DSpace collection: 會議論文
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/825
The collection's search engineSearch the Channels
https://tkuir.lib.tku.edu.tw/dspace/simple-search
Apply Machine-Learning Model for Clustering Rowing Players
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/125337
title: Apply Machine-Learning Model for Clustering Rowing Players abstract: Rowing, as a sport composed of single player or multiple players, performs body movements under certain rhythm with slight variation. The analysis of rhythm alternation or match is important on rowing research and merit our study. Therefore, this study analyzes the rowing movements by the following three procedures, rowing cycle segmentation, feature extraction, rowing cycle clustering. The rowing cycle segmentation procedure segments each player's video to videos of single cycle under the analysis of MediaPipe detected joint points. The feature extraction procedure calculates features from each rowing cycle by selecting amplitudes, angles, angular speeds of 4 selected joint points. At last, the rowing cycle clustering procedure analyzes all one-cycled videos using above features by different clustering and scoring methods. Three clustering methods, including K-means, Birch, and Gaussian-mixture, are experimented in this study for finding the most efficient one. A hybrid measurement from the Silhouette score, the Calinski-Harabasz index, and the Davies-Bouldin index, is proposed for finding the optimal clusters number. Experimental results of 15 players’ videos show that applying K-means clustering algorithm with the proposed hybrid measurement performs better for finding the rowing group.
<br>Automatic Anomaly Mark Detection on Fabric Production Video by Artificial Intelligence Techniques
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/125336
title: Automatic Anomaly Mark Detection on Fabric Production Video by Artificial Intelligence Techniques abstract: In the previous era, humans played important roles in all aspects of industrial work. However, they indisputably made many errors that can be mitigated by automated manufacturing, thus revealing the importance of the latter. In this paper, an autoencoder-based fabric-defect detection method via video is presented. The fabric-production video is segmented using frames to produce images, and then a VGG16-based autoencoder is applied to reconstruct the original image. In the proposed scheme, each fabric-production image is normalized to 256 x 256 pixels, which provided excellent results compared with using various margin sizes in our experiments. We used the structural similarity index (SSIM), which measures similarity when checking whether image regions are normal or defective. Moreover, a masking algorithm is utilized to improve detection accuracy. Based on our experiments, we found that 0.5 is an appropriate value for setting the SSIM threshold as it produced the best detection performance with a defect detection accuracy of ~99%.
<br>Infant Vomit Detection Using ShuffleNetV2
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124897
title: Infant Vomit Detection Using ShuffleNetV2 abstract: Infants often experience phenomena such as vomiting during the neonatal period. Typically, vomiting is when the food or milk in the baby's stomach comes out of the mouth after feeding, usually due to poor digestion or overeating. Therefore, this paper proposes a monitoring system to assist parents in being aware of whether their infant is vomiting even when they briefly divert their attention. The system consists of two stages: first, the utilization of YOLO5FACE for detecting the infant's face and capturing the baby's mouth region. Image processing techniques enhance color saturation in the captured mouth images. And then, ShuffleNetV2 extracts image features, and finally, frame differencing is utilized to determine if vomiting has occurred. Parents can be immediately notified of abnormal conditions by promptly detecting such incidents. The proposed method successfully and rapidly detects vomiting in videos with an accuracy of 98.30%.
<br>Air Pollution Source Tracing Framework: Leveraging Microsensors and Wind Analysis for Pollution Source Identification
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124896
title: Air Pollution Source Tracing Framework: Leveraging Microsensors and Wind Analysis for Pollution Source IdentificationReal-time hydroponic nutrient solution monitoring using a solid-state multi-ion sensing chip for an IoT application
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124831
title: Real-time hydroponic nutrient solution monitoring using a solid-state multi-ion sensing chip for an IoT application物聯網多離子監測系統應用於水耕作物營養素管理
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124827
title: 物聯網多離子監測系統應用於水耕作物營養素管理Air Pollution Source Tracing Framework: Leveraging Microsensors and Wind Analysis for Pollution Source Identification
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124825
title: Air Pollution Source Tracing Framework: Leveraging Microsensors and Wind Analysis for Pollution Source Identification abstract: In the face of rapid urbanization, air pollution has emerged as a pressing and pervasive concern. Urban areas experiencing robust development not
only contend with intrinsic air pollution sources like high traffic regions, densely inhabited zones, and industrial emissions but also grapple with the impact of external sources of pollution. The ramifications of air pollution extend beyond ecological disruptions, posing a formidable threat to human health. Prolonged exposure to contaminated air and airborne particulate matter in polluted environments exacerbates chronic ailments and elevates mortality risks.
