The bio-inspired emerging dynamic vision sensor (DVS), characterized by its exceptional high temporal resolution and immediate response, possesses an innate advantage in capturing rapidly changing scenes. Nevertheless...
详细信息
The bio-inspired emerging dynamic vision sensor (DVS), characterized by its exceptional high temporal resolution and immediate response, possesses an innate advantage in capturing rapidly changing scenes. Nevertheless, it is also susceptible to severe noise interference, especially in challenging conditions like low illumination and high exposure. Notably, the existing noise processing approaches tend to oversimplify data into 2-dimensional (2D) patterns, disregarding the sparse and irregular crucial event structure information that the DVS intrinsically provides via its asynchronous output. Aiming at these problems, we propose a residual graph neural network (RGNN) framework based on density spatial clustering for event denoising, called DBRGNN. Leveraging the temporal window rule, we extract non-overlapping event segments from the DVS event stream and adopt a density-based spatial clustering algorithm to obtain event groups with spatial correlations. To fully exploit the inherent sparsity and plentiful spatiotemporal information of the raw event stream, we transform each event group as compact graph representations via directed edges and feed them into a graph coding module composed of a series of graph convolutional and pooling layers to learn robust geometric features from event sequences. Importantly, our approach effectively reduces noise levels without compromising the spatial structure and temporal coherence of spike events. Compared with other baseline methods, our DBRGNN achieves competitive performance by quantitative and qualitative evaluations on publicly available datasets under varying lighting conditions and noise ratios.
作者：
Tao, ZixingCao, WeihuaGan, Chao
Wuhan China
School of Future Technology School of Automation Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems Wuhan China
Air quality data exhibit nonlinearity, sensitivity to environmental factors, and long-term dependencies. Numerous factors influence air quality, making accurate predictions based on a single-dimensional dataset imprac...
详细信息
This paper addresses the finite-time consensus (FTC) issue for second-order multi-agent systems (MASs) with nonlinear disturbances. To tackle the challenges posed by increasingly complex communication environments, an...
详细信息
Numerous high-dimensional solutions for many-objective optimization problems (MaOPs) usually impose a high cognitive burden on decision makers (DMs). Pareto front (PF) of MaOPs can express the problem characteristics,...
详细信息
Numerous high-dimensional solutions for many-objective optimization problems (MaOPs) usually impose a high cognitive burden on decision makers (DMs). Pareto front (PF) of MaOPs can express the problem characteristics, and then provide prior knowledge for solving the MaOPs. However, the existing high-dimensional visualization methods usually do not establish the relationship between PF information and decision making. Therefore, a novel radial visualization (RadViz) method called KRadViz that incorporates knee point information is proposed to visualize the information of PF shape and aid decision making. The relationship between the optimized performance information and PF shape is established, and the PF shape identification method is constructed. KRadViz is constructed by combining the optimization performance and PF shape. Three preferred solution selection methods are proposed to quickly screen out a few preferred solutions in different scenarios. The proposed KRadViz is compared with three high-dimensional visualization methods. The experimental results show that KRadViz can effectively display the high-dimensional PF shape, and give the optimization performance information of different solutions. The selection preferences of the three methods are also analyzed, and the effectiveness of the assisted decision process is verified. For the DTLZ2 and real-world MaOPs, the individual hypervolume (HV) contribution of preferred solutions increased by 9.98 % and 10.95 %, respectively.
Accurately and promptly detecting the pipeline anomaly is crucial to the safe operation of pipeline systems, while a difficulty lies in that many existing methods require massive data for training models. However, pip...
详细信息
作者：
Kuang, ZehuiMao, FanZhao, XingyuWan, XiongboSchool of Automation
China University of Geosciences Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems Engineering Research Center of Intelligent Technology for Geo-Exploration Ministry of Education Wuhan China School of Future Technology
China University of Geosciences Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems Engineering Research Center of Intelligent Technology for Geo-Exploration Ministry of Education Wuhan China
Domain adaptation methods and appropriate feature extractors are usually applied to solve the problem that the variable working conditions of bearings affect the effectiveness of the fault diagnostic framework. Howeve...
详细信息
作者：
Huang, NanxinXu, ChiSchool of Automation
China University of Geosciences Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems Engineering Research Center of Intelligent Technology for Geo-Exploration Ministry of Education Wuhan China
Driven by advancements in industrial production and artificial intelligence, the need for pose estimation of new ob-jects in areas like robotic manipulation and virtual reality is increasing. We introduce a zero-shot ...
详细信息
Facial expression recognition plays a key role in promoting the development of comprehensive intelligence and building friendly human-computer interaction. Due to the interference of feature noise in expression data, ...
详细信息
Using electric motors instead of diesel engines as the driving system for mining excavators can reduce the energy consumption and operating costs. However, pure electric-driven mining excavators are prone to unexpecte...
详细信息
The ever-increasing amount of solid waste generated globally has given rise to issues of environmental pollution and climate change impact associated with the resource recovery and ultimate disposal. Given the complex...
详细信息
The ever-increasing amount of solid waste generated globally has given rise to issues of environmental pollution and climate change impact associated with the resource recovery and ultimate disposal. Given the complexity of waste categories and multi-objective management needs both in developed and developing countries, it is urgent to employ innovative strategies, such as the machine learning techniques. In present study, we conduct a systematic review of the application of machine learning in the waste life-cycle management with consideration of advancement, challenges, limitations, and future directions. We found that the challenges currently faced by solid waste management systems are extremely complex, necessitating the use of knowledge in the field to constrain machine learning models to enable a deep integration between machine learning and sustainable waste management. Due to its proficiency in modeling complex nonlinear processes, particularly the great advantages in prediction and multi-objective optimization, existing studies on machine learning mainly focus on waste generation projection, collection and classification, transportation, recycling, and disposal route optimization. However, the performance of machine learning applications in the field of solid waste is hindered by limitations in data quality and quantity, as well as the insufficient interpretability of individual models. These challenges can be addressed by integrating methods such as multi-source data fusion, data augmentation techniques, and ensemble learning, along with the development of highly interpretable machine learning models. Overall, the key to achieving environmentally sound management of solid waste lies in optimizing designs and balancing multidimensional evaluation criteria.
暂无评论