NetWork
Capability-Aware Fair Complex Task Assignment in Spatial Crowdsourcing
FANG Xiu;ZHU Tiancheng;YI Yuqiong;HUANG Xinwei;SUN Guohao;Spatial crowdsourcing(SC) requires efficient task assignment algorithms,particularly for complex tasks requiring sequential completion.Despite extensive research on complex task assignments,most studies overlook the importance of fairness and the worker capability,leading to resource wastage.Fairness means ensuring that each worker is treated justly in the task assignment process.The worker capability refers to the maximum number of tasks that they can complete.This paper defines a capability-aware fair complex task assignment(CAFCTA) problem in SC,aiming to maximize the number of completed tasks,the total utility of workers,and capability-aware fairness under worker capability,task dependency,and time constraints.To effectively address the CAFCTA problem,we propose two algorithms:a fast assignment algorithm using a greedy strategy to match tasks to the most suitable workers quickly;an enhanced greedy algorithm with simulated annealing(SA),introducing a random search mechanism to avoid local optima.Experimental results demonstrate the effectiveness and efficiency of our algorithms.
Integrating Hierarchical Dependencies and Correlation for Multi-Label Text Classification
ZHU Jiaxin;LIU Guohua;This study addresses the challenges in scientific paper classification caused by the exponential growth of scientific papers,which intensifies demands for efficient manuscript categorization and cross-disciplinary knowledge retrieval.To overcome limitations of conventional methods in handling multi-label scenarios and domain-specific nuances,a novel multi-label text classification model is proposed,integrating semantic-aware label correlation modeling with advanced text representation learning.This model constructs a discipline-aware label graph through a graph attention network,incorporating hierarchical information and co-occurrence patterns from domain knowledge,while employing DeBERTaV3 to capture deep contextual semantics through its enhanced Transformer architecture.Text embeddings and label representations undergo systematic fusion,with hybrid sampling strategies and text augmentation techniques mitigating data imbalance issues.Experimental results demonstrate that the proposed model achieves higher performance in cross-disciplinary text classification compared to existing state-of-the-art methods.
Minibatch-Enhanced Generative Adversarial Networks for IoT Network Intrusion Detection
YU Shoujian;LIU Jiakang;ZHANG Hong;SHEN Shigen;With the rapid expansion of the Internet of Things(IoTs),network connectivity has exposed systems to increasingly complex and previously unseen security threats.Conventional intrusion detection systems(IDSs) often struggle to identify subtle anomalies in high-dimensional and dynamically evolving IoT traffic.To address this problem,we propose a generative adversarial network(GAN)-based IDS tailored for IoT networks.The proposed method integrates minibatch discrimination to enhance sensitivity to inter-sample anomalies,replaces the conventional L1 distance with cosine similarity to better characterize relationships among high-dimensional feature vectors,and incorporates L1/L2 regularization to improve training stability and generalization.These enhancements enable more accurate discrimination between benign and malicious traffic while reducing overfitting.Experiments on the TON_IoT dataset demonstrate that the proposed method achieves strong overall detection performance in terms of accuracy and F1 score while maintaining a low false positive rate.
Polarization Enhancement in Y3+-Doped Bismuth Titanate Nanosheets for Boosted Piezo-Photocatalytic Performance toward Efficient Water Purification
HANG Jiahe;HU Chunyan;LIU Baojiang;The development of efficient piezo-photocatalytic systems is crucial for environmental remediation and the purification of wastewater.In this study,the yttrium ion(Y3+) was introduced into the A-site of Aurivillius-phase bismuth titanate(Bi_4Ti_3O12,BTO) via a structural modulation strategy to fabric a novel piezo-photocatalyst(denoted as Y-BTO) for the synergistic degradation of tetracycline hydrochloride(TC).The results reveal that replacing Bi3+ with smaller-radius Y3+ not only induces lattice distortion but also enhances the intrinsic piezoelectric polarization within the material.Under simultaneous ultrasound and light irradiation,the optimally doped YBTO catalyst exhibits outstanding piezo-photocataly tic activity,achieving a TC removal rate of 88.8% within60 min.The apparent pseudo-first-order rate constant for the piezo-photocatalytic degradation of TC is determined to be 29.40 × 10-3 min-1,which is 3.6 and 2.3 times the values for individual piezocatalysis and photocatalysis,respectively.This remarkable enhancement is attributed to the piezo-photocatalytic synergy:the piezoelectric field generated by ultrasonic vibration effectively facilitates the separation and migration of photogenerated electron-hole pairs,thereby significantly improving the quantum efficiency.This study demonstrates that A-site ion substitution is an effective strategy for designing highperformance piezo-photocatalysts for environmental remediation.
