Pub Date : 2023-12-22DOI: 10.5755/j01.itc.52.4.33460
Guoxiang Li, Xuejun Wang, Yun Li, Zhitian Li
UAV images are characterized by small targets, difficult to identify in the background image, clustering and sparse distribution of targets, etc. Many researchers have proposed the clustering target detection method (ClusDet) for UAV images. However, due to the large differences in target scales and uneven distribution of targets in UAV images, showing long-tailed distribution, the traditional ClusDet algorithm tends to truncate large and medium targets in the process of clustering; in the detection process, the fixed-threshold NMS method in the ClusDet algorithm is difficult to adaptively detect targets of different sizes, clustering and mutual occlusion. To address the above problems, this paper proposes an adaptive clustered target detection algorithm based on UAV images under long-tail distribution. The method is divided into three sub-networks: the adaptive clustering sub-network, which outputs several segmented images of small target clustering regions by extracting potential small target clustering regions in UAV aerial images; the segmentation and filling sub-network, which fills the images with disproportionate aspect ratio for the output of the adaptive clustering network to keep the size of the images within the reasonable range required by the detection network; and the detection sub-network, which detects the targets within the reasonable range required by the detection network by introducing attention mechanism, using variable threshold NMS, and training using sample balancing strategy effectively improve the detection accuracy of targets in the clustered region. Trained in VisDrone 2019 dataset, the simulation results show that the UAV image adaptive clustering target detection method based on long-tailed distribution has a large improvement in the detection accuracy of small targets, and can effectively improve the detection accuracy of the model for targets in the aggregation region, while the model has good generalization ability.
{"title":"Adaptive clustering object detection method for UAV images under long-tailed distributions","authors":"Guoxiang Li, Xuejun Wang, Yun Li, Zhitian Li","doi":"10.5755/j01.itc.52.4.33460","DOIUrl":"https://doi.org/10.5755/j01.itc.52.4.33460","url":null,"abstract":"UAV images are characterized by small targets, difficult to identify in the background image, clustering and sparse distribution of targets, etc. Many researchers have proposed the clustering target detection method (ClusDet) for UAV images. However, due to the large differences in target scales and uneven distribution of targets in UAV images, showing long-tailed distribution, the traditional ClusDet algorithm tends to truncate large and medium targets in the process of clustering; in the detection process, the fixed-threshold NMS method in the ClusDet algorithm is difficult to adaptively detect targets of different sizes, clustering and mutual occlusion. To address the above problems, this paper proposes an adaptive clustered target detection algorithm based on UAV images under long-tail distribution. The method is divided into three sub-networks: the adaptive clustering sub-network, which outputs several segmented images of small target clustering regions by extracting potential small target clustering regions in UAV aerial images; the segmentation and filling sub-network, which fills the images with disproportionate aspect ratio for the output of the adaptive clustering network to keep the size of the images within the reasonable range required by the detection network; and the detection sub-network, which detects the targets within the reasonable range required by the detection network by introducing attention mechanism, using variable threshold NMS, and training using sample balancing strategy effectively improve the detection accuracy of targets in the clustered region. Trained in VisDrone 2019 dataset, the simulation results show that the UAV image adaptive clustering target detection method based on long-tailed distribution has a large improvement in the detection accuracy of small targets, and can effectively improve the detection accuracy of the model for targets in the aggregation region, while the model has good generalization ability.","PeriodicalId":54982,"journal":{"name":"Information Technology and Control","volume":"7 1","pages":""},"PeriodicalIF":1.1,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138947793","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-22DOI: 10.5755/j01.itc.52.4.32642
Min Xu, Guanyu Zhang, Lin Duan
Abstract: There are many and complex big data for digital power grid regulation, which makes it more difficult to manage big data assets. Therefore, a model of big data asset management system for digital power grid regulation has been built. The model consists of three parts: data collection, data security storage and data index. The data acquisition architecture is designed, and the grey prediction method is used to fill the missing values and correct the abnormal values of the data acquisition results. Store the filled and amended data in the blockchain to ensure data security. The AR-tree index organization is used to realize the digital grid regulation big data index and achieve the goal of high-quality management of digital grid regulation big data assets. The experimental results show that the average recall and precision of this method are 96.9% and 97.9% respectively, and the data collection quality is high; After the application of this method, there is almost no non security data, and the proportion of security data is higher, which shows that this method can ensure the security of big data storage; The response time of digital power grid regulation big data index is less than 0.21s, and the index efficiency is higher.
