Expanded Training Data Research Articles

Plasmonic sensor using generative adversarial networks integration

Machine learning (ML) has emerged as a pivotal force in enhancing the capabilities of sensing technologies across a broad spectrum of applications, from environmental monitoring and biosensing to agriculture, industrial automation, and so on. This study explores integrating ML techniques with photonic crystal fiber (PCF)-based plasmonic sensing techniques to elevate sensor performance. The PCF has two open channels to augment mode coupling, effectively reducing the gap between the analyte channel and core. Moreover, a thin layer of gold within the open channels of the PCF initiates efficient plasmon generation. The results demonstrate a maximum wavelength sensitivity of 9000 nm/refractive index unit (RIU), which can detect a wide range of analyte refractive index (RI) values from 1.33 to 1.40. The sensor exhibits the maximum amplitude sensitivity of 490.41 RIU−1. It also boasts a resolution of 1.11 × 10−5 RIU and the maximum figure-of-merit (FOM) achieved is 138.04 RIU−1 at an analyte RI of 1.39. Furthermore, this research introduces a method utilizing generative adversarial networks (GAN) to expand training data for an artificial neural network (ANN) model. This approach substantially improves the prediction of confinement loss across various analytes and wavelengths in a unique geometric configuration. The sensor’s versatility makes it ideal for various applications, including chemical sensing and medical diagnostics.

Addressing Data Scarcity in Solar Energy Prediction with Machine Learning and Augmentation Techniques

The accurate prediction of global horizontal irradiance (GHI) is crucial for optimizing solar power generation systems, particularly in mountainous areas with complex topography and unique microclimates. These regions face significant challenges due to limited reliable data and the dynamic nature of local weather conditions, which complicate accurate GHI measurement. The scarcity of precise data impedes the development of reliable solar energy prediction models, impacting both economic and environmental outcomes. To address these data scarcity challenges in solar energy prediction, this paper focuses on various locations in Europe and Asia Minor, predominantly in mountainous regions. Advanced machine learning techniques, including random forest (RF) and extreme gradient boosting (XGBoost) regressors, are employed to effectively predict GHI. Additionally, optimizing training data distribution based on cloud opacity values and integrating synthetic data significantly enhance predictive accuracy, with R2 scores ranging from 0.91 to 0.97 across multiple locations. Furthermore, substantial reductions in root mean square error (RMSE), mean absolute error (MAE), and mean bias error (MBE) underscore the improved reliability of the predictions. Future research should refine synthetic data generation, optimize additional meteorological and environmental parameter integration, extend methodology to new regions, and test for predicting global tilted irradiance (GTI). The studies should expand training data considerations beyond cloud opacity, incorporating sky cover and sunshine duration to enhance prediction accuracy and reliability.

Brain-inspired semantic data augmentation for multi-style images.

Data augmentation is an effective technique for automatically expanding training data in deep learning. Brain-inspired methods are approaches that draw inspiration from the functionality and structure of the human brain and apply these mechanisms and principles to artificial intelligence and computer science. When there is a large style difference between training data and testing data, common data augmentation methods cannot effectively enhance the generalization performance of the deep model. To solve this problem, we improve modeling Domain Shifts with Uncertainty (DSU) and propose a new brain-inspired computer vision image data augmentation method which consists of two key components, namely, using Robust statistics and controlling the Coefficient of variance for DSU (RCDSU) and Feature Data Augmentation (FeatureDA). RCDSU calculates feature statistics (mean and standard deviation) with robust statistics to weaken the influence of outliers, making the statistics close to the real values and improving the robustness of deep learning models. By controlling the coefficient of variance, RCDSU makes the feature statistics shift with semantic preservation and increases shift range. FeatureDA controls the coefficient of variance similarly to generate the augmented features with semantics unchanged and increase the coverage of augmented features. RCDSU and FeatureDA are proposed to perform style transfer and content transfer in the feature space, and improve the generalization ability of the model at the style and content level respectively. On Photo, Art Painting, Cartoon, and Sketch (PACS) multi-style classification task, RCDSU plus FeatureDA achieves competitive accuracy. After adding Gaussian noise to PACS dataset, RCDSU plus FeatureDA shows strong robustness against outliers. FeatureDA achieves excellent results on CIFAR-100 image classification task. RCDSU plus FeatureDA can be applied as a novel brain-inspired semantic data augmentation method with implicit robot automation which is suitable for datasets with large style differences between training and testing data.