Long-Term Returns in CIR Models Driven by α-Stable Lévy Process
LU Wenhan;YAN Litan;TAO Qing;Interest rates play a crucial role in national economic operations,capital flows,asset pricing,and insurance product returns,yet their long-term dynamics under non-Gaussian noise remain under explored.This study extends the traditional Cox-Ingersoll-Ross(CIR) model by introducing an α-stable Lévy process and investigates whether the long-term time average of the short-term interest rate converges to a finite limit in the long run under the stable CIR(SCIR) model.Numerical simulations are conducted across various parameter sets to validate the robustness of the model's convergence behavior under different scenarios,and an empirical analysis using data from the Shanghai interbank offered rate(Shibor) is performed.The results confirm that under suitable conditions,the long-run time average of the short-term interest rate under the SCIR model converges to a finite limit,supported by both simulations and Shibor data.These findings provide theoretical insights into the long-term dynamics of interest rates driven by non-Gaussian noise,with practical implications for financial asset pricing and risk management.
Dynamic Tracking Model for Spacecraft Center of Mass in Gravitational Wave Detection Based on IAROA-XGBoost
SUN Zihao;SHENG Xiaowei;XU Yang;CAI Zhiming;During the in-orbit operation of gravitational wave detection spacecraft,the spacecraft center of mass(CoM) undergoes continuous changes due to various factors,which adversely affect detection accuracy.Existing CoM identification algorithms fail to meet real-time acquisition requirements and may disrupt the mission.To address this issue,a CoM dynamic tracking model is proposed,taking the rotation angles of the internal moving optical subassemblies(MOSA) and high-gain antenna(HGA) as input features and the spacecraft's CoM position coordinates as output variables.First,an improved attraction-repulsion optimization algorithm(IAROA) is developed to optimize the hyperparameters of the eXtreme Gradient Boosting(XGBoost) model,forming the IAROAXGBoost model.Second,a discrete CoM-rotation angle mapping dataset is established by integrating a highprecision spacecraft finite element model with Latin hypercube sampling.Based on this dataset,a dynamic CoM tracking model is obtained by training with IAROAXGBoost,and the accuracy of the model is compared with five commonly used machine learning models.Finally,SHapley Additive exPlanations(SHAP) is employed for interpretability analysis to identify the dominant factors influencing CoM position and investigate how feature interactions affect the model outputs.The results demonstrate that the proposed model achieves a CoM prediction accuracy better than 0.5 μm,outperforming the comparative models.It enables high-precision dynamic tracking of the spacecraft's CoM as it changes with component rotation angles,providing real-time and reliable CoM data for the spacecraft's Drag-Free Attitude and Control System,formation flight control,and CoM adjustment,ensuring the successful execution of the gravitational wave detection mission.
EST-NSS: An Enhanced Swin Transformer for Nighttime Semantic Segmentation
TAO Ran;DING Yi;LU Hailun;LU Xiaohui;YIN Wenxiu;ZHANG Qi;Although semantic segmentation of daytime images has achieved remarkable success,nighttime semantic segmentation still faces significant challenges due to the low brightness and contrast in nighttime images.To address these challenges,we proposed an enhanced Swin Transformer for nighttime semantic segmentation(ESTNSS) for nighttime urban streetscape segmentation.The EST-NSS utilizes Swin Transformer as the backbone network.In the image preprocessing stage,we designed a nighttime light enhancer(NLE),which integrates receptive-field attention convolution(RFAConv),attention mechanisms,and a Transformer to enhance image brightness,contrast,and semantic representation,thereby improving the model's perception of key regions.In the feature processing stage,we designed a dual-domain context unit(termed SCBlock),which integrates attention mechanisms and depthwise separable convolution to enhance multi-scale feature representations,thereby improving the model's accuracy in segmenting object boundaries.Extensive experimental results on the NightCity dataset show that our EST-NSS significantly enhances the performance of nighttime semantic segmentation.The mean intersection over union(mIoU) improved by 3.6 percentage points compared with Swin Transformer.Compared with the state-of-the-art models,the EST-NSS achieved superior performance in 15 out of 19 categories.