摘 要： 数字电网调控大数据多而复杂，增加了大数据资产管理的难度。因此，构建了数字电网调控大数据资产管理系统模型。该模型由数据采集、数据安全存储和数据索引三部分组成。设计了数据采集架构，采用灰色预测法对数据采集结果进行缺失值填充和异常值修正。将填充和修正后的数据存储在区块链中，确保数据安全。采用 AR 树索引组织实现数字电网监管大数据索引，实现数字电网监管大数据资产高质量管理的目标。实验结果表明，该方法的平均召回率和精度分别为96.9%和97.9%，数据采集质量较高；应用该方法后，几乎没有非安全数据，且安全数据比例较高，说明该方法能够保证大数据存储的安全性；数字电网监管大数据索引响应时间小于0.21s，索引效率较高。
{"title":"Model construction of big data asset management system for digital power grid regulation","authors":"Min Xu, Guanyu Zhang, Lin Duan","doi":"10.5755/j01.itc.52.4.32642","DOIUrl":"https://doi.org/10.5755/j01.itc.52.4.32642","url":null,"abstract":"Abstract: There are many and complex big data for digital power grid regulation, which makes it more difficult to manage big data assets. Therefore, a model of big data asset management system for digital power grid regulation has been built. The model consists of three parts: data collection, data security storage and data index. The data acquisition architecture is designed, and the grey prediction method is used to fill the missing values and correct the abnormal values of the data acquisition results. Store the filled and amended data in the blockchain to ensure data security. The AR-tree index organization is used to realize the digital grid regulation big data index and achieve the goal of high-quality management of digital grid regulation big data assets. The experimental results show that the average recall and precision of this method are 96.9% and 97.9% respectively, and the data collection quality is high; After the application of this method, there is almost no non security data, and the proportion of security data is higher, which shows that this method can ensure the security of big data storage; The response time of digital power grid regulation big data index is less than 0.21s, and the index efficiency is higher.","PeriodicalId":54982,"journal":{"name":"Information Technology and Control","volume":"6 11","pages":""},"PeriodicalIF":1.1,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138944542","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-22DOI: 10.5755/j01.itc.52.4.34476
Caihua Qiu, Feng Ding, Xiu He, Mengbo Wang
This study explored the feasibility of using hyperspectral imaging technology and to identify Osmanthus fragrans seeds with different vigor under computer algorithm and physical system. Two varieties of Osmanthus seeds (JinQiGui and RiXiangGui) were artificially aged and then hyperspectral data were collected. Multivariate scattering correction (MSC) was used for spectral preprocessing. The selection of characteristic wavelength was realized by competitive adaptive reweighted sampling algorithm (CARS). The extreme learning machine (ELM) and k-nearest neighbor (KNN) were used to establish the spectral discriminant model, and convolutional neural network was used in the computer image discriminant model. The results show that the ability to recognize different vigor JQG was better than RXG. MSC preprocessing can not only make the data distribution more aggregated, but also effectively improve the accuracy of the model. MSC+CARS combined with discriminant model can be realized close to 100% recognition with fewer bands. Compared with machine learning model, image- depth learning model can get higher model accuracy for different vigor JQG and RXG without complex preprocessing. These results indicate that hyperspectral imaging technology can effectively distinguish different vigor of Osmanthus fragrans seeds based on computer technology and physical system, which is of great significance for future research.
