检索结果-内蒙古大学图书馆

Sample Efficient Deep Reinforcement Learning With Online State Abstraction and Causal Transformer Model prediction

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2024年第11期35卷 16574-16588页

作者： Lan, Yixing Xu, Xin Fang, Qiang Hao, Jianye Natl Univ Def Technol Coll Intelligence Sci & Technol Changsha 410073 Peoples R China Tianjin Univ Coll Intelligence & Comp Tianjin 300372 Peoples R China

Deep reinforcement learning (RL) typically requires a tremendous number of training samples, which are not practical in many applications. State abstraction and world models are two promising approaches for improving sample efficiency in deep RL. However, both state abstraction and world models may degrade the learning performance. In this article, we propose an abstracted model-based policy learning (AMPL) algorithm, which improves the sample efficiency of deep RL. In AMPL, a novel state abstraction method via multistep bisimulation is first developed to learn task-related latent state spaces. Hence, the original Markov decision processes (MDPs) are compressed into abstracted MDPs. Then, a causal transformer model predictor (CTMP) is designed to approximate the abstracted MDPs and generate long-horizon simulated trajectories with a smaller multistep prediction error. Policies are efficiently learned through these trajectories within the abstracted MDPs via a modified multistep soft actor-critic algorithm with a lambda-target. Moreover, theoretical analysis shows that the AMPL algorithm can improve sample efficiency during the training process. On Atari games and the DeepMind Control (DMControl) suite, AMPL surpasses current state-of-the-art deep RL algorithms in terms of sample efficiency. Furthermore, DMControl tasks with moving noises are conducted, and the results demonstrate that AMPL is robust to task-irrelevant observational distractors and significantly outperforms the existing approaches.

关键词： Transformers Task analysis Predictive models Heuristic algorithms prediction algorithms Computational modeling Training Causal transformer feature representation model-based reinforcement learning (MBRL) state abstraction

来源：评论

学校读者我要写书评

暂无评论

A Novel Hybrid Model Based on Sentiment Analysis Using Community Guided Link prediction for Connecting People With Similar Interest

引用

IEEE ACCESS 2024年 12卷 139435-139455页

作者： Luthra, Snigdha Kumar, Rakesh Gupta, Meenu Chandigarh Univ Dept Comp Sci & Engn Mohali 140413 Punjab India

In this study, a novel hybrid recommendation model is designed for popular social media platforms such as Facebook, Twitter, WhatsApp, and Instagram. Recognizing the global significance of these platforms, the model aims to facilitate connections among users with similar interests, leveraging community detection algorithms. By integrating sentimental analysis, link prediction, and community detection techniques, a comprehensive network is constructed that enhances user experience and addresses the escalating issue of cyberbullying in the era of increasing connectivity. The proposed model strategically combines nodes and edges to form communities tailored to individual user interests. Social media comments, gathered randomly from the Facebook API, undergo sentimental analysis to evaluate the model's performance. The innovative Community-guided Link prediction algorithm using Sentimental Analysis (CLPSA) for multilayer networks is employed, aiding in the identification of nodes associated with fake comments and thereby mitigating cyberbullying problems. Integrating sentiment analysis enhances understanding of human emotions, and users in the same cluster are more likely to connect in the future, promoting positive engagement. The model's novelty lies in its utilization of the Proposed Cluster Edge Betweenness algorithm, which identifies clusters of users with denser connections among them while being loosely connected to other separate modules. This approach helps create a network structure that fosters positive interactions and reduces the occurrence of intolerable comments. The experimental comparison of community detection algorithms underscores the effectiveness of our approach, particularly the superior performance of the Cluster Edge Betweenness algorithm. It achieves high accuracy in identifying user interests and forming cohesive clusters, essential for enhancing user engagement and community cohesion within social media networks. The qualitative analysis of c

关键词： Cyberbullying prediction algorithms Clustering algorithms Blogs Detection algorithms Analytical models Semantics Sentiment analysis Community detection social media networks (SMNs) sentimental analysis cyberbullying link prediction