Gaussian process regression based inspection robot for predicting and locating pipeline anticorrosion coating defects

The direct current voltage gradient (DCVG) technology is adept at identifying defects and corrosion issues within the anti-corrosion layer of buried pipelines by measuring changes in voltage gradient above the ground. Its widespread adoption in the field of anti-corrosion layer defect detection for its high precision and accuracy. However, the current DCVG inspection process relies on experienced operators holding electrodes to walk along the pipeline, resulting in a huge workload. To address these challenges, this paper proposes an innovative method that combines Gaussian process regression (GPR) with an intelligent inspection robot for autonomous pipeline anti-corrosion coating defect detection. This method uses environmental data to directly predict the location of defects within a pipeline’s anti-corrosion coating. Through incremental learning, the GPR model is trained to be continuously updated based on new samples such as position coordinates and voltage measurements during autonomous inspections. In addition, the intelligent inspection robot operates collaboratively with crawler wheels and UR robotic arms, enhancing motion stability and flexibility in expanding training data sets. Experimental results confirm that the intelligent inspection robot driven by Gaussian process prediction can achieve accurate defect positioning within 25 iteration cycles, with a positioning accuracy within 0.12 m. This method enhances defect detection accuracy, alleviating operator burden and offering an efficient solution for buried pipeline maintenance.

Balancing Innovation and Regulation: A Comprehensive Analysis and Neural Network Approach to AI Copyright Challenges

This article explores emerging issues surrounding artificial intelligence (AI) and copyright through a two-pronged approach. First, it provides an extensive literature review analyzing government and industry strategies for addressing AI copyright concerns and evaluates their rationality. Second, it details experiments conducted using neural networks to examine relevant information and investigate image copyright challenges, assessing mainstream large language models’ efficacy in handling copyright matters.
 The literature review explores AI copyright perspectives of the United Kingdom, China, the European Union, and the United States. It finds that countries emphasize balanced regulation and innovation (UK), ethical content creation (China), regulating high-risk applications (EU), or principles like non-discrimination and privacy (US). However, comprehensive governance frameworks are needed to navigate AI’s ethical, social, and legal intricacies.
 The experimental portion trains a convolutional neural network on a dataset of 41 infringing and non-infringing image sets to identify copyright infringement. While achieving over 80% accuracy, enhancements through expanded training data, segmentation, and multi-domain detection could improve generalization. The paper concludes with an analysis advocating copyright adaptation for AI creations, measured protections for standalone AI works, and constructive policies from interdisciplinary dialogue.

Analysis of methods for image recognition and image synthesis for expanding the training sample

The article examines in detail an important aspect in the field of machine learning — image recognition and synthesis for effective training sample expansion. Methods for image recognition are considered, the method of convolutional neural networks is highlighted. One of the main problems when using these methods was also revealed, namely the insufficient training sample for training the network. In modern machine learning, where access to large amounts of real data is often limited, the problem of lack of training examples emerges as a key one. The importance of data expansion is undeniable in conditions of limited access to a large amount of real data. The article discusses and analyzes augmentation methods that allow you to increase the amount of data by introducing various transformations and modifications to the original set. Particular emphasis is placed on generative models such as the Variational Autoencoder (VAE). The authors consider in detail their ability to synthesize new, realistic images and their impact on improving the quality of model training. The ability to create realistic and context-sensitive images is examined, considering their pros and cons in the context of expanding training data. The importance of using such methods in scenarios with a small amount of available data is emphasized. The paper also examines the effect of balanced synthesis and recognition on the performance of models, taking into account the representativeness and diversity of the data. Examples and research results highlight the practicality of using the considered methods in various machine learning scenarios. The purpose of the article is not only to consider existing approaches, but also to point to the prospects and directions of further research in this important direction.