{"title":"Apply Physical System Model and Computer Algorithm to Identify Osmanthus Fragrans Seed Vigor Based on Hyperspectral Imaging and Convolutional Neural Network","authors":"Caihua Qiu, Feng Ding, Xiu He, Mengbo Wang","doi":"10.5755/j01.itc.52.4.34476","DOIUrl":"https://doi.org/10.5755/j01.itc.52.4.34476","url":null,"abstract":"This study explored the feasibility of using hyperspectral imaging technology and to identify Osmanthus fragrans seeds with different vigor under computer algorithm and physical system. Two varieties of Osmanthus seeds (JinQiGui and RiXiangGui) were artificially aged and then hyperspectral data were collected. Multivariate scattering correction (MSC) was used for spectral preprocessing. The selection of characteristic wavelength was realized by competitive adaptive reweighted sampling algorithm (CARS). The extreme learning machine (ELM) and k-nearest neighbor (KNN) were used to establish the spectral discriminant model, and convolutional neural network was used in the computer image discriminant model. The results show that the ability to recognize different vigor JQG was better than RXG. MSC preprocessing can not only make the data distribution more aggregated, but also effectively improve the accuracy of the model. MSC+CARS combined with discriminant model can be realized close to 100% recognition with fewer bands. Compared with machine learning model, image- depth learning model can get higher model accuracy for different vigor JQG and RXG without complex preprocessing. These results indicate that hyperspectral imaging technology can effectively distinguish different vigor of Osmanthus fragrans seeds based on computer technology and physical system, which is of great significance for future research.","PeriodicalId":54982,"journal":{"name":"Information Technology and Control","volume":"4 21","pages":""},"PeriodicalIF":1.1,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138945070","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-22DOI: 10.5755/j01.itc.52.4.33828
Wuliang Gao
In deep learning, model quality is extremely important. Consequently, the quality and the sufficiency of the datasets for training models have attracted considerable attention from both industry and academia. Automatic data augmentation, which provides a means of using image processing operators to generate data from existing datasets, is quite effective in searching for mutants of the images and expanding the training datasets. However, existing automatic data augmentation techniques often fail to fully exploit the potential of the data, failing to balance the search efficiency and the model accuracy. This paper presents CAugment, a novel approach to diversifying image datasets by combining image processing operators. Given a training image dataset, CAugment is composed of: 1) the three-level evolutionary algorithm (TLEA) that employs three levels of atomic operations for augmenting the dataset and an adaptive strategy for decreasing granularity and 2) a design that uses the three-dimensional evaluation method (TDEM) and a dHash algorithm to measure the diversity of the dataset. The search space can be expanded, which further improves model accuracy during training. We use CAugment to augment the CIFAR-10/100 and SVHN datasets and use the augmented datasets to train the WideResNet and Shake-Shake models. Our results show that the amount of data increases linearly along with the training epochs; in addition, the models trained by the CAugment-augmented datasets outperform those trained by the datasets augmented by the other techniques by up to 17.9% in accuracy on the SVHN dataset.
