Convolution Algorithm Research Articles

A GAN-based cyclic iterative unsupervised change detection network

ABSTRACT Remote sensing image change detection has significant applications across various fields. In recent years, the powerful feature extraction capability of deep learning technology has introduced innovative approaches to change detection methodologies. However, the accuracy of convolutional neural network algorithms hinges heavily on labelled data, necessitating substantial manpower and resources to generate labelled samples. Therefore, this paper proposes an unsupervised change detection model called CIUCD. This model operates on the fundamental premise of unsupervised change detection, employing a generator and a segmenter for iterative training. During each iteration, the generator and segmenter update and optimize their parameters, with the segmenter leveraging the generated change images as labels for training purposes. This iterative labelling process circumvents the need for manually labelled data, thereby mitigating the difficulty of obtaining change detection labels for actual remote sensing images. To validate the efficacy of the proposed model, we conducted experiments using actual remotely sensed images from the 2023 Turkey earthquake and the 2022 Afghanistan earthquake events. The results show that our method enhances the F1, Pr, Re and OA by 5.98%, 2.05%, 11.82% and 8.50% (Turkey earthquake), and by 9.54%, 5.87%, 4.52% and 11.34% (Afghanistan earthquake), respectively, when compared to the unsupervised change detection algorithm KPCA-Mnet. Experiments on real remote sensing images of 1000 × 1000 pixels showed that the CIUCD algorithm not only accurately detect larger change areas but is also capable of identifying changes in subtle objects such as solar panels. Additionally, the CIUCD algorithm is resilient to natural environmental factors like the angle of illumination, enabling precise predictions of changes in the actual remote sensing images and yielding clearer detection results. The performance of the CIUCD algorithm remains outstanding even when applied to larger images with a resolution of 3000 × 3000 pixels.

Rapid Convolutional Algorithm for the Discovery of Blueberry Honey Authenticity Markers via Nontargeted LC-MS Analysis.

Bees produce honey through the collection and transformation of nectar, whose botanical origin impacts the taste, nutritional value, and, therefore, the market price of the resulting honey. This phenomenon has led some to mislabel their honey so that it can be sold at a higher price. Metabolomics has been gaining popularity in food authentication, but rapid data mining algorithms are needed to facilitate the discovery of new authenticity markers. A nontargeted high-resolution liquid chromatography-mass spectrometry (HR/LC-MS) analysis of 262 monofloral honey samples, of which 50 were blueberry honey, was performed. Data mining methods were demonstrated for the discovery of binary single-markers (compound was only detected in blueberry honey), threshold single-markers (compound had the highest concentration in blueberry honey), and interval ratio-markers (the ratio of two compounds was within a unique interval in blueberry honey). A novel convolutional algorithm was developed for the discovery of interval ratio-markers, which trained 14× faster and achieved a 0.2 Matthews correlation coefficient (MCC) units higher classification score than existing open-source implementations. The convolutional algorithm also had classification performance similar to that of a brute-force search but trained 1521× faster. A pipeline for shortlisting candidate authenticity markers from the LC-MS spectra that may be suitable for chemical structure identification was also demonstrated and led to the identification of niacin as a blueberry honey threshold single-marker. This work demonstrates an end-to-end approach to mine the honey metabolome for novel authenticity markers and can readily be applied to other types of food and analytical chemistry instruments.

Characterizing the progression of subclinical cardiac amyloidosis through artificial intelligence applied to electrocardiographic images and echocardiograms

