This paper proposes a method for expanding the metadata of three-dimensional point cloud data using Large Language Models (LLMs). Currently, point cloud data plays a crucial role in various fields such as autonomous d...
详细信息
ISBN:
(数字)9791188428137
ISBN:
(纸本)9798331507602
This paper proposes a method for expanding the metadata of three-dimensional point cloud data using Large Language Models (LLMs). Currently, point cloud data plays a crucial role in various fields such as autonomous driving and medical image reconstruction, necessitating the expansion of metadata for efficient processing. Traditionally, metadata construction has relied on manual input, which is prone to errors. In this study, we propose a method that utilizes LLMs, particularly the Llama 3.1 model, to extract the center points of each class in the point cloud data and expand the metadata by adding these center points to the annotation files. By using center points, computational costs are reduced, and the performance of segmentation and detection models based on this data is improved.
Visual place recognition techniques based on deep learning, which have imposed themselves as the state-of-the-art in recent years, do not generalize well to environments visually different from the training set. Thus,...
详细信息
Epilepsy is a prevalent neurological disorder and has been studied through the analysis of Electroencephalogram (EEG) signals. However, the identification and classification of epileptic seizure patterns remains chall...
Epilepsy is a prevalent neurological disorder and has been studied through the analysis of Electroencephalogram (EEG) signals. However, the identification and classification of epileptic seizure patterns remains challenging due to the non-stationary nature of EEG signals and the presence of artifacts. In this paper, we investigate the applicability of a transformer-based deep learning model to classify seizure patterns observed in epileptic patients. We employed the self-attention mechanism inherent in transformers to capture complex temporal relationships in the EEG recordings. By prepossessing the EEG signals into suitable input sequences and adapting the transformer architecture, we achieved 78.11% in distinguishing between different epileptic seizure patterns. Our findings indicate that the transformer model, with its ability to manage long-range dependencies, offers a robust approach to EEG-based seizure pattern classification. This work is important for building advanced automated diagnostic tools for epilepsy and related neurological disorders.
Broadcasting is an information dissemination problem in a connected graph in which one vertex, called the originator, must distribute a message to all other vertices by placing a series of calls along the edges of the...
详细信息
ISBN:
(数字)9781665488020
ISBN:
(纸本)9781665488037
Broadcasting is an information dissemination problem in a connected graph in which one vertex, called the originator, must distribute a message to all other vertices by placing a series of calls along the edges of the graph. Every time the informed vertices aid the originator in distributing the message. Finding the broadcast time of any vertex in an arbitrary graph is NP-complete. We designed an efficient heuristic, which improves the results of existing heuristics in most cases. Extensive simulations show that our new heuristic outperforms the existing ones for most of the commonly used interconnection networks in some network models generated by network simulator ns-2.
The increasing complexity and memory demands of Deep Neural Networks (DNNs) for real-time systems pose new significant challenges, one of which is the GPU memory capacity bottleneck, where the limited physical memory ...
详细信息
ISBN:
(数字)9798350358414
ISBN:
(纸本)9798350358421
The increasing complexity and memory demands of Deep Neural Networks (DNNs) for real-time systems pose new significant challenges, one of which is the GPU memory capacity bottleneck, where the limited physical memory inside GPUs impedes the deployment of sophisticated DNN models. This paper presents, to the best of our knowledge, the first study of addressing the GPU memory bottleneck issues, while simultaneously ensuring the timely inference of multiple DNN tasks. We propose RT-Swap, a real-time memory management framework, that enables transparent and efficient swap scheduling of memory objects, employing the relatively larger CPU memory to extend the available GPU memory capacity, without compromising timing guarantees. We have implemented RT-Swap on top of representative machine-learning frameworks, demonstrating its effectiveness in making significantly more DNN task sets schedulable at least 72% over existing approaches even when the task sets demand up to 96.2% more memory than the GPU's physical capacity.
As social networking services and e-commerce are growing rapidly, the number of online users also dynamically growing which facilitates the contribution of huge content to the digital world. In such a dynamic environm...