{"title":"CAugment: An Approach to Diversifying Dataset by Combining Image Processing Operations","authors":"Wuliang Gao","doi":"10.5755/j01.itc.52.4.33828","DOIUrl":"https://doi.org/10.5755/j01.itc.52.4.33828","url":null,"abstract":"In deep learning, model quality is extremely important. Consequently, the quality and the sufficiency of the datasets for training models have attracted considerable attention from both industry and academia. Automatic data augmentation, which provides a means of using image processing operators to generate data from existing datasets, is quite effective in searching for mutants of the images and expanding the training datasets. However, existing automatic data augmentation techniques often fail to fully exploit the potential of the data, failing to balance the search efficiency and the model accuracy. This paper presents CAugment, a novel approach to diversifying image datasets by combining image processing operators. Given a training image dataset, CAugment is composed of: 1) the three-level evolutionary algorithm (TLEA) that employs three levels of atomic operations for augmenting the dataset and an adaptive strategy for decreasing granularity and 2) a design that uses the three-dimensional evaluation method (TDEM) and a dHash algorithm to measure the diversity of the dataset. The search space can be expanded, which further improves model accuracy during training. We use CAugment to augment the CIFAR-10/100 and SVHN datasets and use the augmented datasets to train the WideResNet and Shake-Shake models. Our results show that the amount of data increases linearly along with the training epochs; in addition, the models trained by the CAugment-augmented datasets outperform those trained by the datasets augmented by the other techniques by up to 17.9% in accuracy on the SVHN dataset.","PeriodicalId":54982,"journal":{"name":"Information Technology and Control","volume":"58 4","pages":""},"PeriodicalIF":1.1,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138945802","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-22DOI: 10.5755/j01.itc.52.4.34479
Hui Xie
With the ever-increasing amount of vehicle data being generated, the collection and transmission of this data-to-data processing centers is consuming significant amounts of communication resources. The traditional method of compressing and transmitting the vehicle data is not effective in addressing the issue of efficient utilization of this data. In order to overcome this challenge, we propose an adaptive federated learning approach that avoids the need for transmitting data per vehicle. Our approach leverages the vehicle as a distributed training device node and enables the training of vehicle data using the vehicle's own computing power, thereby eliminating the need to transmit the data over the network. To further enhance the efficiency of the federated learning aggregation calculation, we introduce the information entropy function and cosine similarity calculation. By computing the similarity between the model and the benchmark model, we are able to give a new round of model aggregation calculation weight. Finally, we validate the proposed algorithm using the actual MNIST dataset, demonstrating its high effectiveness.
{"title":"Weight Coefficient Based Adaptive Federated Learning for Vehicular Data Transmission","authors":"Hui Xie","doi":"10.5755/j01.itc.52.4.34479","DOIUrl":"https://doi.org/10.5755/j01.itc.52.4.34479","url":null,"abstract":"With the ever-increasing amount of vehicle data being generated, the collection and transmission of this data-to-data processing centers is consuming significant amounts of communication resources. The traditional method of compressing and transmitting the vehicle data is not effective in addressing the issue of efficient utilization of this data. In order to overcome this challenge, we propose an adaptive federated learning approach that avoids the need for transmitting data per vehicle. Our approach leverages the vehicle as a distributed training device node and enables the training of vehicle data using the vehicle's own computing power, thereby eliminating the need to transmit the data over the network. To further enhance the efficiency of the federated learning aggregation calculation, we introduce the information entropy function and cosine similarity calculation. By computing the similarity between the model and the benchmark model, we are able to give a new round of model aggregation calculation weight. Finally, we validate the proposed algorithm using the actual MNIST dataset, demonstrating its high effectiveness.","PeriodicalId":54982,"journal":{"name":"Information Technology and Control","volume":"12 6","pages":""},"PeriodicalIF":1.1,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138947240","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-22DOI: 10.5755/j01.itc.52.4.33527
Yüksel Eraslan, Tugrul Oktay
An aerial vehicle design process usually aims to maximize performance in a specific flight phase regarding a particular topic such as aerodynamics, flight qualities, or control. This paper proposes a multidisciplinary enhancement both in aerodynamics and longitudinal autonomous flight performance (LAFP) via modern simultaneous design methodology conducted with a novel morphing idea. In this regard, the main wing of a fixed-wing unmanned aerial vehicle (UAV) is redesigned with wingtips capable of altering its taper ratio which results in a semi-tapered planform. The dynamic model of morphing aircraft is constituted from data obtained by numerical and analytical approaches for a number of morphing scenarios. The LAFP is identified as the sum of trajectory tracking parameters which are rise time, settling time, and maximum overshoot, while aerodynamic performance is the lift-to-drag ratio. A hierarchically structured control system is designed and the proportional-integral-differential (PID) controller coefficients and the taper ratio of the morphing wingtip are optimized via the Simultaneous Perturbation Stochastic Ap-proximation (SPSA) algorithm. The k-nearest neighbor (k-NN) machine learning algorithm is also conducted to expand the data limited within the investigated range of morphing scenarios so as to have higher accuracy in optimization. Finally, flight simulations of the morphing UAV with optimal wing and control system design are carried out, closed-loop responses are examined in the presence of the von-Karman turbulence model, and the obtained satisfactory results are presented for both disciplines.
