检索结果-内蒙古大学图书馆

A Hybrid SARIMAX Model in Conjunction with Neural Networks for the Forecasting of Life Insurance Industry Growth in Thailand

A Hybrid SARIMAX Model in Conjunction with Neural Networks f...

引用

International ECTI Northern Section Conference on Electrical, Electronics, computer and Telecommunications engineering (ECTI-NCON)

作者： Supika Huadsri Sakorn Mekruksavanich Anuchit Jitpattanakul Wikanda Phaphan Department of Applied Statistics Faculty of Applied Science King Mongkut’s University of Technology North Bangkok Bangkok Thailand Department of Computer Engineering School of Information and Communication Technology University of Phayao Phayao Thailand Department of Mathematics Faculty of Applied Science King Mongkut’s University of Technology North Bangkok Bangkok Thailand Intelligent and Nonlinear Dynamic Innovations Research Center Research Group in Statistical Learning and Inference

This article is conducted with the primary aim of comparing various forecasting models to identify the optimal model for forecasting the growth of the life insurance industry in Thailand. The models under consideration include the SARIMAX model, the Recurrent Neural Network (RNN), and a hybrid model introduced as RNN-SARIMAX. The dataset utilized for this investigation is extensive, consisting of monthly direct premium data from the Thai life insurance sector spanning January 2003 to December 2022. The dataset is segregated into two distinct segments. Initially, data from January 2003 to December 2021 is employed for constructing the forecasting model. Subsequently, data from January 2022 to December 2022 is utilized for a comparative analysis of the forecasting models, to determine the most suitable and accurate model for predicting the growth of the life insurance industry in Thailand. The evaluation criteria for comparing models include the Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE), and the Coefficient of Determination (R-squared). The consistent findings from our analysis indicate that the SARIMAX model consistently outperforms both the RNN and RNN-SARIMAX models.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Cross-Modal Few-Shot Learning with Second-Order Neural Ordinary Differential Equations 39

Cross-Modal Few-Shot Learning with Second-Order Neural Ordin...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Zhang, Yi Cheng, Chun-Wun He, Junyi He, Zhihai Schönlieb, Carola-Bibiane Chen, Yuyan Aviles-Rivero, Angelica I. Harbin Institute of Technology China Department of Electrical and Electronic Engineering Southern University of Science and Technology China Department of Applied Mathematics and Theoretical Physics University of Cambridge United Kingdom Pengcheng Laboratory China Shanghai Key Laboratory of Data Science School of Computer Science Fudan University China Yau Mathematical Sciences Center Tsinghua University China

ISBN: (纸本)157735897X

We introduce SONO, a novel method leveraging Second-Order Neural Ordinary Differential Equations (Second-Order NODEs) to enhance cross-modal few-shot learning. By employing a simple yet effective architecture consisting of a Second-Order NODEs model paired with a cross-modal classifier, SONO addresses the significant challenge of overfitting, which is common in few-shot scenarios due to limited training examples. Our second-order approach can approximate a broader class of functions, enhancing the model's expressive power and feature generalization capabilities. We initialize our cross-modal classifier with text embeddings derived from class-relevant prompts, streamlining training efficiency by avoiding the need for frequent text encoder processing. Additionally, we utilize text-based image augmentation, exploiting CLIP's robust image-text correlation to enrich training data significantly. Extensive experiments across multiple datasets demonstrate that SONO outperforms existing state-of-the-art methods in few-shot learning performance. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Differential equations

来源：评论

学校读者我要写书评

暂无评论

Polygon relations and subadditivity of entropic measures for discrete and continuous multipartite entanglement

arXiv

引用

arXiv 2024年

作者： Liu, Lijun Ge, Xiaozhen Cheng, Shuming College of Mathematics and Computer Science Shanxi Normal University Linfen041000 China Shanghai Institute of Intelligent Science and Technology Tongji University Shanghai201804 China Department of Applied Mathematics The Hong Kong Polytechnic University Kowloon999077 Hong Kong The Department of Control Science and Engineering Tongji University Shanghai201804 China Institute for Advanced Study Tongji University Shanghai200092 China

In a recent work by us [Ge et al., Phys. Rev. A 110, L010402 (2024)], we have derived a series of polygon relations of bipartite entanglement measures that is useful to reveal entanglement properties of discrete, continuous, and even hybrid multipartite quantum systems. In this work, with the information-theoretical measures of Rényi and Tsallis entropies, we study the relationship between the polygon relation and the subadditivity of entropy. In particular, the entropy-polygon relations are derived for pure multi-qubit states and then generalized to multi-mode Gaussian states, by utilizing the known results from the quantum marginal problem. Then the equivalence between the polygon relation and subadditivity is established, in the sense that for all discrete or continuous multipartite states, the polygon relation holds if and only if the underlying entropy is subadditive. As a byproduct, the subadditivity of Rényi and Tsallis entropies is proven for all bipartite Gaussian states. Finally, the difference between polygon relations and monogamy relations is clarified, and generalizations of our results are discussed. Our work provides a better understanding of the rich structure of multipartite states, and hence is expected to be helpful for the study of multipartite entanglement. Copyright © 2024, The Authors. All rights reserved.