Abstract Background Transthyretin amyloid cardiomyopathy (ATTR-CM) follows an indolent course and may remain undiagnosed for many years before resulting in clinical disease. Purpose To assess the utility of scalable artificial intelligence (AI) methods applied to electrocardiographic (ECG) images and echocardiograms in detecting longitudinal changes that precede the clinical diagnosis of ATTR-CM. Methods Across 5 hospitals of a large U.S.-based hospital system, we identified a total of 234 patients (age 79 [IQR:70-85] years, n=56 (23.9%) women) with ATTR-CM, defined by the presence of a positive bone scintigraphy scan (or pharmacotherapy with an approved transthyretin stabilizer) between 2015 and 2023. Using 10:1 age- and sex-matched controls, we independently trained convolutional neural network algorithms to discriminate ATTR-CM from controls using i) ECG images (AUROC 0.915) or ii) echocardiographic studies by incorporating all available views (AUROC 0.93). In a leave-out testing set of n=82 individuals with serial ECGs and n=49 individuals with serial echocardiograms we characterized longitudinal patterns in the evolution of the AI ECG and echo-based scores, and estimated the percentage of individuals that would have screened positive across distinct time intervals. Results In the leave-out testing set 52.4% and 34.7% of all individuals who went on to receive a diagnosis of ATTR-CM would have screened positive by AI deployed to ECGs and echocardiograms, respectively, 1 to 3 years before their confirmatory testing, with the percentage increasing to 75.6% and 69.4% in the 12 months before their eventual diagnosis (A). Across both modalities, there was a longitudinal increase in AI predictions when comparing the median of all studies performed more than a year before diagnosis versus the median of any studies performed within 12 months of diagnosis (AI-Echo: 0.20 [IQR: 0.04-0.35] to 0.49 [0.24-0.59] & AI-ECG: 0.07 [0.02-0.20] to 0.32 [0.11-0.49], p<0.001 for both) (B), consistent with evolution in the ECG and echocardiographic signature of ATTR-CM before its clinical diagnosis. Conclusions We demonstrate that AI applied directly to ECG images and echocardiography may enable scalable monitoring of subclinical ATTR-CM progression.

Collaborative Optimization of Multi-regional Integrated Energy System Based on Improved Beluga Whale Optimization Algorithm

With the installed capacity of the renewable energy power generation is growing at a high speed, a large number of the renewable energy is connected to the grid, the energy problem has been solved to a large extent, but the ensuing problems can not be ignored, the first is the consumption of the renewable energy, which is followed by the stability problem of the power system, and a multi-regional integrated energy system (MRIES) is constructed to solve the problem. In view of the wind power uncertainty, the absorption treatment is carried out by the equipment layer and the optimization layer respectively. A hybrid energy storage system (HESS) is introduced in the equipment layer to suppress the influence of the wind power uncertainty on the system stability. A combination of the convolutional filtering algorithm, the Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise algorithm and the fuzzy control algorithm is introduced in the optimization layer to develop the charging and discharging strategy of the HESS. With the strategy, the impact of the electric power fluctuations on the power system during the optimization process can be reduced largely. Then, on the basis of fully considering the energy coupling in different equipments, a collaborative optimization model of the MRIES is constructed. And the integrated demand response model considering the time-of-use electric price is introduced in the load side. Finally, the improved Beluga optimization(IBWO) algorithm is developed to optimize the model. The optimization results show that the IBWO algorithm plays a good optimization effect both in the participation of energy supply equipments and the economy, plays a collaborative optimization role in the MRIES and ensures the stability of the whole power system.

Implementation of Smart Farm Systems Based on Fog Computing in Artificial Intelligence of Things Environments.

Cloud computing has recently gained widespread attention owing to its use in applications involving the Internet of Things (IoT). However, the transmission of massive volumes of data to a cloud server often results in overhead. Fog computing has emerged as a viable solution to address this issue. This study implements an Artificial Intelligence of Things (AIoT) system based on fog computing on a smart farm. Three experiments are conducted to evaluate the performance of the AIoT system. First, network traffic volumes between systems employing and not employing fog computing are compared. Second, the performance of the communication protocols-hypertext transport protocol (HTTP), message queuing telemetry transport protocol (MQTT), and constrained application protocol (CoAP)-commonly used in IoT applications is assessed. Finally, a convolutional neural network-based algorithm is introduced to determine the maturity level of coffee tree images. Experimental data are collected over ten days from a coffee tree farm in the Republic of Korea. Notably, the fog computing system demonstrates a 26% reduction in the cumulative data volume compared with a non-fog system. MQTT exhibits stable results in terms of the data volume and loss rate. Additionally, the maturity level determination algorithm performed on coffee fruits provides reliable results.

