Search Results (28)

Human action recognition (HAR) technology based on radar signals has garnered significant attention from both industry and academia due to its exceptional privacy-preserving capabilities, noncontact sensing characteristics, and insensitivity to lighting conditions. However, the scarcity of accurately labeled human radar data poses a significant challenge in meeting the demand for large-scale training datasets required by deep model-based HAR technology, thus substantially impeding technological advancements in this field. To address this issue, a semi-supervised learning algorithm, MF-Match, is proposed in this paper. This algorithm computes pseudo-labels for larger-scale unsupervised radar data, enabling the model to extract embedded human behavioral information and enhance the accuracy of HAR algorithms. Furthermore, the method incorporates contrastive learning principles to improve the quality of model-generated pseudo-labels and mitigate the impact of mislabeled pseudo-labels on recognition performance. Experimental results demonstrate that this method achieves action recognition accuracies of 86.69% and 91.48% on two widely used radar spectrum datasets, respectively, utilizing only 10% labeled data, thereby validating the effectiveness of the proposed approach. Full article

Support vector data description (SVDD) is widely regarded as an effective technique for addressing anomaly detection problems. However, its performance can significantly deteriorate when the training data are affected by outliers or mislabeled observations. This study introduces a universal truncated loss function framework into the SVDD model to enhance its robustness and employs the fast alternating direction method of multipliers (ADMM) algorithm to solve various truncated loss functions. Moreover, the convergence of the fast ADMM algorithm is analyzed theoretically. Within this framework, we developed the truncated generalized ramp, truncated binary cross entropy, and truncated linear exponential loss functions for SVDD. We conducted extensive experiments on synthetic and real-world datasets to validate the effectiveness of these three SVDD models in handling data with different noise levels, demonstrating their superior robustness and generalization capabilities compared to other SVDD models. Full article

Distantly supervised relation extraction (DSRE), first used to address the limitations of manually annotated data via automatically annotating the data with triplet facts, is prone to issues such as mislabeled annotations due to the interference of noisy annotations. To address the interference of noisy annotations, we leveraged a novel knowledge distillation (KD) method which was different from the conventional models on DSRE. More specifically, we proposed a model-agnostic KD method, Multi-Level Knowledge Distillation with Adaptive Temperature (MKDAT), which mainly involves two modules: Adaptive Temperature Regulation (ATR) and Multi-Level Knowledge Distilling (MKD). ATR allocates adaptive entropy-based distillation temperatures to different training instances for providing a moderate softening supervision to the student, in which label hardening is possible for instances with great entropy. MKD combines the bag-level and instance-level knowledge of the teacher as supervisions of the student, and trains the teacher and student at the bag and instance levels, respectively, which aims at mitigating the effects of noisy annotation and improving the sentence-level prediction performance. In addition, we implemented three MKDAT models based on the CNN, PCNN, and ATT-BiLSTM neural networks, respectively, and the experimental results show that our distillation models outperform the baseline models on bag-level and instance-level evaluations. Full article

With the significant increase in cyber-attacks and attempts to gain unauthorised access to systems and information, Network Intrusion-Detection Systems (NIDSs) have become essential detection tools. Anomaly-based systems use machine learning techniques to distinguish between normal and anomalous traffic. They do this by using training datasets that have been previously gathered and labelled, allowing them to learn to detect anomalies in future data. However, such datasets can be accidentally or deliberately contaminated, compromising the performance of NIDS. This has been the case of the UGR’16 dataset, in which, during the labelling process, botnet-type attacks were not identified in the subset intended for training. This paper addresses the mislabelling problem of real network traffic datasets by introducing a novel methodology that (i) allows analysing the quality of a network traffic dataset by identifying possible hidden or unidentified anomalies and (ii) selects the ideal subset of data to optimise the performance of the anomaly detection model even in the presence of hidden attacks erroneously labelled as normal network traffic. To this end, a two-step process that makes incremental use of the training dataset is proposed. Experiments conducted on the contaminated UGR’16 dataset in conjunction with the state-of-the-art NIDS, Kitsune, conclude with the feasibility of the approach to reveal observations of hidden botnet-based attacks on this dataset. Full article