来源：评论

学校读者我要写书评

暂无评论

prediction Window Selection in FCS-MPC for Two-Level VSI Applications

引用

IEEE TRANSACTIONS ON POWER ELECTRONICS 2024年第3期39卷 3135-3143页

作者： Zafra, Eduardo Vazquez, Sergio Dragicevic, Tomislav Alcaide, Abraham M. Leon, Jose I. Franquelo, Leopoldo G. Univ Seville Elect Dept Seville 41092 Spain Univ Seville Lab Engn Energy & Environm Sustainabil Seville 41092 Spain DTU Dept Wind & Energy Syst DK-2800 Kongens Lyngby Denmark

Semiconductors in power converters are operated in switched-mode. For a given sampling frequency, a digital controller computes the next actuation according to several objectives. Typically one desires to regulate current or voltage values with minimal error and harmonic distortion. Higher switching frequency of the semiconductor is helpful as the granularity of the problem is improved and the harmonic spectrum is shifted toward high-order harmonics that are easier to filter with passive elements. However, the switching losses increase. Finite control set model predictive control (FCS-MPC) is proven to be an excellent control method with a great ability to maximize harmonic quality in terms of the switching frequency, particularly when the controller is configured to consider long prediction horizons. Nonetheless, in order to avoid computational overruns, the prediction horizon and the sampling frequency cannot be indefinitely increased. In this article, the tradeoff between these two parameters is studied, providing guidelines on FCS-MPC tuning for different configurations.

关键词： Control systems Predictive models Optimization Switching frequency prediction algorithms Switches Regulation Digital control model predictive control (MPC) three-phase dc-ac power inverters

来源：评论

学校读者我要写书评

暂无评论

Lightweight Fault prediction Method for Edge Networks

引用

IEEE INTERNET OF THINGS JOURNAL 2024年第6期11卷 11117-11136页

作者： Yang, Yang Mei, Jingting Zhang, Zhenwei Long, Yuhan Liu, Aolun Gao, Zhipeng Rui, Lanlan Beijing Univ Posts & Telecommun State Key Lab Network & Switching Technol Beijing 100876 Peoples R China Beijing Univ Posts & Telecommun Beijing 100876 Peoples R China Beijing Univ Posts & Telecommun State Key Lab Networking & Switching Beijing 100876 Peoples R China Beijing Univ Posts & Telecommun Network Management Ctr Beijing Peoples R China

The occurrence of faults increases in edge networks as the service types and component architecture become increasingly complex. Traditional centralized cloud fault prediction technology cannot be directly applied to edge network systems that are typically required to handle real-time data due to their high-computational complexity. Therefore, this article proposes a lightweight fault prediction algorithm for edge networks that realizes fault prediction with cross-layer cooperation. First, for edge devices with limited computation resources, the time feature lightweight extraction method of brain neurology fusion long short term memory (LSTM) based on scene reappearance mechanisms is proposed, which solves the single-step dependency problem of neurons and improves the accuracy. And the LSTM neuron connection method based on pulse dynamics is designed. This method prunes the structure of the LSTM network based on relevant knowledge of biological neurology to realize a lightweight time feature extraction model of fault information. Then, on the edge server side, a spatial feature lightweight extraction method based on a two-way residual structure is proposed. This method uses decomposition convolution to reduce the number of network parameters and realize a lightweight model. Finally, the spatio-temporal correlation features of the extracted fault information are spliced to realize fault prediction. To verify the effectiveness of the proposed model, we compare the improved model with the existing fault prediction model. The experiments show that the algorithm proposed in this article has higher accuracy, lower complexity and lower memory requirements. Therefore, it has high-deployment potential in edge network scenarios.