A novel extension model for predicting the friction coefficient of fluorinated ethylene propylene based on temporal convolutional networks expansion algorithms

A novel extension model for predicting the friction coefficient of fluorinated ethylene propylene based on temporal convolutional networks expansion algorithms

Pose Estimation of a Cobot Implemented on a Small AI-Powered Computing System and a Stereo Camera for Precision Evaluation.

The precision of robotic manipulators in the industrial or medical field is very important, especially when it comes to repetitive or exhaustive tasks. Geometric deformations are the most common in this field. For this reason, new robotic vision techniques have been proposed, including 3D methods that made it possible to determine the geometric distances between the parts of a robotic manipulator. The aim of this work is to measure the angular position of a robotic arm with six degrees of freedom. For this purpose, a stereo camera and a convolutional neural network algorithm are used to reduce the degradation of precision caused by geometric errors. This method is not intended to replace encoders, but to enhance accuracy by compensating for degradation through an intelligent visual measurement system. The camera is tested and the accuracy is about one millimeter. The implementation of this method leads to better results than traditional and simple neural network methods.

Continuous Signal Convolution Using FFT with Overlap for Transportation Robot Coordinate Determination

The paper discusses the development of a continuous discrete signal convolution algorithm using fast Fourier transform (FFT) with overlap for more accurate determination of the coordinates of a transportation robot. Two methods for determining the coordinates of a transportation robot in two-dimensional space are presented, where the convolution operation is used and where this algorithm can be applied. The algorithm is based on a method that allows the FFT to be performed on multiple computational units in parallel, such as on a TMS320F28377D digital signal processor, where three computational units are available that operate independently in parallel. The use of a FFT with overlap allows signal processing without data loss, which is critical for continuous signal processing.The study of the influence of the FFT window offset value on the frequency of obtaining the spectrum of the signal has been carried out, and as it was found that reducing the FFT window offset leads to an increase in the frequency of obtaining the spectrum of the signal, which, in turn, allows you to more accurately restore the amplitude of the original signal, that is, there is less loss of spectrum data. As results, the paper provides illustrations of the study in the form of three-dimensional graphs, where the spectra of the signal in time during the FFT with overlap, where as the initial signal is used a sinusoidal signal, which over time increases its frequency. Comparison of the results of the algorithm without overlap and with overlap at different offset values showed that the method with overlap provides significantly more accurate and frequent acquisition of spectrum data, avoiding data loss.The presented results are useful in robotics when it is required to accurately determine the coordinates of a transportation robot.

Tennis teaching assistance model based on double chain shared unsupervised action recognition algorithm

The traditional self-supervised methods based on skeleton data frequently categorize the various enhancements of a given sample as positive examples, while the remaining samples are designated as negative examples. This approach results in a significant imbalance in the ratio of positive to negative samples, which in turn constrains the efficacy of samples with identical semantic information. Therefore, to further improve the training quality in tennis teaching and enhance the accuracy of action recognition, a dual chain sharing unsupervised action recognition algorithm has been proposed. This study first designs a skeleton topology data augmentation method based on the physical connections of human joints to obtain advanced semantic embeddings. Then, an improved positive sample expansion strategy is utilized to enhance the diversity and quality of training data. Next, an unsupervised learning mechanism is employed to autonomously learn and recognize complex patterns of tennis movements. To measure the performance of the model, tests were conducted on its accuracy, F1 score, recognition time, and fitting degree. The experimental results showed that the proposed algorithm could reach its optimal state after 23 iterations of training, and its F1 value reached 0.918, with an average accuracy of 92.9 % and an average recognition time of 7.4 s. The research algorithms were superior to the multi-input branch graph convolutional network action recognition algorithms used for comparison, pull-push contrastive loss action recognition algorithms, and multi-granularity anchor contrastive representation learning action recognition algorithms. Its accuracy was leading by 8.6 %-23.5 %, and the recognition time has been reduced by 2.3s-7.4 s. In addition, in terms of the fitting degree of action recognition, compared with other methods, the proposed method had a fitting degree of up to 95.7 %, which was at least 8.1 % higher. This research method can improve the timeliness of feedback from trainers by accurately and quickly recognizing movements, helping coaches develop more targeted training plans, thereby improving the efficiency and quality of tennis teaching.