With the increasing volume of collected Earth observation (EO) data, artificial intelligence (AI) methods have become state-of-the-art in processing and analyzing them. However, there is still a lack of high-quality, large-scale EO datasets for training robust networks. This paper presents AgriSen-COG, a large-scale benchmark dataset for crop type mapping based on Sentinel-2 data. AgriSen-COG deals with the challenges of remote sensing (RS) datasets. First, it includes data from five different European countries (Austria, Belgium, Spain, Denmark, and the Netherlands), targeting the problem of domain adaptation. Second, it is multitemporal and multiyear (2019–2020), therefore enabling analysis based on the growth of crops in time and yearly variability. Third, AgriSen-COG includes an anomaly detection preprocessing step, which reduces the amount of mislabeled information. AgriSen-COG comprises 6,972,485 parcels, making it the most extensive available dataset for crop type mapping. It includes two types of data: pixel-level data and parcel aggregated information. By carrying this out, we target two computer vision (CV) problems: semantic segmentation and classification. To establish the validity of the proposed dataset, we conducted several experiments using state-of-the-art deep-learning models for temporal semantic segmentation with pixel-level data (U-Net and ConvStar networks) and time-series classification with parcel aggregated information (LSTM, Transformer, TempCNN networks). The most popular models (U-Net and LSTM) achieve the best performance in the Belgium region, with a weighted F1 score of 0.956 (U-Net) and 0.918 (LSTM).The proposed data are distributed as a cloud-optimized GeoTIFF (COG), together with a SpatioTemporal Asset Catalog (STAC), which makes AgriSen-COG a findable, accessible, interoperable, and reusable (FAIR) dataset. Full article

Arabica coffee, one of Indonesia’s economically important coffee commodities, is commonly subject to fraud due to mislabeling and adulteration. In many studies, spectroscopic techniques combined with chemometric methods have been massively employed in classification issues, such as principal component analysis (PCA) and discriminant analyses, compared to machine learning models. In this study, spectroscopy combined with PCA and a machine learning algorithm (artificial neural network, ANN) were developed to verify the authenticity of Arabica coffee collected from four geographical origins in Indonesia, including Temanggung, Toraja, Gayo, and Kintamani. Spectra from pure green coffee were collected from Vis–NIR and SWNIR spectrometers. Several preprocessing techniques were also applied to attain precise information from spectroscopic data. First, PCA compressed spectroscopic information and generated new variables called PCs scores, which would become inputs for the ANN model. The discrimination of Arabica coffee from different origins was conducted with a multilayer perceptron (MLP)-based ANN model. The accuracy attained ranged from 90% to 100% in the internal cross-validation, training, and testing sets. The error in the classification process did not exceed 10%. The generalization ability of the MLP combined with PCA was superior, suitable, and successful for verifying the origin of Arabica coffee. Full article

Tri-training expands the training set by adding pseudo-labels to unlabeled data, which effectively improves the generalization ability of the classifier, but it is easy to mislabel unlabeled data into training noise, which damages the learning efficiency of the classifier, and the explicit decision mechanism tends to make the training noise degrade the accuracy of the classification model in the prediction stage. This study proposes the Tri-training algorithm for adaptive nearest neighbor density editing and cross-entropy evaluation (TTADEC), which is used to reduce the training noise formed during the classifier iteration and to solve the problem of inaccurate prediction by explicit decision mechanism. First, the TTADEC algorithm uses the nearest neighbor editing to label high-confidence samples. Then, combined with the relative nearest neighbor to define the local density of samples to screen the pre-training samples, and then dynamically expand the training set by adaptive technique. Finally, the decision process uses cross-entropy to evaluate the completed base classifier of training and assign appropriate weights to it to construct a decision function. The effectiveness of the TTADEC algorithm is verified on the UCI dataset, and the experimental results show that compared with the standard Tri-training algorithm and its improvement algorithm, the TTADEC algorithm has better classification performance and can effectively deal with the semi-supervised classification problem where the training set is insufficient. Full article

The detection of cracks in concrete structures is crucial for the assessment of their structural integrity and safety. To this end, detection with deep neural convolutional networks has been extensively researched in recent years. Despite their success, these methods are limited in classifying concrete as cracked or non-cracked and disregard other characteristics, such as the severity of the cracks. Furthermore, the classification process can be affected by various sources of interference and noise in the images. In this paper, an integrated methodology for analysing concrete crack images is proposed using transfer and unsupervised learning. The method extracts image features using pre-trained networks and groups them based on similarity using hierarchical clustering. Three pre-trained networks are used for this purpose, with Inception v3 performing the best. The clustering results show the ability to divide images into different clusters based on image characteristics. In this way, various clusters are identified, such as clusters containing images of obstruction, background debris, edges, surface roughness, as well as cracked and uncracked concrete. In addition, dimensionality reduction is used to further separate and visualise the data, making it easier to analyse clustering results and identify misclassified images. This revealed several mislabelled images in the dataset used in this study. Additionally, a correlation was found between the principal components and the severity of cracks and surface imperfections. The results of this study demonstrate the potential of unsupervised learning for analysing concrete crack image data to distinguish between noisy images and the severity of cracks, which can provide valuable information for building more accurate predictive models. Full article