Current methods for monitoring air pollution typically revolve around air quality indices and forecasting; however, they fall short in pinpointing pollution sources. Leveraging the widespread deployment of Smart City and Rural Air Quality Microsensors
across Taiwan, this study monitors areas affected by air pollution. By analyzing pollution group movement paths through continuous time series of air
pollution data and integrating wind factors, the study employs backtracking to identify the emission sources contributing to air pollution. Furthermore, innovative
air pollution corridors are formulated to assess the extent of pollution impact. Thus, when air pollution incidents arise, this method can unveil the pollution
sources at the specific location, elucidate propagation and movement paths during emission, and outline zones at risk of pollution in the near future.
This paper introduces the Air Pollution Source Tracing Problem (APSTP), proposing the APSTF (Air Pollution Source Tracing Framework) to address this
challenge. The APSTF encompasses three key phases: Identification, Matching and Backtracking, and Pathway Generation. It effectively identifies stations experiencing air pollution, correlates affected areas across different time slots, and predicts pollution impact areas, thereby shedding light on pollution dynamics and aiding in source identification and future pollution prediction. The APSTF stands
as a valuable tool for understanding and mitigating air pollution, leveraging data from air quality micro-sensors, meteorological stations, and advanced mathematical algorithms.
<br>以卷積與遞迴類神經網路為基礎之動態網路鏈結預測
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124801
title: 以卷積與遞迴類神經網路為基礎之動態網路鏈結預測An Augmented User Context-Awareness Recommendation System
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124800
title: An Augmented User Context-Awareness Recommendation System應用生成對抗網路於動態社群網路預測
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124799
title: 應用生成對抗網路於動態社群網路預測在低軌衛星與地面網路的整合環境中基於軟體定義與 QoS 導向之傳輸路徑選擇方法
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124798
title: 在低軌衛星與地面網路的整合環境中基於軟體定義與 QoS 導向之傳輸路徑選擇方法Cooperative Fall Detection with Multiple Cameras
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124797
title: Cooperative Fall Detection with Multiple Cameras abstract: We propose a fall detection mechanism based on multi-camera cooperation in home space. Cameras capture image-based falling events, and self-organize a group using deep reinforcement learning. Neighbor cameras exchange sensing data and statuses in local network proximity. With information sharing in a group, cameras can improve the accuracy of decision making on falling events and cope with the limited fields of view against physical deployment of cameras in residential areas.
<br>Enhanced Multipath QUIC Protocol with Lower Path Delay and Packet Loss Rate
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124796
title: Enhanced Multipath QUIC Protocol with Lower Path Delay and Packet Loss RateA novel social network prediction based on virtual knowledge distillation.
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124795
title: A novel social network prediction based on virtual knowledge distillation.Enhancements of visual secret sharing with quick response codes
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124794
title: Enhancements of visual secret sharing with quick response codes abstract: Visual secret sharing is the method for distributing a secret among a group by using images which can be seen by user groups. Each group holds a share, and with the sufficient amount of shares, the secret can be obtained. When the number of shares is below the threshold, secret would be unable to be acquired. We consider using the quick response (QR) codes, which can be captured by camera phones, to produce the shares for visual secret sharing. With the conventional method, reconstructed QR code can be captured and linked to the webpage while receiving insufficient number of shares, implying the leakage of secret. With our method, sufficient number of shares should be obtained to make the reconstructed QR code possible of delivering the information therein after capturing with the mobile phone camera. Hence, the security with visual secret sharing can be assured.
<br>Building a Reliable Trust Model in Wireless Sensor Networks
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124793
title: Building a Reliable Trust Model in Wireless Sensor NetworksA Novel RNN-Based Ensemble Model for Link Prediction on Dynamic Social Network
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124792
title: A Novel RNN-Based Ensemble Model for Link Prediction on Dynamic Social NetworkSimulation analysis of detecting pollution emissions from industrial areas by IoT mechanism
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124791
title: Simulation analysis of detecting pollution emissions from industrial areas by IoT mechanismError control for block-based compressed sensing with quadtree partition concepts
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124790
title: Error control for block-based compressed sensing with quadtree partition concepts abstract: Block-based compressed sensing has been a popular topic in image compression researches. A major portion of papers with this topic concentrates on compression performances based on rate-distortion theory. Here, we focus on the performances of reconstructed images with the transmission over lossless and lossy channels. Conventional block-based compressed sensing chooses fixed-sized square blocks to perform encoding. With quadtree partition that is based on the characteristics of image regions, we can separate square blocks with different sizes, and then perform compressed sensing independently. By fixing the measurement rate in both cases, we can observe and compare the performances with fixed-sized blocks and those with quadtree partition with the different percentages of block sizes. Better performances with quadtree partition can be observed in most cases.
