robust spectral representation. Then, a group-shuffle attention module is proposed to capture the discriminative and robust spatial-spectral features in the presence of noisy labels. Finally, an adaptive noise-robust loss function is developed to fight against noisy labels by learning a parameter to balance the normalized cross entropy (NCE) and reverse cross entropy (RCE). Experimental results on three HSI benchmark datasets with simulated noisy labels demonstrate the effectiveness of AAN on HSI classification.

Partial Label Learning via GANs With Multiclass SVMs and Information Maximization

Partial label learning (PLL), an important branch of weakly supervised learning, addresses the problem that each instance is associated with a set of candidate labels and only one is correct. In this paper, a novel adversarial model PL-MIGAN is proposed to simultaneously mitigate two fundamental issues of generative adversarial networks (GANs) in PLL: label disambiguation performance of discriminator and instance synthesis quality of generator. First of all, multi-class support vector machines (SVMs) applied in discriminator to disambiguate the candidate labels and identify fake instances. This strategy not only improves the disadvantage that traditional supervised loss is unable to perform disambiguation but also reduces the influence of cumulative error caused by noise label propagation. Furthermore, a partial contrastive loss is constructed to extend the self-supervised contrastive approach to PLL, allowing us to effectively leverage ambiguous labels information. Finally, the generator jointly employ mutual information (MI) and partial contrastive loss to estimate the latent distribution of each class label. In addition, in order to reduce the impact of ambiguous information, an iteratively optimization procedure is designed to update the label confidence matrix as conditional information guides the generation of instance classes. As adversarial learning proceeds, both the discriminator and the generator alternately and iteratively boost their performance. Simulation results reveal the overwhelming performance of PL-MIGAN.

JoSDW: Combating Noisy Labels by Dynamic Weight

The real world is full of noisy labels that lead neural networks to perform poorly because deep neural networks (DNNs) are prone to overfitting label noise. Noise label training is a challenging problem relating to weakly supervised learning. The most advanced existing methods mainly adopt a small loss sample selection strategy, such as selecting the small loss part of the sample for network model training. However, the previous literature stopped here, neglecting the performance of the small loss sample selection strategy while training the DNNs, as well as the performance of different stages, and the performance of the collaborative learning of the two networks from disagreement to an agreement, and making a second classification based on this. We train the network using a comparative learning method. Specifically, a small loss sample selection strategy with dynamic weight is designed. This strategy increases the proportion of agreement based on network predictions, gradually reduces the weight of the complex sample, and increases the weight of the pure sample at the same time. A large number of experiments verify the superiority of our method.

CycleGAN based confusion model for cross-species plant disease image migration

The identification and classification of plant diseases is of great significance to ecological protection and deep learning methods have made a great of progress in the common plant diseases identification for specific plant. While faced with the same plant disease of other plants, due to the insufficient or low quality training data, current deep learning methods will be difficult to identify the diseases effectively and accurately. Inspired by the advantages of GAN in dataset expansion, we propose the CycleGAN based confusion model in this paper. In this paper, GAN framework is improved by adding noise label and learn together during training stage, which migrates the data of common plant diseases to the plants with insufficient or low quality data. In order to evaluate the quality of the migrated training dataset among different GAN approaches, we introduce the quality indicators of the migration images such as MMD, FID, EMD etc. We compare our model with other GANs model, and the experimental results show that the proposed model obtains better results in the migration process, which make it more effective for the identification of cross species plant diseases.

Sound Event Detection by Pseudo-Labeling in Weakly Labeled Dataset.

Weakly labeled sound event detection (WSED) is an important task as it can facilitate the data collection efforts before constructing a strongly labeled sound event dataset. Recent high performance in deep learning-based WSED’s exploited using a segmentation mask for detecting the target feature map. However, achieving accurate detection performance was limited in real streaming audio due to the following reasons. First, the convolutional neural networks (CNN) employed in the segmentation mask extraction process do not appropriately highlight the importance of feature as the feature is extracted without pooling operations, and, concurrently, a small size kernel forces the receptive field small, making it difficult to learn various patterns. Second, as feature maps are obtained in an end-to-end fashion, the WSED model would be weak to unknown contents in the wild. These limitations would lead to generating undesired feature maps, such as noise in the unseen environment. This paper addresses these issues by constructing a more efficient model by employing a gated linear unit (GLU) and dilated convolution to improve the problems of de-emphasizing importance and lack of receptive field. In addition, this paper proposes pseudo-label-based learning for classifying target contents and unknown contents by adding ’noise label’ and ’noise loss’ so that unknown contents can be separated as much as possible through the noise label. The experiment is performed by mixing DCASE 2018 task1 acoustic scene data and task2 sound event data. The experimental results show that the proposed SED model achieves the best F1 performance with 59.7% at 0 SNR, 64.5% at 10 SNR, and 65.9% at 20 SNR. These results represent an improvement of 17.7%, 16.9%, and 16.5%, respectively, over the baseline.