Particle Clustering in Turbulence: Prediction of Spatial and Statistical Properties with Deep Learning

We investigate the utility of deep learning for modeling the clustering of particles that are aerodynamically coupled to turbulent fluids. Using a Lagrangian particle module within the Athena++ hydrodynamics code, we simulate the dynamics of particles in the Epstein drag regime within a periodic domain of isotropic forced hydrodynamic turbulence. This setup is an idealized model relevant to the collisional growth of micron- to millimeter-sized dust particles in early-stage planet formation. The simulation data are used to train a U-Net deep-learning model to predict gridded three-dimensional representations of the particle density and velocity fields, given as input the corresponding fluid fields. The trained model qualitatively captures the filamentary structure of clustered particles in a highly nonlinear regime. We assess model fidelity by calculating metrics of the density field (the radial distribution function) and of the velocity field (the relative velocity and the relative radial velocity between particles). Although trained only on the spatial fields, the model predicts these statistical quantities with errors that are typically <10%. Our results suggest that, given appropriately expanded training data, deep learning could complement direct numerical simulations in predicting particle clustering within turbulent flows.

Expanding Training Data for Structure-Based Receptor-Ligand Binding Affinity Regression through Imputation of Missing Labels.

The success of machine learning is, in part, due to a large volume of data available to train models. However, the amount of training data for structure-based molecular property prediction remains limited. The previously described CrossDocked2020 data set expanded the available training data for binding pose classification in a molecular docking setting but did not address expanding the amount of receptor-ligand binding affinity data. We present experiments demonstrating that imputing binding affinity labels for complexes without experimentally determined binding affinities is a viable approach to expanding training data for structure-based models of receptor-ligand binding affinity. In particular, we demonstrate that utilizing imputed labels from a convolutional neural network trained only on the affinity data present in CrossDocked2020 results in a small improvement in the binding affinity regression performance, despite the additional sources of noise that such imputed labels add to the training data. The code, data splits, and imputation labels utilized in this paper are freely available at https://github.com/francoep/ImputationPaper.

Learning inter-annual flood loss risk models from historical flood insurance claims

Flooding is a natural hazard that causes substantial loss of lives and livelihoods worldwide. Developing predictive models for flood-induced financial losses is crucial for applications such as insurance underwriting. This research uses the National Flood Insurance Program (NFIP) dataset between 2000 and 2020 to evaluate the predictive skill of past data in predicting near-future flood loss risk. Our approach applies neural networks (Conditional Generative Adversarial Networks), decision trees (Extreme Gradient Boosting), and kernel-based regressors (Gaussian Processes) to estimate pointwise losses. It aggregates them over intervals using a bias-corrected Burr–Pareto distribution to predict risk. The regression models help identify the most informative predictors and highlight crucial factors influencing flood-related financial losses. Applying our approach to quantify the county-level coastal flood loss risk in eight US Southern states results in an R2=0.807, substantially outperforming related work using stage-damage curves. More detailed testing on 11 counties with significant claims in the NFIP dataset reveals that Extreme Gradient Boosting yields the most favorable results, and bias correction significantly improves the similarity between the predicted and reference claim amount distributions. Our experiments also show that, despite the already experienced climate change, the difference in future short-term risk predictions of flood-loss amounts between historical shifting or expanding training data windows is insignificant.

A Chinese Grammatical Error Correction Model Based On Grammatical Generalization And Parameter Sharing

Abstract Chinese grammatical error correction (CGEC) is a significant challenge in Chinese natural language processing. Deep-learning-based models tend to have tens of millions or even hundreds of millions of parameters since they model the target task as a sequence-to-sequence problem. This may require a vast quantity of annotated corpora for training and parameter tuning. However, there are currently few open-source annotated corpora for the CGEC task; the existing researches mainly concentrate on using data augmentation technology to alleviate the data-hungry problem. In this paper, rather than expanding training data, we propose a competitive CGEC model from a new insight for reducing model parameters. The model contains three main components: a sequence learning module, a grammatical generalization module and a parameter sharing module. Experimental results on two Chinese benchmarks demonstrate that the proposed model could achieve competitive performance over several baselines. Even if the parameter number of our model is reduced by 1/3, it could reach a comparable $F_{0.5}$ value of 30.75%. Furthermore, we utilize English datasets to evaluate the generalization and scalability of the proposed model. This could provide a new feasible research direction for CGEC research.