<br>A Novel Few-Shot Learning with Meta-Gradient Memory
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124389
title: A Novel Few-Shot Learning with Meta-Gradient MemoryFast Detection of Fabric Defects based on Neural Networks
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124378
title: Fast Detection of Fabric Defects based on Neural NetworksConsiderations of Charging Vehicles and Environmental Charging in Sensor Deployment Problems
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124363
title: Considerations of Charging Vehicles and Environmental Charging in Sensor Deployment Problems心智圖法融入演算法大班教學以提升學習成效之研究
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124332
title: 心智圖法融入演算法大班教學以提升學習成效之研究Budget-Constrained Cost-Covering Job Assignment for a Total Contribution-Maximizing Platform
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/124082
title: Budget-Constrained Cost-Covering Job Assignment for a Total Contribution-Maximizing Platform abstract: We propose an optimization problem to model a situation when a platform with a limited budget wants to pay a group of workers to work on a set of jobs with possibly worker-job-dependent execution costs.
The platform needs to assign workers to jobs and at the same time decides how much to pay each worker to maximize the total “contribution” from the workers by using up the limited budget. The binary effort assignment problem, in which an effort from a worker is indivisible and can only be dedicated to a single job, is reminiscent of bipartite matching problems. Yet, a matched worker and job pair neither incurs cost nor enforces a compulsory effort in a standard matching setting while we consider such
cost to be covered by payment and certain level of effort to be made when a job is executed by a worker. The fractional effort assignment problem, in which generally a worker’s effort can be divisible and split among multiple jobs, bears a resemblance to a labor economy or online labor platform, and the platform needs to output an arrangement of efforts and the corresponding payments.
There are six settings in total to consider by combining the conditions on payments and efforts. Intuitively, we study how to come up with the best assignment under each setting and how different these assignments
under different settings can be in terms of the total contribution from workers when the information of each worker’s quality of service and cost is available. NP-completeness results and approximation algorithms are given for different settings. We then compare the solution quality of some settings in the end.
<br>On The Development of a Legal Penalty Prediction System for Drunk Driving Cases
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123464
title: On The Development of a Legal Penalty Prediction System for Drunk Driving Cases abstract: Recent years, computer-aided penalty prediction have been promoted to gain people's trust to the judicial systems, especially in developing Chinese region. In this paper, we propose machine learning based models to predict the legal penalty of criminal cases. Particularly, we focus on drunk driving cases as they are frequent, and the regulations are clear. Unlike western text which words are separated by spaces, words in Chinese text are continuum. In our proposed method, we first use a word segmentation method to separate the Chinese words in text and apply a pre-trained model to convert words into vectors. In the vector space, words with similar meanings have short distance with each other. As the amount of each penalty varies greatly, resulting a data imbalance problem. Therefore, we adapt the Synthetic Minority Oversampling Technique (SMOTE) algorithm as a solution. Finally, we apply deep learning-based models, including Bi-GRU and TextCNN to perform penalty prediction, and compare their advantages and disadvantages.In the experimental result, for drunk driving case penalty prediction, our propose SMOTE + TextCNN solution can reach 73.96% of accuracy. If we allow the prediction to be plus or minus one month from the actual, the accuracy is 95.60%. As for the computation time, our proposed method can predict the penalty of 1,524 drunk driving cases per second.
<br>Distilling One-Stage Object Detection Network via Decoupling Foreground and Background Features
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123463
title: Distilling One-Stage Object Detection Network via Decoupling Foreground and Background Features以卷積與遞迴類神經網路為基礎之動態網路鏈結預測
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123462
title: 以卷積與遞迴類神經網路為基礎之動態網路鏈結預測Simulation analysis of detecting pollution emissions from industrial areas by IoT mechanism
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123461
title: Simulation analysis of detecting pollution emissions from industrial areas by IoT mechanismMulti-Secret Image Sharing Scheme by Boolean Operations
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123460
title: Multi-Secret Image Sharing Scheme by Boolean Operations abstract: Multi-secret image sharing technology shares multiple secret images among participants. The proposed (k, n, m) multi-secret image sharing scheme shares m secret images among n participants and gathering k participants' shared images perfectly recovers these m secret images. The usage of sharing matrix based strategy acquires perfect reconstruction of the secret images under (k, n) thresholds. The proposed source-random- separate (SRS) method exhibits good sharing on multiple secret images. Experimental results show that the proposed scheme performs well on sharing efficiency and sharing security.