关键词： Feature extraction Servers Predictive models prediction algorithms Telecommunications Probability Neurons Edge networks fault prediction feature extraction spatio-temporal correlation

来源：评论

学校读者我要写书评

暂无评论

A Deep Learning Based Expert Framework for Portfolio prediction and Forecasting

引用

IEEE ACCESS 2024年 12卷 103810-103829页

作者： Jeribi, Fathe Martin, R. John Mittal, Ruchi Jari, Hassan Alhazmi, Abdulrahman Hassan Malik, Varun Swapna, S. L. Goyal, S. B. Kumar, Manoj Singh, Shubhranshu Vikram Jazan Univ Coll Engn & Comp Sci Jazan 45142 Saudi Arabia Chitkara Univ Chitkara Inst Engn & Technol Rajpura 140401 Punjab India Applexus Technol Technopk Thiruvananthapuram 695583 Kerala India City Univ Malaysia Fac Informat Technol Petaling Jaya 46100 Selangor Malaysia Univ Wollongong Dubai Sch Comp Sci FEIS Dubai Knowledge Pk Dubai U Arab Emirates Middle East Univ MEU Res Unit Amman 11831 Jordan Amity Univ Noida 201301 Uttar Pradesh India

Stock market forecasting involves predicting fluctuations and trends in the value of financial assets, utilizing statistical and machine learning models to analyze historical market data for insights into future behavior. This practice aids investors, traders, financial institutions, and governments in making informed decisions, managing risks, and assessing economic conditions. Forecasting financial markets is difficult due to the intricate interplay of global economics, politics, and investor sentiment, making it inherently unpredictable. This study introduces a Deep Learning based Expert Framework for Stock Market forecasting (Portfolio prediction) called DLEF-SM. The methodology begins with an improved jellyfish-induced filtering (IJF-F) technique for preprocessing, effectively analyzing raw data and eliminating artifacts. To address imbalanced data and enhance data quality, pre-trained convolutional neural network (CNN) architectures, VGGFace2 and ResNet-50, are used for feature extraction. Additionally, an improved black widow optimization (IBWO) algorithm is designed for feature selection, reducing data dimensionality and preventing under-fitting. For precise stock market predictions, integrate deep reinforcement learning with artificial neural network (DRL-ANN) is proposed. Simulation outcomes reveal that the proposed framework achieves maximum forecasting accuracy, reaching 99.562%, 98.235%, and 98.825% for S&P500-S, S&P500-L, and DAX markets, respectively.

关键词： Forecasting Stock markets Predictive models Data models Feature extraction Accuracy prediction algorithms Portfolios Deep learning Feature detection Financial management Stock market predictive analytics portfolio management deep learning feature optimization

来源：评论

学校读者我要写书评

暂无评论

Proximal Policy-Optimized Regularized Least Squares Algorithm for Noise-Resilient Motion prediction of UMVs

引用

IEEE JOURNAL OF OCEANIC ENGINEERING 2024年第4期49卷 1397-1410页

作者： Zhong, Yiming Yu, Caoyang Xiang, Xianbo Lian, Lian Shanghai Jiao Tong Univ Sch Oceanog Shanghai 200030 Peoples R China Shanghai Jiao Tong Univ Key Lab Polar Ecosyst & Climate Change Minist Educ Shanghai 200030 Peoples R China Huazhong Univ Sci & Technol Sch Naval Architecture & Ocean Engn Wuhan 430074 Peoples R China

To enhance the accuracy of motion prediction in unmanned marine vehicles (UMVs), an innovative proximal policy-optimized regularized least squares (PPO-RLS) algorithm is proposed in this article. This article begins by developing a dynamics model for UMVs that incorporates viscous damping and external forces to minimize modeling errors. However, this model does not account for data noise, making accurate parameter identification difficult when using traditional least squares (LS) algorithms. To overcome this limitation, the PPO-RLS algorithm is proposed, incorporating a regularization term within the LS framework and utilizing proximal policy optimization for adaptive regularization term tuning. The performance of the PPO-RLS algorithm is thoroughly evaluated using both simulation data and lake trial data, demonstrating significant improvements over both the traditional LS algorithm and a state-of-the-art algorithm. Specifically, in simulation tests, the PPO-RLS algorithm achieves a notable reduction in root mean square error for surge velocity (8.49E-03 m/s) and heading angle (2.32$<^>\circ$), markedly outperforming the LS algorithm (2.36E-02 m/s for surge velocity and 4.14$<^>\circ$ for heading angle). In addition, the PPO-RLS algorithm displays enhanced stability, as indicated by a more than 50% reduction in condition number (1.46E+04 for PPO-RLS versus 2.89E+06 for LS). These improvements are further validated by lake trial data, confirming the algorithm's advanced motion prediction capabilities with quantitatively lower errors and greater robustness.

