Training Instances Research Articles

Earthquake magnitude prediction in Indonesia using a supervised method based on cloud radon data

In the challenging realm of earthquake prediction, the reliability of forecasting systems has remained a persistent obstacle. This study focuses on earthquake magnitude prediction in Indonesia, leveraging supervised machine learning techniques and cloud radon data. We present an analysis of the tele-monitoring system, data collection methods, and the application of regression-based machine learning algorithms. Utilizing a comprehensive dataset spanning 30 training instances and 105 test instances, the study evaluates multiple metrics to ascertain the efficacy of the prediction models. Our findings reveal that the linear regression approach yields the best earthquake magnitude prediction method, with the lowest values across multiple evaluation metrics: standard deviation 0.40, mean absolute error (MAE) 0.30, mean absolute percentage error (MAPE) 6%, root mean square error (RMSE) 0.52, mean squared error (MSE) 0.28, symmetric mean absolute percentage error (SMAPE) 0.06, and conformal normalized mean absolute percentage error (cnSMAPE) 0.97. Additionally, we discuss the implications of the research results and the potential applications in enhancing existing earthquake prediction methodologies.

Upskilling consumers for the digital health era: a content analysis of resources for consumer representative training

Upskilling consumers for the digital health era: a content analysis of resources for consumer representative training

An Unbiased Risk Estimator for Partial Label Learning with Augmented Classes

Partial Label Learning (PLL) is a typical weakly supervised learning task, which assumes each training instance is annotated with a set of candidate labels containing the ground-truth label. Recent PLL methods adopt identification-based disambiguation to alleviate the influence of false positive labels and achieve promising performance. However, they require all classes in the test set to have appeared in the training set, ignoring the fact that new classes will keep emerging in real applications. To address this issue, in this paper, we focus on the problem of Partial Label Learning with Augmented Class (PLLAC), where one or more augmented classes are not visible in the training stage but appear in the inference stage. Specifically, we propose an unbiased risk estimator with theoretical guarantees for PLLAC, which estimates the distribution of augmented classes by differentiating the distribution of known classes from unlabeled data and can be equipped with arbitrary PLL loss functions. Besides, we provide a theoretical analysis of the estimation error bound of the estimator, which guarantees the convergence of the empirical risk minimizer to the true risk minimizer as the number of training data tends to infinity. Furthermore, we add a risk-penalty regularization term in the optimization objective to alleviate the influence of the over-fitting issue caused by negative empirical risk. Extensive experiments on benchmark, UCI and real-world datasets demonstrate the effectiveness of the proposed approach.

On the use of Change History Data to Enhance Class Change-Proneness Prediction Models

As software evolves, new artifacts are created, modified, or removed. One of these main artifacts generated in the development of object-oriented software is the class. Classes have a very dynamic life cycle that can result in additional costs to the project. One way to mitigate this is to detect, in the early stages of the development, classes that are prone to change. Some approaches in the literature adopt Machine Learning (ML) algorithms to predict the change-proneness of a class. However, most of these approaches do not consider the temporal dependency between training instances, i.e., they consider that the instances are independent. To overcome such a limitation, this study presents an approach for predicting change-proneness based on the class change history. The approach adopts the sliding window method and is evaluated to obtain six kinds of models, which are derived by using, as predictors, different sets of metrics: structural, evolutionary, and smell-based. The evaluation uses five systems, four ML algorithms, and also explores some resample techniques to deal with imbalanced data. Regardless of the kind of model analyzed and the algorithm used, our approach overcomes the traditional one in 378 (~80) cases, out of 420, considering all systems, kinds of models, indicators, and algorithms. Moreover, the results show that our approach presents the best performance when the set of evolutionary metrics is used as predictors. There is no improvement when smell-based metrics are added. The Random Forest algorithm with the resampling technique ADA reaches the best performance among the ML algorithms evaluated.