Probability density function based data augmentation for deep neural network automatic modulation classification with limited training data

AbstractDeep neural networks (DNN) based automatic modulation classification (AMC) has achieved high accuracy performance. However, DNNs are data‐hungry models, and training such a model requires a large volume of data. Insufficient training data will cause DNN models to experience overfitting and severe performance degradation. In practical AMC tasks, training the deep model with sufficient data is challenging due to the costly data collection. To this end, a novel probability density function (PDF) based data augmentation scheme and a method to determine the required minimum sampling size for data enlargement is proposed. Compared with the known image‐based augmentation scheme, the proposed waveform‐based PDF technique has low complexity and is easy to implement. Experimental results show that the required size of the training dataset is one order of magnitude smaller than the sufficient dataset in the additive white Gaussian noise channel, and effective recognition can be achieved using around 60% of the total examples under the Rayleigh channel. Moreover, the presented scheme can expand training data under frequency and phase offsets.

Multi-Domain Fusion Graph Network for Semi-Supervised PolSAR Image Classification

The expensive acquisition of labeled data limits the practical use of supervised learning on polarimetric synthetic aperture radar (PolSAR) image analysis. Semi-supervised learning has attracted considerable attention as it can utilize few labeled data and very many unlabeled data. The scattering response of PolSAR data is strongly spatial distribution dependent, which provides rich information about land-cover properties. In this paper, we propose a semi-supervised learning method named multi-domain fusion graph network (MDFGN) to explore the multi-domain fused features including spatial domain and feature domain. Three major factors strengthen the proposed method for PolSAR image analysis. Firstly, we propose a novel sample selection criterion to select reliable unlabeled data for training set expansion. Multi-domain fusion graph is proposed to improve the feature diversity by extending the sample selection from the feature domain to the spatial-feature fusion domain. In this way, the selecting accuracy is improved. By few labeled data, very many accurate unlabeled data are obtained. Secondly, multi-model triplet encoder is proposed to achieve superior feature extraction. Equipped with triplet loss, limited training samples are fully utilized. For expanding training samples with different patch sizes, multiple models are obtained for the fused classification result acquisition. Thirdly, multi-level fusion strategy is proposed to apply different image patch sizes for different expanded training data and obtain the fused classification result. The experiments are conducted on Radarsat-2 and AIRSAR images. With few labeled samples (about 0.003–0.007%), the overall accuracy of the proposed method ranges between 94.78% and 99.24%, which demonstrates the proposed method’s robustness and excellence.

Distress Detection in Subway Tunnel Images via Data Augmentation Based on Selective Image Cropping and Patching.

Distresses, such as cracks, directly reflect the structural integrity of subway tunnels. Therefore, the detection of subway tunnel distress is an essential task in tunnel structure maintenance. This paper presents the performance improvement of deep learning-based distress detection to support the maintenance of subway tunnels through a new data augmentation method, selective image cropping and patching (SICAP). Specifically, we generate effective data for training the distress detection model by focusing on the distressed regions via SICAP. After the data augmentation, we train a distress detection model using the expanded training data. The new image generated based on SICAP does not change the pixel values of the original image. Thus, there is little loss of information, and the generated images are effective in constructing a robust model for various subway tunnel lines. We conducted experiments with some comparative methods. The experimental results show that the detection performance can be improved by our data augmentation.

Enhancing MR image segmentation with realistic adversarial data augmentation.