Surrogate-model-based low-carbon optimization method for tolerance allocation of mechanical components under bidirectional constraints

The low-carbon optimization of tolerance allocation involves reducing production costs and carbon emissions, while ensuring dimensional precision, making it a crucial pathway to achieving low-carbon manufacturing. However, the complex mapping mechanism between tolerance, cost, and carbon emissions makes it challenging to construct effective mechanistic models. This complexity arises from both the forward constraints of multiple-coupled design features of components and backward constraints on the differentiated processing conditions of an enterprise. Therefore, a low-carbon optimization method for tolerance allocation based on surrogate models is proposed. Firstly, the RReliefF algorithm is utilized to analyze the influence of design features and enterprise processing condition factors on the low-carbon performance of tolerance allocation, identifying key influencing factors. Secondly, a convolutional neural network (CNN)-based surrogate model is established to predict the low-carbon performance of tolerance allocation. Specifically, datasets of tolerance, manufacturing costs, and carbon emissions under different design feature constraints are constructed. To improve the training efficiency and accuracy of the prediction model, the heterogeneous data of design features are processed using the one-hot encoding technique. Subsequently, the mapping relationship between tolerance, cost, and carbon emissions is established based on the CNN. Finally, with the aim of minimizing costs and carbon emissions, a low-carbon optimization method for tolerance allocation is established based on the surrogate model and the product size chain. The convolutional neural network-Harris Hawk optimization algorithm (CNN-HHO) is utilized to generate tolerance allocation schemes that minimize both cost and carbon emissions while satisfying dimensional precision. Through a series of multi-peak test functions and practical case studies of the overrunning clutch, the results indicate that the proposed CNN-HHO algorithm demonstrates exceptional accuracy and stability in the validation on the test functions. Furthermore, the optimized overrunning clutch solution achieves approximately 2.61% cost reduction and 2.63% decrease in carbon emissions, providing further support for the effectiveness of the proposed method.

Multi-channel depth segmentation network based on 3D graph convolution algorithm and its application in point cloud segmentation

At the current stage, 3D graph convolutional algorithms face the following issues: (1) The problem of determining the neighborhood in graph convolution. (2) The issue of fusing features at different depths in deep graph convolutional algorithms. (3) The challenge of feature fusion and recognition in multi-path deep graph convolutional algorithms. Based on these, the paper "Multi-Path Deep Segmentation Network Based on 3D Graph Convolutional Algorithm and Its Application in Point Cloud Segmentation" is proposed. Inspired by visual computing theory, the neighborhood for 3D graph convolution is determined based on the receptive field theory, and a 3D graph convolutional algorithm based on visual selectivity is proposed; A single-link deep feature extraction algorithm for 3D graph convolution based on visual selectivity is proposed, it achieve the fusion of features at different depths learned by the graph convolutional algorithm by a pyramid learning method; A multi-link feature extraction algorithm for 3D graph convolution based on visual selectivity is proposed, and achieve multi-link feature fusion and segmentation to utilize the RPN calculation method. compared with algorithms such as PointNet, PointNet++, KPConv deform, 3D GCN, and My Algorithm, The segmentation performance and geometric invariance of the proposed algorithm are validated on the ShapeNetPart dataset and the custom Mortise_and_Tenon_DB dataset, and Experiments demonstrate that the proposed algorithm is correct and feasible, offers superior 3D point cloud segmentation performance, and exhibits strong geometric invariance.