Machine learning-based synthesis of diagnostic algorithms for electromechanical actuators to improve the aerospace flight safety

The relevance of research aimed at developing diagnostic technologies for electromechanical actuators is due to the need to improve flight safety in conditions of increasing intensity of highly electrified aircraft and spacecraft operations. The paper discusses one of the promising approaches to electromechanical actuator health management, which involves the use of machine learning methods to synthesize health monitoring algorithms. Machine learning methods make it possible to build classification models based on empirical data, which are used to generate recommendations for making operational decisions. Empirical data, which is a source of valuable experience and the basis of a training sample necessary for formalizing patterns in classification models, can be formed as a result of life tests, mathematical modeling, and actuator operation. In order to improve the safety of space flights, the article focuses on the integration of electromechanical actuator mathematical model methods, optimal space filling, and machine learning. Optimal space filling methods are used to reduce the computational costs associated with representative training sampling. Examples of developing classification models are given to determine failures associated with changes in gear (backlash, Coulomb friction and viscous friction) which is the most critical actuator link. As a result of computational studies, the main advantages of the proposed approach to the synthesis of electromechanical actuator health assessment algorithms are shown.

A novel approach to estimate land surface temperature from landsat top-of-atmosphere reflective and emissive data using transfer-learning neural network

Land Surface Temperature (LST) is a crucial parameter in studies of urban heat islands, climate change, evapotranspiration, hydrological cycles, and vegetation monitoring. However, conventional satellite-based approaches for LST retrieval often require additional data like land surface emissivity (LSE). Meanwhile, traditional machine learning (ML) techniques face challenges in acquiring representative training data and leveraging data from varied sources effectively. To address these issues, we introduce a novel transfer-learning (TL) neural network approach for LST retrieval using top-of-atmosphere (TOA) reflective and emissive data from Landsat. This method not only improves LST retrieval by integrating various data types but also demonstrates the potential of shortwave data in surrogating LSE information, thereby reducing dependence on explicit LSE data. Our TL approach utilized extensive simulations from the radiative transfer model (RTM) and measurements from the real world. The simulations are comprehensive, covering a wide range of atmospheric and surface scenarios, and the inclusion of real-world data mitigates the discrepancy between simulations and actual observations. When applied to a decade of Landsat-8 observations and ground measurements from 241 stations across diverse regions, our TL method significantly outperforms ML, single-channel (SC), and split-window (SW) algorithms in terms of root mean square error (RMSE), with improvements of 0.46 K, 0.84 K, and 0.57 K, respectively. This superiority underscores the advantage of integrating simulated and observed data, as well as the benefit of utilizing both reflective and emissive data without relying on uncertain LSE inputs. Our findings present a promising new TL framework for estimating LST directly from TOA data, offering a robust approach that we have made publicly available through Google Earth Engine (GEE) for broader use. The LST data retrieved by our proposed method can provide valuable insights for environmental research.

Few-Shot Symbol Detection in Engineering Drawings

ABSTRACT Recently, there has been significant interest in digitizing engineering drawings due to their complexity and practical benefits. Symbol digitization, a critical aspect in this field, is challenging as utilizing Deep Learning-based methods to recognize symbols of interest requires a large number of training instances for each class of symbols. Acquiring and annotating sufficient diagrams is difficult due to concerns about confidentiality and availability. The conventional manual annotation process is time-consuming, costly, and prone to human error. Additionally, obtaining an adequate number of samples for rare classes proves to be exceptionally challenging. This paper introduces a few-shot framework to address these challenges. Several experiments with fewer than ten, and sometimes just one, training instance per class using complex engineering drawings from industry sources were carried out. The results suggest that our method not only significantly improves symbol detection performance compared to other state-of-the-art methods but also decreases the necessary number of training instances.