<br>Automatic Anomaly Mark Detection on Fabric Production Video by Artificial Intelligence Techniques
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123459
title: Automatic Anomaly Mark Detection on Fabric Production Video by Artificial Intelligence TechniquesConditionals support in binary expression tree based genetic programming
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123458
title: Conditionals support in binary expression tree based genetic programming abstract: Inspired by the genetic algorithm (GA), the genetic programming (GP) was proposed for searching a program that fits a certain behavior. There are many aspects that distinguish GP from GA a lot, though GP concepts were originating from GA. In this paper, we focus on the representation scheme for a GP program. A GP program contains both operators and operands. Without proper encoding, the GP crossover and mutation are likely to produce invalid programs. Based on our previous design experiences, we proposed an alternative approach. It is a binary expression tree based representation with conditional behavior of each node. Therefore, the scheme supports unary, binary, and ternary operators. It also reduce the probability of producing invalid programs. A feature of the scheme is that conditional operators are first-class member because each evaluation embeds conditional processing. A few image-processing experiments were conducted to show the effectiveness of the design. The experimental results are also discussed in this paper.
<br>The Design and Implementation of a Practical Meta-Heuristic for Detection and Identification of Denial-of-Service Attack Using Hybrid Approach
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123457
title: The Design and Implementation of a Practical Meta-Heuristic for Detection and Identification of Denial-of-Service Attack Using Hybrid ApproachAutomatic Defense through Fault Localization and Dynamic Patch Creation
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123456
title: Automatic Defense through Fault Localization and Dynamic Patch CreationA Multi-Agent Intelligence Hybrid System Technique for Detection and Defense of DDoS Attacks
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123455
title: A Multi-Agent Intelligence Hybrid System Technique for Detection and Defense of DDoS Attacks abstract: In the paper, we propose the distributed detection and identification multi-agent system (DDIMAS) framework that is the first attempt to apply in solving distributed denial of service (DDoS) problem. It includes three stages which are information heuristic rule, meta-heuristic algorithm and backward and forward search (BFS) rule, respectively. Moreover, the framework is a flexible architecture that can incorporate into other algorithms or rules to improve the overall performance. From the evaluation design, the experiment results show that our method is with higher detection rate and better accuracy than standard repositories. The proposed framework resolves issues in other swarm optimization algorithms and reveals that the performance of DDIMAS is better than existing methods and the adaptive meta-heuristic algorithm framework outperforms other methods for detecting DDoS attacks.
<br>Diagnosing SDN Network Problems by Using Spectrum-based Fault Localization Techniques
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123454
title: Diagnosing SDN Network Problems by Using Spectrum-based Fault Localization TechniquesSpatio-Temporal Patterns and Explanatory Factors of Urban Fire Occurrences in New Taipei
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123453
title: Spatio-Temporal Patterns and Explanatory Factors of Urban Fire Occurrences in New Taipei abstract: In modern metropolitan governance, fire-fighting represents a challenging task given the limited resources and the nature of disastrous impact on the citizen's psyche regarding the loss of both property and human lives. Fire-fighting and reduction of fire loss remain a challenging task for policy makers. The study of past fire incidences and factors empower policy makers for making resource allocation decision. In the wake of availability of large collections of both time-stamped and geocoded fire incident data, the exploration of any possible spatio-temporal patterns can help better allocation of limited fire resources. This paper employs spatio-temporal data analysis techniques such as spatial kernel density and spatial dependency index such as Ripley's K, Moran's I and Lee's L to explore socioeconomical factors and fire incidents. The results showed that in addition to population factors, the amount of electricity usage and ems demands are also relevant indicators for fire occurrences.
<br>Applying Blockchain on Supply Chain Management: Evidence from Scientometric Mapping
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123452
title: Applying Blockchain on Supply Chain Management: Evidence from Scientometric MappingExplore Methods to Generate Quality Input Data for Data-centric Industry Studies
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123451
title: Explore Methods to Generate Quality Input Data for Data-centric Industry StudiesQuadtree-based block compressed sensing for reliable transmission applications
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123450
title: Quadtree-based block compressed sensing for reliable transmission applications abstract: Reliable transmission of compressed media forms an inevitable part in digital communication systems. Compressed media are vulnerable to channel errors, which may cause severe degradation to reconstructed results. Hence error control techniques can work with compression schemes to help alleviate reconstructed image quality. We choose block compressed sensing for compression, and we expect the inherent characteristics of images may be helpful for enhancing reconstructed results. Quadtree decomposition can classify the image into smooth or active areas, which follows the inherent characteristics. After transmitting compressed media, at the decoder, we apply error protection and error regression schemes based on quadtree decomposition with the expectation of enhanced results. Simulations show that proposed method can help to enhance reconstructed image quality.
<br>DPCOVID: Privacy-Preserving federated covid-19 detection
https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/123449
title: DPCOVID: Privacy-Preserving federated covid-19 detection