关键词： Geometry

来源：评论

学校读者我要写书评

暂无评论

A MgNO Method for Multiphase Flow in Porous Media

arXiv

引用

arXiv 2024年

作者： Liu, Xinliang Yang, Xia Zhang, Chen-Song Zhang, Lian Zhao, Li Computer Electrical and Mathematical Science and Engineering Division King Abdullah University of Science and Technology Thuwal Saudi Arabia Xiangtan University Hunan China LSEC ICMSEC Academy of Mathematics and System Sciences Chinese Academy of Sciences University of Chinese Academy of Sciences China Shenzhen International Center for Industrial and Applied Mathematics Shenzhen Research Institute of Big Data China

This research investigates the application of Multigrid Neural Operator (MgNO), a neural operator architecture inspired by multigrid methods, in the simulation for multiphase flow within porous media. The architecture is adjusted to manage a variety of crucial factors, such as permeability and porosity heterogeneity. The study extendes MgNO to time-dependent porous media flow problems and validate its accuracy in predicting essential aspects of multiphase flows. Furthermore, the research provides a detailed comparison between MgNO and Fourier Neural Opeartor (FNO), which is one of the most popular neural operator methods, on their performance regarding prediction error accumulation over time. This aspect provides valuable insights into the models' long-term predictive stability and reliability. The study demonstrates MgNO's capability to effectively simulate multiphase flow problems, offering considerable time savings compared to traditional simulation methods, marking an advancement in integrating data-driven methodologies in geoscience applications. © 2024, CC BY.

关键词： Porous materials

来源：评论

学校读者我要写书评

暂无评论

Multimodal brain tumor segmentation and classification from MRI scans based on optimized DeepLabV3+ and interpreted networks information fusion empowered with explainable AI

引用

computers in Biology and Medicine 2024年 182卷 109183-109183页

作者： Ullah, Muhammad Sami Khan, Muhammad Attique Albarakati, Hussain Mubarak Damaševičius, Robertas Alsenan, Shrooq Department of Computer Science HITEC University Taxila47080 Pakistan Department of Artificial Intelligence College of Computer Engineering and Science Prince Mohammad Bin Fahd University Al Khobar Saudi Arabia Computer and Network Engineering Department College of Computing Umm Al-Qura University Makkah24382 Saudi Arabia Faculty of Applied Mathematics Silesian University of Technology Gliwice44-100 Poland Information Systems Department College of Computer and Information Sciences Princess Nourah Bint Abdulrahman University P.O. Box 84428 Riyadh11671 Saudi Arabia

Explainable artificial intelligence (XAI) aims to offer machine learning (ML) methods that enable people to comprehend, properly trust, and create more explainable models. In medical imaging, XAI has been adopted to interpret deep learning black box models to demonstrate the trustworthiness of machine decisions and predictions. In this work, we proposed a deep learning and explainable AI-based framework for segmenting and classifying brain tumors. The proposed framework consists of two parts. The first part, encoder-decoder-based DeepLabv3+ architecture, is implemented with Bayesian Optimization (BO) based hyperparameter initialization. The different scales are performed, and features are extracted through the Atrous Spatial Pyramid Pooling (ASPP) technique. The extracted features are passed to the output layer for tumor segmentation. In the second part of the proposed framework, two customized models have been proposed named Inverted Residual Bottleneck 96 layers (IRB-96) and Inverted Residual Bottleneck Self-Attention (IRB-Self). Both models are trained on the selected brain tumor datasets and extracted features from the global average pooling and self-attention layers. Features are fused using a serial approach, and classification is performed. The BO-based hyperparameters optimization of the neural network classifiers is performed and the classification results have been optimized. An XAI method named LIME is implemented to check the interpretability of the proposed models. The experimental process of the proposed framework was performed on the Figshare dataset, and an average segmentation accuracy of 92.68 % and classification accuracy of 95.42 % were obtained, respectively. Compared with state-of-the-art techniques, the proposed framework shows improved accuracy. © 2024 Elsevier Ltd

关键词： Data fusion

来源：评论

学校读者我要写书评

暂无评论

A Sensor-Based Deep Learning Approach for Recognizing Daily and Work Activities in Open Environments for Sanitation Workers

A Sensor-Based Deep Learning Approach for Recognizing Daily ...