航空飞行器的设计过程通常旨在最大限度地提高特定飞行阶段的性能，涉及空气动力学、飞行品质或控制等特定主题。本文提出了一种通过现代同步设计方法，采用新颖的变形理念，同时提高空气动力学和纵向自主飞行性能（LAFP）的多学科方法。在这方面，对固定翼无人飞行器（UAV）的主翼进行了重新设计，翼尖能够改变其锥形比，从而形成半锥形平面。变形飞机的动态模型是通过数值和分析方法获得的一些变形方案的数据建立的。LAFP 被确定为上升时间、稳定时间和最大过冲等轨迹跟踪参数的总和，而气动性能则是升阻比。设计了一个分层结构控制系统，并通过同步扰动随机拟合（SPSA）算法优化了比例-积分-微分（PID）控制器系数和变形翼尖的锥度比。此外，还采用了 k 近邻（k-NN）机器学习算法，以扩大所研究的变形场景范围内的数据限制，从而提高优化的准确性。最后，对采用最佳机翼和控制系统设计的变形无人机进行了飞行模拟，并在 von-Karman 湍流模型存在的情况下对闭环响应进行了检验，结果令人满意。
{"title":"Multidisciplinary Performance Enhancement on a Fixed-wing Unmanned Aerial Vehicle via Simultaneous Morphing Wing and Control System Design","authors":"Yüksel Eraslan, Tugrul Oktay","doi":"10.5755/j01.itc.52.4.33527","DOIUrl":"https://doi.org/10.5755/j01.itc.52.4.33527","url":null,"abstract":"An aerial vehicle design process usually aims to maximize performance in a specific flight phase regarding a particular topic such as aerodynamics, flight qualities, or control. This paper proposes a multidisciplinary enhancement both in aerodynamics and longitudinal autonomous flight performance (LAFP) via modern simultaneous design methodology conducted with a novel morphing idea. In this regard, the main wing of a fixed-wing unmanned aerial vehicle (UAV) is redesigned with wingtips capable of altering its taper ratio which results in a semi-tapered planform. The dynamic model of morphing aircraft is constituted from data obtained by numerical and analytical approaches for a number of morphing scenarios. The LAFP is identified as the sum of trajectory tracking parameters which are rise time, settling time, and maximum overshoot, while aerodynamic performance is the lift-to-drag ratio. A hierarchically structured control system is designed and the proportional-integral-differential (PID) controller coefficients and the taper ratio of the morphing wingtip are optimized via the Simultaneous Perturbation Stochastic Ap-proximation (SPSA) algorithm. The k-nearest neighbor (k-NN) machine learning algorithm is also conducted to expand the data limited within the investigated range of morphing scenarios so as to have higher accuracy in optimization. Finally, flight simulations of the morphing UAV with optimal wing and control system design are carried out, closed-loop responses are examined in the presence of the von-Karman turbulence model, and the obtained satisfactory results are presented for both disciplines.","PeriodicalId":54982,"journal":{"name":"Information Technology and Control","volume":"58 41","pages":""},"PeriodicalIF":1.1,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138946329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-22DOI: 10.5755/j01.itc.52.4.34021
Cong Shi, Rui Zhai, Yalin Song, Junyang Yu, Han Li, Yingqi Wang, Longge Wang
Traditional deep learning-based strategies for sentiment analysis rely heavily on large-scale labeled datasets for model training, but these methods become less effective when dealing with small-scale datasets. Fine-tuning large pre-trained models on small datasets is currently the most commonly adopted approach to tackle this issue. Recently, prompt-based learning has gained significant attention as a promising research area. Although prompt-based learning has the potential to address data scarcity problems by utilizing prompts to reformulate downstream tasks, the current prompt-based methods for few-shot sentiment analysis are still considered inefficient. To tackle this challenge, an adaptive prompt-based learning method is proposed, which includes two aspects. Firstly, an adaptive prompting construction strategy is proposed, which can capture the semantic information of texts by utilizing a dot-product attention structure, improving the quality of the prompt templates. Secondly, contrastive learning is applied to the implicit word vectors obtained twice during the training stage to alleviate over-fitting in few-shot learning processes. This improves the model’s generalization ability by achieving data enhancement while keeping the semantic information of input sentences unchanged. Experimental results on the ERPSTMT datasets of FewCLUE demonstrate that the proposed method have great ability to construct suitable adaptive prompts and outperforms the state-of-the-art baselines.