Study on Self-Learning System for Milling Chatter Feature Based on Hybrid Preprocessing Model and Improved Convolutional Clustering

Milling chatter is a self-excited vibration phenomenon during the cutting process, typically occurring in machining operations with lower stiffness. It significantly impacts the machining precision and surface morphology of the workpiece. To summarize the vibration signal features under operating conditions and enhance the precision of the intelligent recognition model for milling chatter features, this paper proposes an adaptive learning model for chatter feature recognition based on the improved convolutional clustering algorithm. This study uses wavelet analysis to extract chatter features and improved convolutional clustering to identify chatter and stable cutting signals. This method selected appropriate mapping techniques based on differences in chatter signals under various conditions and uses convolution for feature extraction. Improve convolutional clustering evaluates sample similarity through energy transfer, less influenced by signal space structure compared to Euclidean or Mahalanobis distances. This method provides a unified criterion for evaluating different signal feature representation spaces, thereby improving the accuracy and efficiency of chatter sample recognition. The recognition accuracy of chatter phenomena at different spindle speed ranges reaches $95$ % and $96.3$ % respectively. The results show that this method can effectively distinguish between chatter signals and stable cutting signals.

Inverse versus convolution treatment planning algorithms for gamma knife radiosurgery.

To compare the convolution and inverse algorithm plans. The cross-sectional study was conducted from January to May 2022 at the Icon Gamma Knife Centre, Al- Taj Hospital, Baghdad, Iraq, and comprised patients with malignant and benign brain tumours who underwent gamma knife therapy. Each patient's brain was imaged using computed tomography and magnetic resonance imaging. The neurosurgeon prescribed the dose depending on the tumour volume and type, while the medical physicist generated the two plans based on inverse and convolution algorithms. The prescribed dose was delivered to 50% of the isodose line of the tumour. Each plan was evaluated with respect to tumour conformity index, coverage, gradient index, number of shots, and time of treatment. Of the 30 patients, 17(56.7%) were males and 13(43.3%) were females. The overall mean age was 46.29±15.20 years (range: 10-71 years). The mean dose delivered was 15.86±3.86Gy, and the mean number of gamma radiation shots was 12.56±6.95. There was significant difference between the two algorithm plans for all dosimetric parameters, with the inverse plan providing higher coverage and selectivity than convolution plan, but taking longer time(p<0.05), while plan was inverse plan better than convolution plan in terms of gradient and conformity (p<0.05). With more extended treatment, the inverse plan was found to have superior selectivity, coverage, gradient index and Paddick conformity index values compared to the convolution plan.

Optimizing Convolutional Neural Network Architectures

Convolutional neural networks (CNNs) are commonly employed for demanding applications, such as speech recognition, natural language processing, and computer vision. As CNN architectures become more complex, their computational demands grow, leading to substantial energy consumption and complicating their use on devices with limited resources (e.g., edge devices). Furthermore, a new line of research seeking more sustainable approaches to Artificial Intelligence development and research is increasingly drawing attention: Green AI. Motivated by an interest in optimizing Machine Learning models, in this paper, we propose Optimizing Convolutional Neural Network Architectures (OCNNA). It is a novel CNN optimization and construction method based on pruning designed to establish the importance of convolutional layers. The proposal was evaluated through a thorough empirical study including the best known datasets (CIFAR-10, CIFAR-100, and Imagenet) and CNN architectures (VGG-16, ResNet-50, DenseNet-40, and MobileNet), setting accuracy drop and the remaining parameters ratio as objective metrics to compare the performance of OCNNA with the other state-of-the-art approaches. Our method was compared with more than 20 convolutional neural network simplification algorithms, obtaining outstanding results. As a result, OCNNA is a competitive CNN construction method which could ease the deployment of neural networks on the IoT or resource-limited devices.