引用

International ECTI Northern Section Conference on Electrical, Electronics, computer and Telecommunications engineering (ECTI-NCON)

作者： Sakorn Mekruksavanich Ponnipa Jantawong Wikanda Phaphan Anuchit Jitpattanakul Department of Computer Engineering School of Information and Communication Technology University of Phayao Phayao Thailand Department of Applied Statistics Faculty of Applied Science Research Group in Statistical Learning and Inference King Mongkut’s University of Technology North Bangkok Bangkok Thailand Department of Mathematics Faculty of Applied Science Intelligent and Nonlinear Dynamic Innovations Research Center King Mongkut’s University of Technology North Bangkok Bangkok Thailand

Sanitation workers play a crucial role in maintaining public hygiene and cleanliness. Understanding their daily routines and job responsibilities is essential for improving their work environment, task distribution, and overall supervision. However, tracking sanitation workers in open environments remains a persistent challenge. This paper introduces a deep learning model named CNN-BiGRU-CBAM, designed to recognize sanitation workers’ daily and working activities using smartwatch sensors. The model leverages advanced convolutional neural networks to extract highly informative features from multi-modal sensor data, including acceleration information, collected from smartwatches. The dataset used in this study is carefully curated and collected from sanitation workers during their daily work routines, encompassing activities such as walking, running, sweeping, lifting, and driving vehicles. These activities are systematically categorized to cover both work-related tasks and non-work activities. Experimental results demonstrate that the proposed model consistently achieves an average F1-score of over 94.50% across seven activity groups, outperforming baseline deep learning methods by an impressive margin of 5.62%. The model excels in accurately identifying operations in diverse and dynamic work settings. This research highlights the potential of smartwatches and deep learning in continuously analyzing sanitation workers’ duties and developing improved working regulations that consider their well-being and daily tasks.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Provably Convergent Plug-and-Play Quasi-Newton Methods

arXiv

引用

arXiv 2023年

作者： Tan, Hong Ye Mukherjee, Subhadip Tang, Junqi Schönlieb, Carola-Bibiane Department of Applied Mathematics and Theoretical Physics University of Cambridge United Kingdom Department of Computer Science University of Bath United Kingdom School of Mathematics University of Birmingham United Kingdom Department of Electronics and Electrical Communication Engineering Indian Institute of Technology Kharagpur India

Plug-and-Play (PnP) methods are a class of efficient iterative methods that aim to combine data fidelity terms and deep denoisers using classical optimization algorithms, such as ISTA or ADMM, with applications in inverse problems and imaging. Provable PnP methods are a subclass of PnP methods with convergence guarantees, such as fixed point convergence or convergence to critical points of some energy function. Many existing provable PnP methods impose heavy restrictions on the denoiser or fidelity function, such as nonexpansiveness or strict convexity, respectively. In this work, we propose a novel algorithmic approach incorporating quasi-Newton steps into a provable PnP framework based on proximal denoisers, resulting in greatly accelerated convergence while retaining light assumptions on the denoiser. By characterizing the denoiser as the proximal operator of a weakly convex function, we show that the fixed points of the proposed quasi-Newton PnP algorithm are critical points of a weakly convex function. Numerical experiments on image deblurring and super-resolution demonstrate 2-8x faster convergence as compared to other provable PnP methods with similar reconstruction *** Codes 49M15, 49J52, 65K15 Copyright © 2023, The Authors. All rights reserved.

关键词： Image reconstruction

来源：评论

学校读者我要写书评

暂无评论

Enhancing Sensor-based Human Activity Recognition Using Hybrid Deep Learning and Data Augmentation

Enhancing Sensor-based Human Activity Recognition Using Hybr...

引用

Research, Invention, and Innovation Congress (RI2C)

作者： Sakorn Mekruksavanich Wikanda Phaphan Anuchit Jitpattanakul Department of Computer Engineering School of Information and Communication Technology University of Phayao Phayao Thailand Department of Applied Statistics Research Group in Statistical Learning and Inference Faculty of Applied Science King Mongkut's University of Technology North Bangkok Bangkok Thailand Department of Mathematics Intelligent and Nonlinear Dynamic Innovations Research Center Faculty of Applied Science King Mongkut's University of Technology North Bangkok Bangkok Thailand