Towards sharper excess risk bounds for differentially private pairwise learning

Pairwise learning is a vital part of machine learning. It depends on pairs of training instances, and is naturally fit for modeling relationships between samples. However, as a data driven paradigm, it faces huge privacy issues. Differential privacy (DP) is a useful tool to protect the privacy of machine learning, but corresponding excess population risk bounds are loose in existing DP pairwise learning analysis. In this paper, we propose a gradient perturbation algorithm for pairwise learning to get better risk bounds under Polyak–Łojasiewicz condition, including both convex and non-convex cases. Specifically, for the theoretical risk bound in expectation, previous best results are of rates O(pn2ϵ2+1n) and O(pnϵ+1n) under strongly convex condition and convex conditions, respectively. In this paper, we use the on-average stability and achieve an O(min{pn1.5ϵ+p1.5n2.5ϵ3,pn2ϵ2+1n}) bound, significantly improving previous bounds. For the high probability risk bound, previous best results are analyzed by the uniform stability, and O(βnU+pnϵ) excess population risk bounds are achieved under strongly convex or convex conditions, where βnU is the traditional pairwise uniform stability parameter, it is large since it considers the worst case of the loss sensitivity. In this paper, we propose the pairwise locally elastic stability and improve the high probability bound to O(βEn+pnϵ), in which the pairwise locally elastic stability parameter βE≪βnU because it considers the average sensitivity of the pairwise loss function.

Residual [formula omitted]-Nearest Neighbors Label Distribution Learning

Label Distribution Learning (LDL) is a novel learning paradigm that assigns label distribution to each instance. It aims to learn the label distribution of training instances and predict unknown ones. Algorithm Adaptation (AA)-kNN is one of the most representative LDL baselines that adapts the kNN algorithm to LDL. Its generalization risk has been proven to approach zero if given infinite training data. Despite its theoretical advantage, AA-kNN generally performs poorly because real-world LDL problems only have finite or small training data. In this paper, we improve AA-kNN and propose a novel method called Residual k-Nearest Neighbors Label Distribution Learning (RkNN-LDL). First, RkNN-LDL introduces residual label distribution learning. Second, RkNN-LDL exploits the neighborhood structure of label distribution. In theoretical analysis, we prove that RkNN-LDL has a tighter generalization bound than AA-kNN. Besides, extensive experiments validate that RkNN-LDL beats several state-of-the-art LDL methods and statistically outperforms AA-kNN.

Spatial attention U-Net model with Harris hawks optimization for retinal blood vessel and optic disc segmentation in fundus images.

The state of the human eye's blood vessels is a crucial aspect in the diagnosis of ophthalmological illnesses. For many computer-aided diagnostic systems, precise retinal vessel segmentation is an essential job. However, it remains a difficult task due to the intricate vascular system of the eye. Although many different vascular segmentation techniques have already been presented, additional study is still required to address the problem of inadequate segmentation of thin and tiny vessels. In this work, we introduce the Spatial Attention U-Net (SAU-Net) modelwith harris hawks' optimization (HHO), a lightweight network that can be applied as a data augmentation technique to improve the efficiency of the existingannotated samples without the need of thousands of training instances for Retinal Blood Vessel and Optic Disc Segmentation. The SAU-Net-HHO implementation uses a spatially inferred attention map multiplied by the input feature map for adaptive feature enhancement. U-Net convolutional blocks have been replaced with structured dropout blocks in the proposed network to prevent overfitting. Data from both vascular extraction (DRIVE) and structured analysis of the retina (STARE) are used to evaluate SAU-Net-HHO performance. The results show that the proposed SAU-Net-HHO performs well on both datasets. Analysing the obtained results, an average of 98.5% accuracy and Specificity 96.7% was achieved for DRIVE dataset and 97.8% accuracy and specificity 94.5% for STARE dataset. The proposed method yields numerical results with average values that are on par with those of state-of-the-art methods. Visual inspection has revealed that the suggested method can segment thin and tiny vessels with greater accuracy than previous methods. It also demonstrates its potential for real-life clinical application.