Deep neural networks (DNNs) require large amounts of labeled data for model training. However, label noise is a common problem in datasets due to the difficulty of classification and high cost of labeling processes. Introducing the concepts of curriculum learning and progressive learning, this paper presents a novel solution that is able to handle massive noisy labels and improve model generalization ability. It proposes a new network model training strategy that considers mislabeled samples directly in the network training process. The new learning curriculum is designed to measures the complexity of the data with their distribution density in a feature space. The sample data in each category are then divided into easy-to-classify (clean samples), relatively easy-to-classify, and hard-to-classify (noisy samples) subsets according to the smallest intra-class local density with each cluster. On this basis, DNNs are trained progressively in three stages, from easy to hard, i.e., from clean to noisy samples. The experimental results demonstrate that the accuracy of image classification can be improved through data augmentation, and the classification accuracy of the proposed method is clearly higher than that of standard Inception_v2 for the NEU dataset after data augmentation, when the proportion of noisy labels in the training set does not exceed 60%. With 50% noisy labels in the training set, the classification accuracy of the proposed method outperformed recent state-of-the-art label noise learning methods, CleanNet and MentorNet. The proposed method also performed well in practical applications, where the number of noisy labels was uncertain and unevenly distributed. In this case, the proposed method not only can alleviate the adverse effects of noisy labels, but it can also improve the generalization ability of standard deep networks and their overall capability. Full article

A small portion of mislabeled data can easily limit the performance of deep neural networks (DNNs) due to their high capacity for memorizing random labels. Thus, robust learning from noisy labels has become a key challenge for deep learning due to inadequate datasets with high-quality annotations. Most existing methods involve training models on clean sets by dividing clean samples from noisy ones, resulting in large amounts of mislabeled data being unused. To address this problem, we propose categorizing training samples into five fine-grained clusters based on the difficulty experienced by DNN models when learning them and label correctness. A novel fine-grained confidence modeling (FGCM) framework is proposed to cluster samples into these five categories; with each cluster, FGCM decides whether to accept the cluster data as they are, accept them with label correction, or accept them as unlabeled data. By applying different strategies to the fine-grained clusters, FGCM can better exploit training data than previous methods. Extensive experiments on widely used benchmarks CIFAR-10, CIFAR-100, clothing1M, and WebVision with different ratios and types of label noise demonstrate the superiority of our FGCM. Full article

Real-world classification data usually contain noise, which can affect the accuracy of the models and their complexity. In this context, an interesting approach to reduce the effects of noise is building ensembles of classifiers, which traditionally have been credited with the ability to tackle difficult problems. Among the alternatives to build ensembles with noisy data, bagging has shown some potential in the specialized literature. However, existing works in this field are limited and only focus on the study of noise based on a random mislabeling, which is unlikely to occur in real-world applications. Recent research shows that other types of noise, such as that occurring at class boundaries, are more common and challenging for classification algorithms. This paper delves into the analysis of the usage of bagging techniques in these complex problems, in which noise affects the decision boundaries among classes. In order to investigate whether bagging is able to reduce the impact of borderline noise, an experimental study is carried out considering a large number of datasets with different noise levels, and several noise models and classification algorithms. The results obtained reflect that bagging obtains a better accuracy and robustness than the individual models with this complex type of noise. The highest improvements in average accuracy are around 2–4% and are generally found at medium-high noise levels (from 15–20% onwards). The partial consideration of noisy samples when creating the subsamples from the original training set in bagging can make it so that only some parts of the decision boundaries among classes are impaired when building each model, reducing the impact of noise in the global system. Full article

In recent years, supervised learning-based methods have achieved excellent performance for hyperspectral image (HSI) classification. However, the collection of training samples with labels is not only costly but also time-consuming. This fact usually causes the existence of weak supervision, including incorrect supervision where mislabeled samples exist and incomplete supervision where unlabeled samples exist. Focusing on the inaccurate supervision and incomplete supervision, the weakly supervised classification of HSI is investigated in this paper. For inaccurate supervision, complementary learning (CL) is firstly introduced for HSI classification. Then, a new method, which is based on selective CL and convolutional neural network (SeCL-CNN), is proposed for classification with noisy labels. For incomplete supervision, a data augmentation-based method, which combines mixup and Pseudo-Label (Mix-PL) is proposed. And then, a classification method, which combines Mix-PL and CL (Mix-PL-CL), is designed aiming at better semi-supervised classification capacity of HSI. The proposed weakly supervised methods are evaluated on three widely-used hyperspectral datasets (i.e., Indian Pines, Houston, and Salinas datasets). The obtained results reveal that the proposed methods provide competitive results compared to the state-of-the-art methods. For inaccurate supervision, the proposed SeCL-CNN has outperformed the state-of-the-art method (i.e., SSDP-CNN) by 0.92%, 1.84%, and 1.75% in terms of OA on the three datasets, when the noise ratio is 30%. And for incomplete supervision, the proposed Mix-PL-CL has outperformed the state-of-the-art method (i.e., AROC-DP) by 1.03%, 0.70%, and 0.82% in terms of OA on the three datasets, with 25 training samples per class. Full article