ISBN: (数字)9798331506018

ISBN: (纸本)9798331506025

Human Activity Recognition (HAR) is essential in various applications, including wellness tracking, automated residences, and fitness monitoring. In the past few decades, sensor-based HAR has become increasingly popular due to advancements in technologies for sensors. Nevertheless, HAR networks' effectiveness dramatically depends on the caliber and volume of the training data, which is frequently restricted and unevenly distributed. This research introduces a novel deep learning method called Multihead-CNN-BiGRU, which integrates one-dimensional convolutional neural networks with bidirectional gated recurrent units (BiGRU) to improve the accuracy of sensor-based HAR. To tackle the problem of insufficient a nd uneven training data, we utilize the synthetic minority over-sampling technique (SMOTE) to augment the data. The suggested model is assessed using the publicly accessible WISDM dataset, which comprises sensor data from diverse human actions. The ID-CNN is employed for extracting localized characteristics from the sensor data, while the BiGRU gathers temporal dependencies and contextual information. The hybrid architecture allows the model to acquire spatial and temporal patterns efficiently. In addition, the SMOTE technique is utilized to create artificial samples of the underrepresented classes, thus equalizing the distribution of classes and enhancing the model's capacity to generalize. The experimental findings show that our hybrid strategy provides exceptional outcomes when paired with SMOTE data augmentation. It obtains the most excellent accuracy of 99.51 % and the highest F1-score of 99.49% compared to the most advanced approaches. The suggested framework provides a reliable and precise solution for sensor-based HAR, which opens up opportunities for improved applications in other fields.

关键词： Deep learning Technological innovation Accuracy Training data Sensor phenomena and characterization Logic gates Data augmentation Human activity recognition Reliability Monitoring

来源：评论

学校读者我要写书评

暂无评论

Leveraging Residual Deep Neural Networks and Multi-Device Sensors for Heterogeneous Activity Recognition

Leveraging Residual Deep Neural Networks and Multi-Device Se...

引用

International Conference on Telecommunications and Signal Processing (TSP)

作者： Sakorn Mekruksavanich Wikanda Phaphan Anuchit Jitpattanakul Department of Computer Engineering School of Information and Communication Technology University of Phayao Phayao Thailand Department of Applied Statistics Faculty of Applied Science Research Group in Statistical Learning and Inference King Mongkut's University of Technology North Bangkok Bangkok Thailand Department of Mathematics Faculty of Applied Science Intelligent and Nonlinear Dynamic Innovations Research Center King Mongkut's University of Technology North Bangkok Bangkok Thailand

ISBN: (数字)9798350365597

ISBN: (纸本)9798350365603

This study introduces a novel approach to identifying human activities using wearable sensors, particularly smart-phones and smartwatches. By leveraging deep learning neural networks and data from the HHAR dataset, which includes accelerometer and gyroscope data from individuals engaged in various activities, our method, centered around the HAR-Res NeXt model, accurately detects six activities. Utilizing residual connections and multi-kernel blocks, our approach effectively captures temporal and spatial relationships in sensor data. Experimental results demonstrate superior performance to standard machine learning algorithms and other deep learning approaches for human activity recognition. HAR-ResNeXt achieves high accuracy rates, particularly in classifying smartphone sensor data, underscoring its adaptability across diverse scenarios. Comparative analysis reveals the effectiveness of smartphone sensors and emphasizes the importance of multi-modal sensor fusion for accurate activity detection.

关键词： Deep learning Accuracy Machine learning algorithms Multimodal sensors Signal processing Telecommunications Gyroscopes

来源：评论

学校读者我要写书评

暂无评论

Deep Learning Networks for Human Knee Abnormality Detection Based on Surface EMG Signals

Deep Learning Networks for Human Knee Abnormality Detection ...

引用

International Conference on Telecommunications and Signal Processing (TSP)

作者： Sakorn Mekruksavanich Wikanda Phaphan Anuchit Jitpattanakul Department of Computer Engineering School of Information and Communication Technology University of Phayao Phayao Thailand Department of Applied Statistics Faculty of Applied Science Research Group in Statistical Learning and Inference King Mongkut's University of Technology North Bangkok Bangkok Thailand Department of Mathematics Faculty of Applied Science Intelligent and Nonlinear Dynamic Innovations Research Center King Mongkut's University of Technology North Bangkok Bangkok Thailand

ISBN: (数字)9798350365597

ISBN: (纸本)9798350365603

Early knee problem management relies on precise identification and classification of abnormalities. Surface electromyography (sEMG) and goniometer signals offer non-invasive screening for muscle activity and joint angle patterns, yet their complexity poses challenges in extracting critical diagnostic information. This paper proposes a novel deep-learning method using sEMG and goniometer data for knee abnormality diagnosis. The proposed ResNeXt model, employing CNNs and multi-kernel modules, is evaluated on the UCIEMG dataset. Experimental results demonstrate ResNeXt's superior accuracy, precision, recall, and F1-score compared to baseline models (CNN and LSTM). Res NeXt achieves the best performance with combined EMG and goniometer data, reaching 96.37% accuracy and 93.77% F1-score, with fewer trainable parameters, indicating computational efficiency. The findings indicate ResNeXt's effectiveness in identifying knee abnormalities using biosensor data, particularly sEM G and goniometer signals, aiding early disease detection and treatment.

关键词： Deep learning Accuracy Goniometers Signal processing Muscles Electromyography Telecommunications Surface treatment Long short term memory Diseases

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：