Correcting Pseudo Labels with Label Distribution for Unsupervised Domain Adaptive Vulnerable Plaque Detection.

Pseudo-label-based unsupervised domain adaptation (UDA) has increasingly gained interest in medical image analysis, aiming to solve the problem of performance degradation of deep neural networks when dealing with unseen data. Although it has achieved great success, it still faced two significant challenges: improving pseudo labels' precision and mitigating the effects caused by noisy pseudo labels. To solve these problems, we propose a novel UDA framework based on label distribution learning, where the problem is formulated as noise label correcting and can be solved by converting a fixed categorical value (pseudo labels on target data) to a distribution and iteratively update both network parameters and label distribution to correct noisy pseudo labels, and then these labels are used to re-train the model. We have extensively evaluated our framework with vulnerable plaques detection between two IVOCT datasets. Experimental results show that our UDA framework is effective in improving the detection performance of unlabeled target images.

Noise label learning through label confidence statistical inference

Noise label exists widely in real-world data, resulting in the degradation of classification performance. Popular methods require a known noise distribution or additional cleaning supervision, which is usually unavailable in practical scenarios. This paper presents a theoretical statistical method and designs a label confidence inference (LISR) algorithm to handle this issue. For data distribution, we define a statistical function for label inconsistency and analyze its relationship with neighbor radius. For data representation, we define trusted-neighbor, nearest-trusted-neighbor and untrusted-neighbor. For noisy label recognition, we present three inference methods to predict the labels and their confidence. The LISR algorithm establishes a practical statistical model, queries the initial trusted instances, iteratively searches for the trusted instances and corrects labels. We conducted experiments on synthetic, UCI and classic image datasets. The results of significance test verified the effectiveness of LISR and its superiority to the state-of-the-art noise label learning algorithms.

Deep Feature Representation and Similarity Matrix based Noise Label Refinement Method for Efficient Face Annotation

Deep Feature Representation and Similarity Matrix based Noise Label Refinement Method for Efficient Face Annotation

Robust Deep Neural Networks for Road Extraction From Remote Sensing Images

The application of deep neural networks (DNNs) for road extraction from remote sensing images has gained broad interest because of the competence concerning complex nonlinear relations; however, the presence of noisy labels in the training data sets adversely affects the performance of DNNs. The existing methods of improving the robustness of DNNs focus on modeling the noise distribution. However, these approaches are not satisfactory because of the inaccurate high-level image features obtained by the DNNs. To address this issue, we develop a noise probabilistic model for learning the label noise based on the relationship between the input images, noisy labels, and true labels. The key idea of the probabilistic model is to directly explore the information from the input images and apply it to model the label noise. Then, a robust deep neural network (RDNN) is proposed to instantiate the noise probabilistic model, which consists of two important modules: the true label predictor (TLP) and the noise label estimator (NLE). Especially, the TLP is made of a DNN with softmax, which is used to learn the true label distribution. The NLE is applied to model the label noise distribution, which aims to absorb the label noise in the training process. Moreover, to tackle the challenges in the optimization, we deduce a loss function with the novel regularization, which allows the RDNN to conduct effective training on the noise data set. The effectiveness of the proposed method is validated by experiments on three road data sets that contain various resolutions and imaging conditions. The results demonstrate its superiority over state-of-the-art methods in visual performance and classification accuracy.

HERA

Partial label learning (PLL) aims to learn from the data where each training instance is associated with a set of candidate labels, among which only one is correct. Most existing methods deal with this type of problem by either treating each candidate label equally or identifying the ground-truth label iteratively. In this article, we propose a novel PLL approach named HERA, which simultaneously incorporates the HeterogEneous Loss and the SpaRse and Low-rAnk procedure to estimate the labeling confidence for each instance while training the desired model. Specifically, the heterogeneous loss integrates the strengths of both the pairwise ranking loss and the pointwise reconstruction loss to provide informative label ranking and reconstruction information for label identification, whereas the embedded sparse and low-rank scheme constrains the sparsity of ground-truth label matrix and the low rank of noise label matrix to explore the global label relevance among the whole training data, for improving the learning model. Comprehensive ablation study demonstrates the effectiveness of our employed heterogeneous loss, and extensive experiments on both artificial and real-world datasets demonstrate that our method achieves superior or comparable performance against state-of-the-art methods.