{"title":"Few-shot Sentiment Analysis Based on Adaptive Prompt Learning and Contrastive Learning","authors":"Cong Shi, Rui Zhai, Yalin Song, Junyang Yu, Han Li, Yingqi Wang, Longge Wang","doi":"10.5755/j01.itc.52.4.34021","DOIUrl":"https://doi.org/10.5755/j01.itc.52.4.34021","url":null,"abstract":"Traditional deep learning-based strategies for sentiment analysis rely heavily on large-scale labeled datasets for model training, but these methods become less effective when dealing with small-scale datasets. Fine-tuning large pre-trained models on small datasets is currently the most commonly adopted approach to tackle this issue. Recently, prompt-based learning has gained significant attention as a promising research area. Although prompt-based learning has the potential to address data scarcity problems by utilizing prompts to reformulate downstream tasks, the current prompt-based methods for few-shot sentiment analysis are still considered inefficient. To tackle this challenge, an adaptive prompt-based learning method is proposed, which includes two aspects. Firstly, an adaptive prompting construction strategy is proposed, which can capture the semantic information of texts by utilizing a dot-product attention structure, improving the quality of the prompt templates. Secondly, contrastive learning is applied to the implicit word vectors obtained twice during the training stage to alleviate over-fitting in few-shot learning processes. This improves the model’s generalization ability by achieving data enhancement while keeping the semantic information of input sentences unchanged. Experimental results on the ERPSTMT datasets of FewCLUE demonstrate that the proposed method have great ability to construct suitable adaptive prompts and outperforms the state-of-the-art baselines.","PeriodicalId":54982,"journal":{"name":"Information Technology and Control","volume":"44 43","pages":""},"PeriodicalIF":1.1,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138946554","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-22DOI: 10.5755/j01.itc.52.4.32693
Dan Zhang, Shaoxin Zheng, Wanchun Fu
Abstract: In view of the complex influencing factors of tax revenue, the highly non-linear relationship among the influencing factors and the difficulty in predicting tax revenue, this paper proposes to use GM (1, 1) Combined with LSSVM, and it calculates the tax forecasting of China. This paper selects the proportion of the first industry, the ratio of import and export trade to GDP, GDP, the number of urban employment population, the proportion of residents' disposable income and tax revenue in fiscal revenue as the influencing factors, and uses GM (1, 1) and LSSVM respectively to predict the tax revenue of our country, establishes the quadratic programming model to determine the optimal combination weight for the formation of the combination predicting model of tax revenue in our country, make an empirical analysis with the tax revenue of our country from 2000 to 2018 as the research object, and compare the prediction results with LSSVM model, GM (1,1) model and improved GM (1,1) model. The results show that the prediction model of China's tax revenue based on GM (1,1) and LSSVM has a high fitting accuracy with the test set, which can reflect the complex non-linear relationship between various factors. It is of great significance for the development of prediction on Chinese tax revenue and the formulation of a scientific and effective national financial budget.