2.5D deep learning based on multi-parameter MRI to differentiate primary lung cancer pathological subtypes in patients with brain metastases

BackgroundBrain metastases (BMs) represents a severe neurological complication stemming from cancers originating from various sources. It is a highly challenging clinical task to accurately distinguish the pathological subtypes of brain metastatic tumors from lung cancer (LC).The utility of 2.5-dimensional (2.5D) deep learning (DL) in distinguishing pathological subtypes of LC with BMs is yet to be determined. MethodsA total of 250 patients were included in this retrospective study, divided in a 7:3 ratio into training set (N=175) and testing set (N=75). We devised a method to assemble a series of two-dimensional (2D) images by extracting adjacent slices from a central slice in both superior-inferior and anterior-posterior directions to form a 2.5D dataset. Multi-Instance learning (MIL) is a weakly supervised learning method that organizes training instances into “bags” and provides labels for entire bags, with the purpose of learning a classifier based on the labeled positive and negative bags to predict the corresponding class for an unknown bag. Therefore, we employed MIL to construct a comprehensive 2.5D feature set. Then we used the single-slice as input for constructing the 2D model. DL features were extracted from these slices using the pre-trained ResNet101. All feature sets were inputted into the support vector machine (SVM) for evaluation. The diagnostic performance of the classification models were evaluated using five-fold cross-validation, with accuracy and area under the curve (AUC) metrics calculated for analysis. ResultsThe optimal performance was obtained using the 2.5D DL model, which achieved the micro-AUC of 0.868 (95% confidence interval [CI], 0.817–0.919) and accuracy of 0.836 in the test cohort. The 2D model achieved the micro-AUC of 0.836 (95 % CI, 0.778–0.894) and accuracy of 0.827 in the test cohort. ConclusionsThe proposed 2.5D DL model is feasible and effective in identifying pathological subtypes of BMs from lung cancer.

Addressing type Ia supernova color variability with a linear spectral template

Type Ia Supernovae (SNeIa) provided the first evidence of an accelerated expansion of the universe and remain a valuable probe to cosmology. They are deemed standardizable candles due to the observed correlations between their luminosity and photometric quantities. This characteristic can be exploited to estimate cosmological distances after accounting for the observed variations. There is however a remaining dispersion unaccounted for in the current state-of-the-art standardization methods. In an attempt to explore this issue, we propose a simple linear 3-component rest-frame flux description for a light-curve fitter. Since SNIa intrinsic color index variations are expected to be time-dependent, our description builds upon the mathematical expression of the well-known Spectral Adaptive Light Curve Template 2 (SALT2) for rest-frame flux, whilst we drop the exponential factor and add an extra model component with time and wavelength dependencies. The model components are obtained by performing either Principal Component Analysis (PCA) or Factor Analysis (FA) onto a representative training set. The constraining power of the model dubbed Pure Expansion Template for Supernovae (PETS) is evaluated and we found compatible results with SALT2 for Ωm0 and ΩΛ0 within 68% uncertainty between the two models, with PETS’ fit parameters exhibiting non-negligible linear correlations with SALT2’ parameters. For both PCA and FA model versions we verified that the first component mainly describes color index variations, proving it is a dominant effect on SNIa spectra. The model nuisance parameter which multiplies the color index variation-like fit parameter shows evolution with redshift in an initial binned cosmology analysis. This behavior can be due to selection effects and should be further investigated with higher redshift SNeIa samples. Overall, our model shows promise, as there are still a few aspects to be refined; however, it still falls short in reducing the unaccounted dispersion.

Applying an ecological dynamics framework to mixed martial arts training

ABSTRACT In this paper, we outline an ecological approach to training in mixed martial arts (MMA) to support coaches in helping fighters coordinate skilled movement solutions to the ever-changing problems they encounter in the cage. Orienting one’s perspective on the fighter-cage system may require coaches to re-align their role towards being designers of alive and representative training sessions that are filled with opportunities for (inter)actions, where constraints are purposefully manipulated to facilitate the emergence of skilled behaviours for mixed martial artists of all levels. We conclude by offering three case examples ranging from amateurs to professional champions to highlight the significance of adopting an ecological dynamics framework in MMA.