The success of neural networks on medical image segmentation tasks typically relies on large labeled datasets for model training. However, acquiring and manually labeling a large medical image set is resource-intensive, expensive, and sometimes impractical due to data sharing and privacy issues. To address this challenge, we propose AdvChain, a generic adversarial data augmentation framework, aiming at improving both the diversity and effectiveness of training data for medical image segmentation tasks. AdvChainaugments data with dynamic data augmentation, generating randomly chained photo-metric and geometric transformations to resemble realistic yet challenging imaging variations to expand training data. By jointly optimizing the data augmentation model and a segmentation network during training, challenging examples are generated to enhance network generalizability for the downstream task. The proposed adversarial data augmentation does not rely on generative networks and can be used as a plug-in module in general segmentation networks. It is computationally efficient and applicable for both low-shot supervised and semi-supervised learning. We analyze and evaluate the method on two MR image segmentation tasks: cardiac segmentation and prostate segmentation with limited labeled data. Results show that the proposed approach can alleviate the need for labeled data while improving model generalization ability, indicating its practical value in medical imaging applications.

A multi-branch redundant adversarial net for intelligent fault diagnosis of multiple components under drastically variable speeds

Intelligent fault diagnosis with small training samples plays an important role in the safety of mechanical equipment. However, affected by sharp speed variation, fault feature is extremely weak, which raises difficulty for fault diagnosis. The mutual coupling of multi-component fault features further increases the difficulty. Considering the ability of redundant second generation wavelet transform in non-stationary feature extraction, a multi-branch redundant adversarial net (RedundancyNet) is proposed to address the above issues. The Net consists of discriminator, the generator based on redundant reconstruction, and the classifier based on redundant decomposition. Firstly, through adversarial training process, the generator fuses multi-scale features to generate the signal with varying speeds, thereby expanding training data. Secondly, through layer-by-layer multi-resolution feature enhancement, the classifier boosts weak fault features of vibration signals at variable speeds. Finally, a multi-branch framework is proposed to realize multi-component fault location and damage identification. The proposed method is validated on two cases. The average classification accuracy in the two cases reach 97.14% and 98.33% respectively. However, other end-to-end intelligent fault diagnosis methods for varying speeds or small samples can only reach the highest classification accuracy of 95.14% in Case 1 and 93.59% in Case2, which is much less than RedundancyNet. The analysis results highlight the effectiveness of the net under drastically variable speeds and small faulty training samples. Besides, the proposed classifier is easy to understand, which reveals the process of feature learning and the extracted feature under varying speeds.

Pill Detection Model for Medicine Inspection Based on Deep Learning

This paper proposes a deep learning algorithm that can improve pill identification performance using limited training data. In general, when individual pills are detected in multiple pill images, the algorithm uses multiple pill images from the learning stage. However, when there is an increase in the number of pill types to be identified, the pill combinations in an image increase exponentially. To detect individual pills in an image that contains multiple pills, we first propose an effective database expansion method for a single pill. Then, the expanded training data are used to improve the detection performance. Our proposed method shows higher performance improvement than the existing algorithms despite the limited imaging and data set size. Our proposed method will help minimize problems, such as loss of productivity and human error, which occur while inspecting dispensed pills.

Multi-scale Convolutional Recurrent Neural Network and Data Augmentation for Polyphonic Sound Event Detection

We propose Multi-scale convolutional recurrent neural networks (MCRNN) and data augmentation methods to detect polyphonic sound event with few training data. MCRNN consists of Multi-scale convolutional neural networks (MCNN) and recurrent neural networks (RNN). MCNN concatenates the higher level features extracted using multiple convolution kernels with different scales from the time domain and frequency domain at the same time. RNN is able to capture the longer term temporal context characteristics. A novel background spectrum random replacement (BSRR) data augmentation method is applied to expand training data, which uses standard normal distribution data with randomly selected position and length instead of the original time-domain, frequency-domain or time-frequency domain background spectrum features. Our method is tested on the datasets of DCASE 2019 Task3 (T3). The experimental results showed that the MCRNN and BSRR data augmentation method are efficient. We achieved better results than the first place and the single advanced on the T3 by applying BSRR and SpecAugment data augmentation method simultaneously. On the evaluation dataset (T3-eval), our best result shows 0.05 and 0.975 of error rate (ER) and F1 respectively. Our method got the best performance and relatively improved 17% and 1% than the corresponding values of the single advanced.