Staining-Independent Malaria Parasite Detection and Life Stage Classification in Blood Smear Images

Malaria is a leading cause of morbidity and mortality in tropical and sub-tropical regions. This research proposed a malaria diagnosis system based on the you only look once algorithm for malaria parasite detection and the convolutional neural network algorithm for malaria parasite life stage classification. Two public datasets are utilized: MBB and MP-IDB. The MBB dataset includes human blood smears infected with Plasmodium vivax (P. vivax). While the MP-IDB dataset comprises 4 species of malaria parasites: P. vivax, P. ovale, P. malariae, and P. falciparum. Four distinct stages of life exist in every species, including ring, trophozoite, schizont, and gametocyte. For the MBB dataset, detection and classification accuracies of 0.92 and 0.93, respectively, were achieved. For the MP-IDB dataset, the proposed algorithms yielded the accuracies for detection and classification as follows: 0.84 and 0.94 for P. vivax; 0.82 and 0.93 for P. ovale; 0.79 and 0.93 for P. malariae; and 0.92 and 0.96 for P. falciparum. The detection results showed the models trained by P. vivax alone provide good detection capabilities also for other species of malaria parasites. The classification performance showed the proposed algorithms yielded good malaria parasite life stage classification performance. The future directions include collecting more data and exploring more sophisticated algorithms.

Hybrid convolutional neural networks in human resource recommendation

The study proposes a new algorithm combining a gradient boosting tree algorithm with a hybrid convolutional neural network in order to design a better human resource recommendation algorithm to solve the problem of employment difficulties and recruitment difficulties. The algorithm combines the excellent feature transformation ability of gradient boosting trees with the excellent classification ability of hybrid convolutional neural networks, complementing the shortcomings of each of the two algorithms. The outcomes showed that the algorithm performed best with the learning rate set to 0.3 and the maximum tree depth set to 3. The algorithm now has the lowest loss percentage and highest F1-Score value. The maximum-median hybrid pooling approach used in the study had considerable improvements over the algorithm’s pooling strategy (PS), and it had the greatest recall and F1-Score values of all pooling strategies (0.8108 and 0.7418, respectively). The new algorithm outperformed the gradient boosting tree algorithm and the hybrid convolutional neural network algorithm in terms of recall and F1-Score values, and regardless of the length of the job recommendation, the recall and F1-Score values of the new algorithm were consistently higher than those of the old algorithm. The recall and F1-Score values of the new algorithm, with a job recommendation length of 70, were 0.8198 and 0.7432, respectively, both greater than those of the other algorithms, according to a comparison of it with other conventional HR recommendation algorithms. The study’s newly created algorithm increases the efficacy and precision of HR suggestions.

Compression of room impulse responses for compact storage and fast low-latency convolution

Room impulse responses (RIRs) are used in several applications, such as augmented reality and virtual reality. These applications require a large number of RIRs to be convolved with audio, under strict latency constraints. In this paper, we consider the compression of RIRs, in conjunction with fast time-domain convolution. We consider three different methods of RIR approximation for the purpose of RIR compression and compare them to state-of-the-art compression. The methods are evaluated using several standard objective quality measures, both channel-based and signal-based. We also propose a novel low-rank-based algorithm for fast time-domain convolution and show how the convolution can be carried out without the need to decompress the RIR. Numerical simulations are performed using RIRs of different lengths, recorded in three different rooms. It is shown that compression using low-rank approximation is a very compelling option to the state-of-the-art Opus compression, as it performs as well or better than on all but one considered measure, with the added benefit of being amenable to fast time-domain convolution.

An approach of 2D convolutional neural network–based seismic data fault interpretation with linear annotation and pixel thinking

AbstractThis article introduces an improved method for geological fault interpretation utilizing a 2D convolutional neural network approach with a focus on coalbed horizons, dealing with the problem as an image classification task. At the beginning, linear annotations, reflecting geological fault features, are applied to multiple seismic sections. By considering texture differences between fault and non‐fault areas, we construct samples that represent these distinct zones for training deep neural networks. Initially, fault annotations are transformed into single dots to facilitate pixel‐based processing. To depict a specific dot's geological structure, we employ a matrix clipped around the point, determined by a combination of range and step parameters. Convolutional layers generate filters equivalent to seismic data transformation, streamlining the need for analysis and selection of seismic attributes. The article discusses enhancing the efficiency of 2D convolutional neural network–based fault interpretation by optimizing sample selection, data extraction and model construction procedures. Through the incorporation of data from two mining areas (totalling 27.09 km2) in sample creation, the overall accuracy exceeds 0.99. Recognition extends seamlessly to unlabelled sections, showcasing the innovative technical route and methodology of fault interpretation with linear annotation and pixel‐based thinking. This study presents a method that integrates planar and raster thinking, transitioning from vision‐oriented geological structure annotation to algorithm‐oriented pixel location. The proposed 2D convolutional neural network–based matrix‐oriented fault/non‐fault binary classification demonstrates feasibility and reproducibility, offering a new automated approach for fault detection in coalbeds through convolutional neural network algorithms.