摘要：针对税收收入影响因素复杂、各影响因素之间高度非线性关系以及税收收入预测难度大等问题，本文提出采用GM（1，1）结合LSSVM，对我国税收预测进行计算。本文选取第一产业比重、进出口贸易占 GDP 比重、GDP、城镇就业人口数、居民可支配收入和税收收入占财政收入比重作为影响因素，分别利用 GM（1，1）和 LSSVM 对我国税收收入进行预测、建立二次编程模型，确定形成我国税收收入组合预测模型的最优组合权重，以我国2000-2018年税收收入为研究对象进行实证分析，并与LSSVM模型、GM（1，1）模型和改进的GM（1，1）模型的预测结果进行比较。结果表明，基于GM（1,1）和LSSVM的我国税收收入预测模型与测试集的拟合精度较高，能够反映各因素之间复杂的非线性关系。这对中国税收预测的发展和制定科学有效的国家财政预算具有重要意义。
{"title":"Research on the Prediction Model of Chinese Tax Revenue Based on GM(1,1) and LSSVM","authors":"Dan Zhang, Shaoxin Zheng, Wanchun Fu","doi":"10.5755/j01.itc.52.4.32693","DOIUrl":"https://doi.org/10.5755/j01.itc.52.4.32693","url":null,"abstract":"Abstract: In view of the complex influencing factors of tax revenue, the highly non-linear relationship among the influencing factors and the difficulty in predicting tax revenue, this paper proposes to use GM (1, 1) Combined with LSSVM, and it calculates the tax forecasting of China. This paper selects the proportion of the first industry, the ratio of import and export trade to GDP, GDP, the number of urban employment population, the proportion of residents' disposable income and tax revenue in fiscal revenue as the influencing factors, and uses GM (1, 1) and LSSVM respectively to predict the tax revenue of our country, establishes the quadratic programming model to determine the optimal combination weight for the formation of the combination predicting model of tax revenue in our country, make an empirical analysis with the tax revenue of our country from 2000 to 2018 as the research object, and compare the prediction results with LSSVM model, GM (1,1) model and improved GM (1,1) model. The results show that the prediction model of China's tax revenue based on GM (1,1) and LSSVM has a high fitting accuracy with the test set, which can reflect the complex non-linear relationship between various factors. It is of great significance for the development of prediction on Chinese tax revenue and the formulation of a scientific and effective national financial budget.","PeriodicalId":54982,"journal":{"name":"Information Technology and Control","volume":"82 13","pages":""},"PeriodicalIF":1.1,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138945362","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-22DOI: 10.5755/j01.itc.52.4.34233
B. M. Brinda, Rajan C
Globally, the prevalence of chronic kidney disease (CKD) is steadily increasing. Computer-aided automated diagnostic (CAD) methods play a significant part in predicting CKD. Due to their highly effective classification accuracy, CAD systems like deep learning algorithms are essential in diagnosing diseases. This research creates an innovative categorization model with a metaheuristic algorithm based on the best characteristic selection to diagnose chronic kidney disease. Data with the absence of values were first removed during the pre-processing phase. Then, the optimal assortment of attributes is chosen using the Squirrel Search algorithm, a metaheuristic method that aids in more precise disorder prediction or categorization. Conditional Variational Generative Adversarial Networks were suggested for classification to identify the presence of CKD. Performance measures such as accuracy, precision, recall, and F1 score were evaluated on the benchmark CKD dataset to determine the efficiency of the suggested feature selection-based classifier. According to the experimental findings, the proposed method outperformed existing classification models with accuracy, precision, recall, and F1 score values of 99.2%, 98.4%, 98.6%, and 98.9%, respectively.