Enhancing Natural Language Inference Using New and Expanded Training Data Sets and New Learning Models

Natural Language Inference (NLI) plays an important role in many natural language processing tasks such as question answering. However, existing NLI modules that are trained on existing NLI datasets have several drawbacks. For example, they do not capture the notion of entity and role well and often end up making mistakes such as “Peter signed a deal” can be inferred from “John signed a deal”. As part of this work, we have developed two datasets that help mitigate such issues and make the systems better at understanding the notion of “entities” and “roles”. After training the existing models on the new dataset we observe that the existing models do not perform well on one of the new benchmark. We then propose a modification to the “word-to-word” attention function which has been uniformly reused across several popular NLI architectures. The resulting models perform as well as their unmodified counterparts on the existing benchmarks and perform significantly well on the new benchmarks that emphasize “roles” and “entities”.

A Network Framework for Small-Sample Learning.

Small-sample learning involves training a neural network on a small-sample data set. An expansion of the training set is a common way to improve the performance of neural networks in small-sample learning tasks. However, improper constraints in expanding training data will reduce the performance of the neural networks. In this article, we present certain conditions for incorporation of additional training data. According to these conditions, we propose a neural network framework for self-training using self-generated data called small-sample learning network (SSLN). The SSLN consists of two parts: the expression learning network and the sample recall generative network, both of which are constructed based on restricted Boltzmann machine (RBM). We show that this SSLN can converge as well as the RBM. Moreover, the experiment results on MNIST Digit, SVHN, CIFAR10, and STL-10 data sets reveal the superiority of the SSLN over other models.

The effects of physics-based data augmentation on the generalizability of deep neural networks: Demonstration on nodule false-positive reduction.

An important challenge for deep learning models is generalizing to new datasets that may be acquired with acquisition protocols different from the training set. It is not always feasible to expand training data to the range encountered in clinical practice. We introduce a new technique, physics-based data augmentation (PBDA), that can emulate new computed tomography (CT) data acquisition protocols. We demonstrate two forms of PBDA, emulating increases in slice thickness and reductions of dose, on the specific problem of false-positive reduction in the automatic detection of lung nodules. We worked with CT images from the lung image database consortium (LIDC) collection. We employed a hybrid ensemble convolutional neural network (CNN), which consists of multiple CNN modules (VGG, DenseNet, ResNet), for a classification task of determining whether an image patch was a suspicious nodule or a false positive. To emulate a reduction in tube current, we injected noise by simulating forward projection, noise addition, and backprojection corresponding to 1.5mAs (a "chest x-ray" dose). To simulate thick slice CT scans from thin slice CT scans, we grouped and averaged spatially contiguous CT within thin slice data. The neural network was trained with 10% of the LIDC dataset that was selected to have either the highest tube current or the thinnest slices. The network was tested on the remaining data. We compared PBDA to a baseline with standard geometric augmentations (such as shifts and rotations) and Gaussian noise addition. PBDA improved the performance of the networks when generalizing to the test dataset in a limited number of cases. We found that the best performance was obtained by applying augmentation at very low doses (1.5mAs), about an order of magnitude less than most screening protocols. In the baseline augmentation, a comparable level of Gaussian noise was injected. For dose reduction PBDA, the average sensitivity of 0.931 for the hybrid ensemble network was not statistically different from the average sensitivity of 0.935 without PBDA. Similarly for slice thickness PBDA, the average sensitivity of 0.900 when augmenting with doubled simulated slice thicknesses was not statistically different from the average sensitivity of 0.895 without PBDA. While there were cases detailed in this paper in which we observed improvements, the overall picture was one that suggests PBDA may not be an effective data enrichment tool. PBDA is a newly proposed strategy for mitigating the performance loss of neural networks related to the variation of acquisition protocol between the training dataset and the data that is encountered in deployment or testing. We found that PBDA does not provide robust improvements with the four neural networks (three modules and the ensemble) tested and for the specific task of false-positive reduction in nodule detection.