An Improved Dice Loss for Pneumothorax Segmentation by Mining the Information of Negative Areas

The lesion regions of a medical image account for only a small part of the image, and a critical imbalance exists in the distribution of the positive and negative samples, which affects the segmentation performance of the lesion regions. Dice loss is beneficial for the image segmentation involving an extreme imbalance of the positive and negative samples but it ignores the background regions, which also contain a large amount of information. In this work, we propose an improved dice loss that can mine the information in background areas and modify network architecture to improve performance. The improved dice loss called weighted soft dice loss (WSDice loss). Our loss function gives a small weight to the background area of the label, so the background area will be added to the calculation when calculating dice loss. It can also soft the hard label in the lesion area to increase the robustness of the model to noise label. What's more, we propose to cascade Focal loss and WSDice loss. Focal Loss is a Distribution-based loss function, WSDice Loss is a Region-based loss function, the optimization directions of them are different. The cascaded loss function can make full use of the advantages of both and greatly improve model performance. In addition, we add a simple but effective channel attention module to the decode module of U-net. We experimented on the ChestX-ray8 datasets. Compared with Dice loss, WSDice loss improves the dice coefficient by 1.59%, cascaded loss function can improve dice coefficient by 7.81%. The improved in model architecture can increase the dice coefficient by 1.36%.

작업장 소음의 발생원대책 강화를 위한 저소음기계 인증제도 개선방안

In advanced countries, various policies are being enforced to reduce equipment noise level at it’s source. This paper analyses the problems of domestic noise policies related to noisy equipment, and presents a plan for improving the implementation of expansion policy of low noise equipment based on this analysis. This paper proposes that the equipment satisfying the compulsory rule be able to use the environmental certification mark, the standards of which are defined in recommended rules, to improve the linkage between the two policy measures. This paper also suggests expanding the range of equipment to be supervised and using new noise label that has grades and shows detailed noise level of the equipment to improve environmental certification system. In addition, supplementing the construction permission system and supporting international certification system are offered to expand the use of low noise equipment.

LCC: Towards efficient label completion and correction for supervised medical image learning in smart diagnosis

Smart technology brings convenience to our lives, and smart medical diagnosis plays a particularly important role in smart cities and the smart world. In smart diagnosis, high-quality image labels are essential for supervised medical image learning. The correctness of labels has a substantial influence on smart disease diagnosis. However, the labelling of disease images by professional doctors is a time-consuming and labour-intensive project; therefore, obtaining correct and high-quality labels is a difficult task. This paper proposes a deep convolutional neural network (CNN)-based label completion and correction method (LCC) for supervised medical image learning in smart diagnosis. We use a small number of labelled images in a dataset to train a model, and design a strategy to complement most of the unlabelled data and correct noise label data, and then use the results of completion and correction to continually modify the model to achieve better results. We take the diagnosis of Seborrheic Keratosis (SK) and Flat Wart (FW) as examples; the experimental results show that this strategy can use limited labelled data to complete the labels for most of the unlabelled data and correct some noise labels in the dataset, which improves the accuracy rate of the model to identify diseases.

A New Structure and Criterion for Dataset of Image Classification

What is the correct label for an image in Machine Learning? I investigate the process of label assign-ment of MNIST, CIFAR-10 and Fashion-MNIST and the results show that even in these simple da-tasets classes are not always as easy to distinguish as lived and died, and sometimes even impossible. The reasons include: lack of mutually exclusive definition for each class, and the definition lack of features to distinguish the classes, and the instances are incoherent to the definition and overlap, and the label can’t present uncertainty. In current research, these images and labels are treated as noise label. To give a clearer definition for these cases, I pro-pose a new structure of dataset including: mutually exclusive definition of classes and the label of la-bels, which classify labels into coherent, wrong and uncertain (including multi-objects, mid-object, un-known and unclear), and instances of different la-bels, and criterion to assess algorithms performance. Such structure could also apply to more complex dataset such as ChestXray14 in medical field to make learning and prediction more accurate and meaningful.