关键词： prediction algorithms Optimization Support vector machines Accuracy Vehicle dynamics Mathematical models Predictive models Data noise motion prediction proximal policy optimization (PPO) regularized least squares (RLS) algorithm unmanned marine vehicles (UMVs)

来源：评论

学校读者我要写书评

暂无评论

Compressing Vehicle Trajectory Data Using Hybrid Coding With Kinematic Motion prediction

引用

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2024年第6期25卷 5389-5401页

作者： Xu, Lipeng Wu, Zhizhou Wang, Yinhai Tang, Jinjun Liang, Yunyi Xinjiang Univ Sch Mech Engn Urumqi 830049 Peoples R China Univ Washington Dept Civil & Environm Engn Seattle WA 98195 USA Cent South Univ Sch Transportat Engn Changsha 410017 Peoples R China Cent South Univ Sch Traff & Transportat Engn Changsha Peoples R China Univ Shanghai Sci & Technol Business Sch Shanghai 200093 Peoples R China

This paper proposes a methodology combining Long-Short-Term-Memory (LSTM)-assisted kinematic motion prediction with a hybrid coding algorithm for compressing the trajectory data of Connected Autonomous Vehicles (CAVs). The vehicle locations after the first two time steps are predicted based on the vehicle positions at the first two time steps and the kinematic equation. The vehicle velocities and accelerations are predicted based on the vehicle locations and LSTM. The hybrid coding algorithm integrates differential coding, Binary Coded Decimal (BCD) coding and arithmetic coding. Differential coding converts the original data into the difference between the original data and the predicted data. Since the length of the original data is large but the difference between it and predicted data is small, the required space for storing the data can be greatly reduced. BCD coding converts subsequences of different lengths to the subsequences with the same length so that the original information can be correctly reproduced after decompression. Arithmetic coding expresses the information in small space by converting the character sequence into a decimal between 0 and 1. The proposed algorithm is evaluated on the Next Generation Simulation Trajectory dataset. The experiment results show that the compression ratio and compression rate obtained by the proposed algorithm are respectively higher and lower than those obtained by the baseline algorithms. Also, the sum of compression time, decompression time and transmission time associated with the proposed algorithm is less than that associated with most baseline algorithms and transmission without compression.

关键词： Trajectory prediction algorithms Logic gates Kinematics Channel coding Arithmetic Data compression Trajectory data compression connected autonomous vehicles long short-term memory differential coding BCD coding arithmetic coding

来源：评论

学校读者我要写书评

暂无评论

Frequency Domain prediction of Tonal Signals With Time-Varying Pitches

引用

IEEE SIGNAL PROCESSING LETTERS 2025年 32卷 31-35页

作者： Guo, Ning Edler, Bernd Joint Inst Friedrich Alexander Univ Erlangen Nurnb Int Audio Labs Erlangen IIS D-91058 Erlangen Germany

In this letter, we propose an Extended Frequency Domain Joint Harmonics prediction (EFDJHP) algorithm, which does long term prediction directly in the transform domain on tonal signals with time-varying pitches for transform speech and audio coding. EFDJHP is an algorithm extension of a previously proposed Frequency Domain Joint Harmonics prediction (FDJHP) algorithm where a constant pitch between neighboring frames was assumed (Guo and Edler, 2021). A linearly changing pitch between adjacent frames is assumed in EFDJHP, where the linearity can be assumed to be local and updated across frames. EFDJHP works in a backward prediction fashion without additional algorithmic delay, and needs very few side information for the forward adaption of the predictor coefficients. Bitrate saving analysis and a listening test show that EFDJHP can improve the coding efficiency on tonal signals with frequent pitch variations such as singing voices and speech.