详细信息
Credit card use is becoming more and more commonplace every day. Financial organizations and credit card customers lose a lot of money because of complicated illegal transactions. Fraudsters constantly stay on top of ...
详细信息
Credit card use is becoming more and more commonplace every day. Financial organizations and credit card customers lose a lot of money because of complicated illegal transactions. Fraudsters constantly stay on top of new technology to quickly perpetrate fraud against customer transaction patterns. We analyze credit card transaction networks and identify suspicious patterns, such as transactions connected to multiple accounts or unusual transaction patterns, transactions made at unusual times, and to monitor credit card transactions in real-time and quickly identify suspicious transactions. TigerGraph is used to analyze data, display results on a dashboard, and send notifications via email. One meth’\ Vc 1``13-od commonly used in anomaly detection is to compare data values against the standard deviation. In this research, we explain the use of TigerGraph as a platform for anomaly detection above the standard deviation, as well as the use of the Louvain algorithm in finding merchant communities used by fraudsters. The data used in this study comes from Sparkov simulation data obtained from Kaggle. Our results show that by using TigerGraph, we managed to achieve a very high accuracy rate of 99.77%, precision 82.84%, recall 72.38%, and f1-score 77,26% in predicting transaction fraud on Sparkov simulation data. This is much better than the results reported in a paper that uses the supervised machine learning method with the AdaBoost algorithm which achieves the highest accuracy of 77%.
Person Re-Identification (ReID) is a crucial task in computer vision that aims to match persons captured from non-overlapping camera views. In this paper, to alleviate the impacts of background and occlusion, we propo...
Person Re-Identification (ReID) is a crucial task in computer vision that aims to match persons captured from non-overlapping camera views. In this paper, to alleviate the impacts of background and occlusion, we propose to use instance segmentation and pose estimation methods to create masks for global feature extraction. Furthermore, we divide the pedestrian images into three regions according to pedestrian keypoints, trying to eliminate the alignment errors. This part-based matching strategy also helps to address the occlusion issue. Overall, we construct a deep learning network with three branches, including two global branches and a part-based branch. The two global branches extract global features using segmentation-based mask and the mask derived from pedestrian keypoint heatmaps, respectively. In the end, a weighted fusion strategy is used to combine the global scores and part-based scores for final classification. This network enables us to acquire robust global feature of pedestrians by excluding background and occlusion, and simultaneously address the alignment errors to some extent. Experimental results on these three widely used datasets demonstrate the effectiveness of our method: Specifically, it achieves 64.5% rank-1 accuracy and 54.3% mAP on Occluded-Duke, and 94.4% and 87.1% rank-1 accuracies on Market-1501 and DukeMTMC-reID, respectively.
The cases of dengue hemorrhagic fever (DHF) in Indonesia have increased significantly since 2020. Data shown by the Central Statistics Agency, for example, in South Sumatra Province, there were 6,348 cases of DHF duri...
详细信息
Loans generate the majority of revenue forbanks and other financial institutions. Loan approval is a critical process in banking organizations because they can only lend to specific people or organizations due to rest...
详细信息
ISBN:
(数字)9798350379990
ISBN:
(纸本)9798350391558
Loans generate the majority of revenue forbanks and other financial institutions. Loan approval is a critical process in banking organizations because they can only lend to specific people or organizations due to restricted resources or credit. In order to forecastif a specific individual is eligible for a loan or not, a variety of machine learning techniques are combined with algorithms such as bagging classifiers. These algorithms include logistic regression classifiers, support vector classifiers (SVC), decision trees, and random forest. This system aims to enhance the accuracy and robustness of bank loan approvalpredictions through the implementation of an ensemble machine learning model with 97% accuracy. Leveraging the strengths of Random Forest (RF) and XGBoost (eXtreme Gradient Boosting), this ensemble approach seeks to mitigate the limitations of individual models, providing a more reliable and precise decision-making system for loan approval processes. However, these systems encounter challenges like managing imbalanced datasets and ensuring model interpretability. Furthermore, integrating diverse data sources while maintaining data quality and consistency presents significant difficulties.
暂无评论