Organizational and pedagogical conditions for the formation of the readiness of future physical training and sports instructors for professional activity

Today, physical training and sports instructors play a key role in the training of military personnel of various categories and age groups, because in the process of professional activity they ensure the development and improvement of: physical qualities and military-applied skills; psychophysical readiness to act in extreme conditions, etc. The theoretical research was carried out in accordance with the plan of the initiative scientific project "SPT Instructor, 2022-2025" with the participation of leading specialists in physical training (physical education of various population groups) of representatives of specialized departments of higher military educational institutions (institutions of higher education) of Ukraine (in accordance with agreements on interdepartmental cooperation). According to the results of the analysis of the scientific and methodological literature by the members of the scientific research group, it was established that empirical studies that comprehensively reveal the peculiarities of the organization of the training process of instructors in physical training and sports in institutions of the security and defense sector of Ukraine are rare, which emphasizes the relevance of the chosen direction of scientific research. Research methods: axiomatic, experience of organizing the system of physical and special physical training of military personnel, idealization, instrumental, historical and logical, departure from the concrete, formalization. As a result of the research, we have developed organizational and pedagogical conditions for the formation of the readiness of future instructors in physical training and sports for professional activity (using the example of servicemen of the National Guard of Ukraine who study at the courses of instructors in physical training and sports).

Saliency-based video summarization for face anti-spoofing

With the growing availability of databases for face presentation attack detection, researchers are increasingly focusing on video-based face anti-spoofing methods that involve hundreds to thousands of images for training the models. However, there is currently no clear consensus on the optimal number of frames in a video to improve face spoofing detection. Inspired by the visual saliency theory, we present a video summarization method for face anti-spoofing detection that aims to enhance the performance and efficiency of deep learning models by leveraging visual saliency. In particular, saliency information is extracted from the differences between the Laplacian and Wiener filter outputs of the source images, enabling the identification of the most visually salient regions within each frame. Subsequently, the source images are decomposed into base and detail images, enhancing the representation of the most important information. Weighting maps are then computed based on the saliency information, indicating the importance of each pixel in the image. By linearly combining the base and detail images using the weighting maps, the method fuses the source images to create a single representative image that summarizes the entire video. The key contribution of the proposed method lies in demonstrating how visual saliency can be used as a data-centric approach to improve the performance and efficiency for face presentation attack detection. By focusing on the most salient images or regions within the images, a more representative and diverse training set can be created, potentially leading to more effective models. To validate the method’s effectiveness, a simple CNN–RNN deep learning architecture was used, and the experimental results showcased state-of-the-art performance on four challenging face anti-spoofing datasets.

A refined reweighing technique for nondiscriminatory classification.

Discrimination-aware classification methods remedy socioeconomic disparities exacerbated by machine learning systems. In this paper, we propose a novel data pre-processing technique that assigns weights to training instances in order to reduce discrimination without changing any of the inputs or labels. While the existing reweighing approach only looks into sensitive attributes, we refine the weights by utilizing both sensitive and insensitive ones. We formulate our weight assignment as a linear programming problem. The weights can be directly used in any classification model into which they are incorporated. We demonstrate three advantages of our approach on synthetic and benchmark datasets. First, discrimination reduction comes at a small cost in accuracy. Second, our method is more scalable than most other pre-processing methods. Third, the trade-off between fairness and accuracy can be explicitly monitored by model users. Code is available at https://github.com/frnliang/refined_reweighing.

Automatic similarity feature selection for ontology matching with semantic sampling

Ontology provides a formal representation of knowledge as a set of concepts within a domain and the relationships between them. Ontology Matching (OM) identifies similar entities across different ontologies. Similarity Features (SFs) characterize the similarity between two entities from different perspectives, which serves as the foundation of OM method. However, noisy and redundant SFs can obstruct the relevance of useful ones, reducing the quality of matching results. To offer a practical and efficient SF selection, this work proposes a new automatic SF selection framework for OM, which consists of two new components. First, a semantic sampling method is proposed to automatically construct a balanced and representative training dataset, without the need for expert intervention. This method first generates a basic sample set that considers both the representativeness of the concepts and the heterogeneous characteristics of the two ontologies, and then over-samples the minority samples based on their semantic context. Second, a wrapper-based SF selection method is designed, which includes an Adaptive Compact Genetic Algorithm (ACGA) and a Relaxed Naive Bayes (RNB) classifier. The ACGA is able to efficiently identify the SF subset, reducing the computational cost by adaptively maintaining a population probability model based on the diversity and confidence of the SFs. Moreover, the RNB can effectively evaluate the quality of selected SFs by modifying the feature independence assumption with recombination and weighting strategies. Experimental results on the ontology alignment evaluation initiative’s Benchmark demonstrate that our method can efficiently find accurate OM results, and significantly outperform the state-of-the-art matching techniques.