关键词： Harmonic analysis Transforms Time-frequency analysis prediction algorithms Audio coding Speech coding Signal processing algorithms Time-domain analysis Mathematical models Chirp Long term prediction time varying pitch audio coding FDJHP

来源：评论

学校读者我要写书评

暂无评论

Exploring Potential of ML-Aided Mobile Traffic prediction for Energy-Efficient Optimization of Network Resources Using Real World Dataset

引用

IEEE ACCESS 2024年 12卷 93606-93622页

作者： Kolackova, Aneta Sevgican, Salih Fatih Ulu, Muhammet Sadreddini, Jale Masek, Pavel Hosek, Jiri Jerabek, Jan Tugcu, Tuna Brno Univ Technol Dept Telecommun Brno 60190 Czech Republic Bogazici Univ NETLAB Dept Comp Engn TR-34342 Istanbul Turkiye Istanbul Arel Univ Dept Comp Engn TR-34537 Istanbul Turkiye Tampere Univ Unit Elect Engn Tampere 33100 Finland

To meet the extremely stringent but diverse requirements of Beyond Fifth-Generation (B5G) networks, traffic-aware adaptive utilization of network resources is becoming essential. To cope with that, a detailed traffic data analysis enables opportunities for mobile network operators to improve the Quality of Service (QoS) in the next-generation mobile communication systems. This paper presents a comprehensive analysis of the real world data collected from an operator's 4G+ and 5G infrastructure during a seven-month campaign. Efficient Machine Learning (ML) based network traffic predictions are presented together with a statistical model to develop optimal resource allocation strategies by using the data gathered during the pandemic, an era when the data volume, as well as the bandwidth requirements and the end users' expectations, were significantly elevated in terms of QoS, given the huge shift to the online world. Data analysis confirmed the assumption that there are traffic changes during the day and the whole week, which helped us to find new research directions regarding resource allocation optimization of next-generation mobile networks. Furthermore, we introduce the Predictive Energy Saver for Baseband Units (PESBiU) algorithm, which utilizes traffic prediction and power consumption analysis to manage the power states (sleep or active) of BBUs in a network. The PESBiU algorithm utilizes the results from ML predictions to effectively balance energy efficiency and network performance, demonstrating its potential for practical deployment in future mobile communication networks by transitioning BBUs to sleep mode during low-traffic periods, thereby achieving significant power savings.

关键词： Long short term memory Quality of service Resource management Telecommunication traffic Logic gates 5G mobile communication prediction algorithms Data analysis Machine learning Energy management Beyond 5G networks data analysis machine learning mobile traffic resource allocation BBU energy saving

来源：评论

学校读者我要写书评

暂无评论

Collision Probability Field Based Interaction-Aware Longitudinal Motion prediction

引用

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2024年第9期25卷 12095-12107页

作者： Na, Yuseung Lee, Minchul Kang, Jeonghun Sunwoo, Myoungho Jo, Kichun Hanyang Univ Dept Automot Engn Seoul 04763 South Korea ACEWORKS Inc Seoul 06222 South Korea Konkuk Univ Dept Smart Vehicle Engn Seoul 05029 South Korea Korea Univ Dept Automot Convergence Seoul 02841 South Korea

In the realm of autonomous driving, motion planning for the ego vehicle necessitates the prediction of surrounding vehicles' motions. This prediction traditionally relies on object tracking modules containing several sensors to gauge vehicle positions and velocities. However, existing physical or maneuver-based model approaches overlook important aspects of vehicle interactions that significantly affect actual vehicle movement. Ignoring these interactions can lead to inadequate ego vehicle's motion planning. Addressing this gap, this paper proposes a novel approach: the Collision Probability Field (CPF)-based interaction-aware longitudinal motion prediction. Our methodology uniquely integrates the CPF, derived from the uncertainty of sensing information, to account for the probabilistic state of vehicle positions and velocities. This allows the prediction algorithm to consider not just the static data, but also the dynamic interactions between vehicles such as collision. Our approach was tested in various scenarios, including lane changes with an approaching vehicle from behind and different driver behavior models in real-world conditions. Our findings demonstrate a significant improvement in prediction accuracy for the motion planning of ego vehicle, highlighting the importance of interaction-aware predictions in autonomous driving systems.

关键词： Predictive models Uncertainty prediction algorithms Mathematical models Sensors Measurement uncertainty Estimation Collision probability field collision risk assessment interaction-aware motion prediction measurement uncertainty

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：