Metal oxide (MOX) sensors offer a low-cost solution to detect volatile organic compound (VOC) mixtures. However, their operation involves time-consuming heating cycles, leading to a slower data collection and data classification process. This work introduces a few-shot learning approach that promotes rapid classification. In this approach, a model trained on several base classes is fine-tuned to recognize a novel class using a small number (n = 5, 25, 50 and 75) of randomly selected novel class measurements/shots. The used dataset comprises MOX sensor measurements of four different juices (apple, orange, currant and multivitamin) and air, collected over 10-minute phases using a pulse heater signal. While high average accuracy of 82.46 is obtained for five-class classification using 75 shots, the model’s performance depends on the juice type. One-shot validation showed that not all measurements within a phase are representative, necessitating careful shot selection to achieve high classification accuracy. Error analysis revealed contamination of some measurements by the previously measured juice, a characteristic of MOX sensor data that is often overlooked and equivalent to mislabeling. Three strategies are adopted to overcome this: (E1) and (E2) fine-tuning after dropping initial/final measurements and the first half of each phase, respectively, (E3) pretraining with data from the second half of each phase. Results show that each of the strategies performs best for a specific number of shots. E3 results in the highest performance for five-shot learning (accuracy 63.69), whereas E2 yields the best results for 25-/50-shot learning (accuracies 79/87.1) and E1 predicts best for 75-shot learning (accuracy 88.6). Error analysis also showed that, for all strategies, more than 50% of air misclassifications resulted from contamination, but E1 was affected the least. This work demonstrates how strongly data quality can affect prediction performance, especially for few-shot classification methods, and that a data-centric approach can improve the results. Full article

In the era of Big Data, the digitization of texts and the advancements in Artificial Intelligence (AI) and Natural Language Processing (NLP) are enabling the automatic analysis of literary works, allowing us to delve into the structure of artifacts and to compare, explore, manage and preserve the richness of our written heritage. This paper proposes a deep-learning-based approach to discovering semantic relationships in literary texts (19th century Greek Literature) facilitating the analysis, organization and management of collections through the automation of metadata extraction. Moreover, we provide a new annotated dataset used to train our model. Our proposed model, REDSandT_Lit, recognizes six distinct relationships, extracting the richest set of relations up to now from literary texts. It efficiently captures the semantic characteristics of the investigating time-period by finetuning the state-of-the-art transformer-based Language Model (LM) for Modern Greek in our corpora. Extensive experiments and comparisons with existing models on our dataset reveal that REDSandT_Lit has superior performance (90% accuracy), manages to capture infrequent relations (100%F in long-tail relations) and can also correct mislabelled sentences. Our results suggest that our approach efficiently handles the peculiarities of literary texts, and it is a promising tool for managing and preserving cultural information in various settings. Full article

A magnetic resonance imaging (MRI) exam typically consists of the acquisition of multiple MR pulse sequences, which are required for a reliable diagnosis. With the rise of generative deep learning models, approaches for the synthesis of MR images are developed to either synthesize additional MR contrasts, generate synthetic data, or augment existing data for AI training. While current generative approaches allow only the synthesis of specific sets of MR contrasts, we developed a method to generate synthetic MR images with adjustable image contrast. Therefore, we trained a generative adversarial network (GAN) with a separate auxiliary classifier (AC) network to generate synthetic MR knee images conditioned on various acquisition parameters (repetition time, echo time, and image orientation). The AC determined the repetition time with a mean absolute error (MAE) of 239.6 ms, the echo time with an MAE of 1.6 ms, and the image orientation with an accuracy of 100%. Therefore, it can properly condition the generator network during training. Moreover, in a visual Turing test, two experts mislabeled 40.5% of real and synthetic MR images, demonstrating that the image quality of the generated synthetic and real MR images is comparable. This work can support radiologists and technologists during the parameterization of MR sequences by previewing the yielded MR contrast, can serve as a valuable tool for radiology training, and can be used for customized data generation to support AI training. Full article