Research and Modification on the Park-Ang Damage Index for Railway Rectangular Piers

ABSTRACT This study addressed the issue of distortion in evaluating the seismic damage state of railway bridge piers using the original Park-Ang damage index. By constructing a parameterized co-simulation program with two finite element models, OpenSees and Abaqus, five sets of flexural failure and five sets of flexural-shear failure rectangular pier quasi-static test results were selected to validate the rationality of the calculation program. Four hundred and eighty parameterized calculations were conducted using this program. The combination coefficient β and critical damage index DI within the Park-Ang damage index were modified using nonlinear regression algorithm and convolutional neural network algorithm. The adaptability of the damage index calculation was ultimately verified through an engineering case study of a simply supported beam bridge. The research findings are as follows: 1) The main factor influencing β is the longitudinal reinforcement ratio, while the axial compression ratio, shear span ratio, volumetric stirrup ratio, aspect ratio, concrete strength, and steel strength are secondary parameters influencing β. 2) Compared to the nonlinear regression algorithm, the convolutional neural network algorithm can better establish a calculation model between pier structural parameters and β, clarifying their relationship. 3) It is recommended to use critical values of 0.11, 0.29, 0.49, and 1.00 for assessing pier damage as no, slight, moderate, severe, and collapse damage when using the damage index calculation model. 4) The Park-Ang index corrected by convolutional neural networks demonstrates improved accuracy and computational stability, making it suitable for practical assessment of seismic damage in bridge structures.

Deep learning-based ultrasonographic classification of canine chronic kidney disease.

In veterinary medicine, attempts to apply artificial intelligence (AI) to ultrasonography have rarely been reported, and few studies have investigated the value of AI in ultrasonographic diagnosis. This study aimed to develop a deep learning-based model for classifying the status of canine chronic kidney disease (CKD) using renal ultrasonographic images and assess its diagnostic performance in comparison with that of veterinary imaging specialists, thereby verifying its clinical utility. In this study, 883 ultrasonograms were obtained from 198 dogs, including those diagnosed with CKD according to the International Renal Interest Society (IRIS) guidelines and healthy dogs. After preprocessing and labeling each image with its corresponding IRIS stage, the renal regions were extracted and classified based on the IRIS stage using the convolutional neural network-based object detection algorithm You Only Look Once. The training scenarios consisted of multi-class classification, categorization of images into IRIS stages, and four binary classifications based on specific IRIS stages. To prevent model overfitting, we balanced the dataset, implemented early stopping, used lightweight models, and applied dropout techniques. Model performance was assessed using accuracy, recall, precision, F1 score, and receiver operating characteristic curve and compared with the diagnostic accuracy of four specialists. Inter- and intra-observer variabilities among specialists were also evaluated. The developed model exhibited a low accuracy of 0.46 in multi-class classification. However, a significant performance improvement was observed in binary classifications, with the model designed to distinguish stage 3 or higher showing the highest accuracy of 0.85. In this classification, recall, precision, and F1 score values were all 0.85, and the area under the curve was 0.89. Compared with radiologists, whose accuracy ranged from 0.48 to 0.62 in this experimental scenario, the AI model exhibited superiority. Intra-observer reliability among radiologists was substantial, whereas inter-observer variability showed a moderate level of agreement. This study developed a deep-learning framework capable of reliably classifying CKD IRIS stages 3 and 4 in dogs using ultrasonograms. The developed framework demonstrated higher accuracy than veterinary imaging specialists and provided more objective and consistent interpretations. Therefore, deep-learning-based ultrasound diagnostics are potentially valuable tools for diagnosing CKD in dogs.