Hydrogel Functionalized Textiles for Smart and Sustainable Applications
HOSSAIN Khan Rajib;JALIL M.Abdul;KHAN Ayub Nabi;FUAD Khan Md.Salman Adib;MONIRUZZAMAN Md.;SAYEK Shabir;SAGOR S.M.Shahadat Hossain;Hydrogels have garnered significant interest due to their remarkable hydrophilicity,biocompatibility,and customizable biomimetic properties.However,their use is limited by inherently poor mechanical properties,arising from their low-density chains and irregular microstructures.It remains challenging to improve the strength and toughness.To address these drawbacks,hydrogels are incorporated into textiles to create functional hydrogel-based composites.This paper reviews advances in hydrogel textile composites,focusing on design strategies and property enhancement.It also discusses the progress of these composites in areas such as medical dressings,wearable electronic devices,and flame-retardant protection while looking ahead to future research directions.
Covalent Organic Framework with Electron Push-Pull Effect for Selective Photocatalytic Oxidation of Sulfides
HUANG Zhenxuan;ZHAO Qixin;JIANG Zhiqiang;ZENG Yang;SHEN Jiefeng;XUAN Weimin;The selective aerobic oxidation of sulfides enables efficient production of sulfoxide-containing pharmaceuticals and agrochemicals.However,conventional catalytic systems frequently suffer from compromised selectivity due to undesired over-oxidation of sulfides to sulfones.Photocatalysis has emerged as a transformative strategy to address this persistent selectivity challenge,prompting extensive exploration of photocatalytic materials.In particular,covalent organic frameworks(COFs)distinguish themselves as superior platforms for the photocatalytic selective aerobic oxidation of sulfides,due to their structural tunability,tailorable optoelectronic properties,and operational durability under photocatalytic conditions.Herein,two COFs(TP-BPDA and BTT-BPDA)were synthesized via Schiff-base condensation.Photoelectrochemical experiments demonstrated that TP-BPDA,benefiting from a more pronounced electron push-pull effect,showed better photocatalytic performance for the selective aerobic oxidation of sulfides to sulfoxides.Moreover,TP-BPDA exhibited broad substrate generality.Based on the scavenger experiment,a putative reaction mechanism was proposed.This study provides guidance for the further development of high-efficiency photocatalysts for selective aerobic oxidation by leveraging the electron pushpull effect.
Effect of Cotton Waste-Based MCC on Interface and Mechanical Properties of Jute/Glass Fiber-Reinforced Epoxy Composites
LI Hongyang;YANG Jiahua;WANG Zechuan;YAO Lan;WANG Ni;With the increasingly prominent issues of energy shortage and environmental pollution,it has become imperative to develop natural fiber-reinforced composites for applications in infrastructure,automobiles,etc.In this study,jute fibers,as low-cost natural fibers,were selected as the reinforcement for the composites.To enhance the overall mechanical properties,glass fibers were also added as the reinforcement to form the hybrid composites.To improve the interfacial properties between the fibers and the epoxy resin,the microcrystalline cellulose(MCC)prepared from cotton waste was employed to modify the composites.The introduction of MCC in the designed four jute/glass epoxy composites changed the mechanical properties.Especially for the composite with two layers of glass fiber fabrics on each of the top and bottom surfaces and two layers of jute fabrics in the middle(designated as 2G2J2G),its bending strength and bending modulus after MCC addition reached 113.66 MPa and 11.87 GPa,respectively,which were 34.19% and 23.26% higher than those of the composite without MCC.Scanning electron microscopy(SEM) observations revealed that MCC was uniformly dispersed in the epoxy resin matrix,which would have a positive effect on the interfacial strengthening.