LLM-Assisted Data Augmentation for Chinese Dialogue-Level Dependency Parsing

Abstract Dialogue-level dependency parsing, despite its growing academic interest, often encounters underperformance issues due to resource shortages. A potential solution to this challenge is data augmentation. In recent years, large language models (LLMs) have demonstrated strong capabilities in generation, which can facilitate data augmentation greatly. In this study, we focus on Chinese dialogue-level dependency parsing, presenting three simple and effective strategies with LLM to augment the original training instances, namely word-level, syntax-level, and discourse-level augmentations, respectively. These strategies enable LLMs to either preserve or modify dependency structures, thereby assuring accuracy while increasing the diversity of instances at different levels. We conduct experiments on the benchmark dataset released by Jiang et al. (2023) to validate our approach. Results show that our method can greatly boost the parsing performance in various settings, particularly in dependencies among elementary discourse units. Lastly, we provide in-depth analysis to show the key points of our data augmentation strategies.

Partial label learning via identifying outlier features

Partial label learning (PLL) is a classification problem where each training instance is ambiguously annotated with a set of candidate labels, among which only one is the ground truth label. The topological structure in the feature space is frequently used by existing PLL approaches to clarify if a candidate label is the ground truth label for a training example. However, these techniques frequently experience cumulative errors brought on by the error-prone labeling confidence estimate throughout the topological structure because of the redundant and noisy features occurring naturally in the feature space. In this paper, we propose a novel approach through Identifying Outlier Features (PL-IOF) to address this challenge. The feature space is decomposed into class prototype features and outlier features, where the former captures the discriminative features of each class prototype, and the latter captures the outlier features in each training instance. A unified framework is specifically suggested to simultaneously optimize class prototypes and outlier features, as well as to estimate labeling confidences over incomplete label training instances. This framework ensures the high quality of the extracted class prototypes by recognizing the outliers. Experiments on both synthetic and real-world datasets demonstrate the out-performance of PL-IOF over the state-of-the-art.

Bidirectional brain image translation using transfer learning from generic pre-trained models

Brain imaging plays a crucial role in the diagnosis and treatment of various neurological disorders, providing valuable insights into the structure and function of the brain. Techniques such as magnetic resonance imaging (MRI) and computed tomography (CT) enable non-invasive visualization of the brain, aiding in the understanding of brain anatomy, abnormalities, and functional connectivity. However, cost and radiation dose may limit the acquisition of specific image modalities, so medical image synthesis can be used to generate required medical images without actual addition. CycleGAN and other GANs are valuable tools for generating synthetic images across various fields. In the medical domain, where obtaining labeled medical images is labor-intensive and expensive, addressing data scarcity is a major challenge. Recent studies propose using transfer learning to overcome this issue. This involves adapting pre-trained CycleGAN models, initially trained on non-medical data, to generate realistic medical images. In this work, transfer learning was applied to the task of MR-CT image translation and vice versa using 18 pre-trained non-medical models, and the models were fine-tuned to have the best result. The models’ performance was evaluated using four widely used image quality metrics: Peak-signal-to-noise-ratio, Structural Similarity Index, Universal Quality Index, and Visual Information Fidelity. Quantitative evaluation and qualitative perceptual analysis by radiologists demonstrate the potential of transfer learning in medical imaging and the effectiveness of the generic pre-trained model. The results provide compelling evidence of the model’s exceptional performance, which can be attributed to the high quality and similarity of the training images to actual human brain images. These results underscore the significance of carefully selecting appropriate and representative training images to optimize performance in brain image analysis tasks.