在全球范围内，慢性肾脏病（CKD）的发病率正在稳步上升。 计算机辅助自动诊断（CAD）方法在预测慢性肾脏病方面发挥着重要作用。由于其高效的分类准确性，深度学习算法等计算机辅助自动诊断系统在疾病诊断中至关重要。这项研究利用基于最佳特征选择的元启发式算法创建了一个创新的分类模型，用于诊断慢性肾病。在预处理阶段，首先移除没有数值的数据。然后，使用松鼠搜索算法（一种元启发式方法，有助于更精确地预测疾病或进行分类）选择最佳属性组合。建议使用条件变异生成对抗网络进行分类，以识别是否存在 CKD。在基准 CKD 数据集上评估了准确率、精确度、召回率和 F1 分数等性能指标，以确定所建议的基于特征选择的分类器的效率。实验结果表明，建议的方法优于现有的分类模型，准确率、精确率、召回率和 F1 分数分别为 99.2%、98.4%、98.6% 和 98.9%。
{"title":"Chronic Kidney Disease Diagnosis Using Conditional Variational Generative Adversarial Networks and Squirrel Search Algorithm","authors":"B. M. Brinda, Rajan C","doi":"10.5755/j01.itc.52.4.34233","DOIUrl":"https://doi.org/10.5755/j01.itc.52.4.34233","url":null,"abstract":"Globally, the prevalence of chronic kidney disease (CKD) is steadily increasing. Computer-aided automated diagnostic (CAD) methods play a significant part in predicting CKD. Due to their highly effective classification accuracy, CAD systems like deep learning algorithms are essential in diagnosing diseases. This research creates an innovative categorization model with a metaheuristic algorithm based on the best characteristic selection to diagnose chronic kidney disease. Data with the absence of values were first removed during the pre-processing phase. Then, the optimal assortment of attributes is chosen using the Squirrel Search algorithm, a metaheuristic method that aids in more precise disorder prediction or categorization. Conditional Variational Generative Adversarial Networks were suggested for classification to identify the presence of CKD. Performance measures such as accuracy, precision, recall, and F1 score were evaluated on the benchmark CKD dataset to determine the efficiency of the suggested feature selection-based classifier. According to the experimental findings, the proposed method outperformed existing classification models with accuracy, precision, recall, and F1 score values of 99.2%, 98.4%, 98.6%, and 98.9%, respectively.","PeriodicalId":54982,"journal":{"name":"Information Technology and Control","volume":"36 6","pages":""},"PeriodicalIF":1.1,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138946053","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-22DOI: 10.5755/j01.itc.52.4.34183
Zhiyuan Tan, Bin Chen, Liying Sun, Huimin Xu, Kun Zhang, Feng Chen
In order to improve visual recognition accuracy of pepper and provide reliable technical support for agricultural production, an improved YOLOv4 algorithm for pepper target recognition and detection was proposed in this paper. By adding Mosaic data enhancement and CBAM (Conventional block attention module) attention mechanism to the primitive character extraction network, the method enhanced the learning ability of the target detection algorithm, made the network effectively suppress the interference features, and increased the attention to effective features. To improve the accuracy of identification. The improved network model was trained, verified and tested on the self-made data set. The results showed that the proposed algorithm could effectively improve the accuracy of pepper recognition under natural light, and finally improved the mean Average Precision (mAP) of the existing YOLOv4 algorithm from 88.95% to 98.36%.
{"title":"Pepper Target Recognition and Detection Based on Improved YOLO v4","authors":"Zhiyuan Tan, Bin Chen, Liying Sun, Huimin Xu, Kun Zhang, Feng Chen","doi":"10.5755/j01.itc.52.4.34183","DOIUrl":"https://doi.org/10.5755/j01.itc.52.4.34183","url":null,"abstract":"In order to improve visual recognition accuracy of pepper and provide reliable technical support for agricultural production, an improved YOLOv4 algorithm for pepper target recognition and detection was proposed in this paper. By adding Mosaic data enhancement and CBAM (Conventional block attention module) attention mechanism to the primitive character extraction network, the method enhanced the learning ability of the target detection algorithm, made the network effectively suppress the interference features, and increased the attention to effective features. To improve the accuracy of identification. The improved network model was trained, verified and tested on the self-made data set. The results showed that the proposed algorithm could effectively improve the accuracy of pepper recognition under natural light, and finally improved the mean Average Precision (mAP) of the existing YOLOv4 algorithm from 88.95% to 98.36%.","PeriodicalId":54982,"journal":{"name":"Information Technology and Control","volume":"13 4","pages":""},"PeriodicalIF":1.1,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138947646","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：