30793 research outputs found
Sort by
Industrial Anomaly Detection Based on Improved Diffusion Model: A Review
No full textData Availability:
No datasets were generated or analysed during the current study.As a class of highly effective generative models, diffusion models have attracted considerable attention in recent years and have been extensively applied to industrial anomaly detection tasks. In this review, a comprehensive discussion is presented on industrial anomaly detection based on improved diffusion models. Initially, a brief introduction to diffusion models and anomaly detection is provided, covering fundamental concepts, widely used datasets, and evaluation metrics. Recent advancements in diffusion models are then outlined from three key perspectives: inference speed, generalization ability, and reconstruction quality. Furthermore, their applications in industrial anomaly detection are examined, including sample generation, data augmentation, and the reconstruction of anomalous images. Finally, recent developments in diffusion models are summarized, and several potential research directions are suggested for future investigation.This work was supported in part by the Royal Society of the UK under Grant IES\R3\243021 and the Alexander von Humboldt Foundation of Germany
Multi-sensor Particle Filtering for Nonlinear Complex Networks With Heterogeneous Measurements Under Non-Gaussian Noises
Get PDFIn this article, the multisensor particle filtering problem is investigated for a class of nonlinear complex networks with multirate heterogeneous measurements. The underlying complex networks are subject to non-Gaussian noises and randomly switching couplings, while the multirate heterogeneous measurements (including fast-rate binary measurements and slow-rate integral measurements) are transmitted to remote filters via imperfect wireless communication channels. Both the deterministic and stochastic channel gains, along with possible transmission failures, are taken into account to characterize the properties of wireless communication channels. The purpose of this article is to propose a channel-related filtering scheme in the particle filtering framework to address these engineering-oriented complexities. To achieve this, a mixture distribution is established to reflect the effects of randomly switching couplings and generate new particle candidates. By utilizing the Monte Carlo approximation method, two types of update expressions for importance weights are explicitly derived based on the channel properties and the likelihood functions. Finally, numerical simulations are presented to demonstrate the viability and effectiveness of the proposed particle filtering algorithms.10.13039/501100001809-National Natural Science Foundation of China (Grant Number: 62203016, 62425301, U2241214, 62373008 and 61933007);
10.13039/501100002858-China Postdoctoral Science Foundation (Grant Number: 2021TQ0009);
10.13039/501100001809-Royal Society of the U.K.;
Alexander von Humboldt Foundation of Germany
Recursive Quadratic Filter Design for Non-Gaussian Systems Under Random Access Protocol: A Zero-Order Hold Strategy
Get PDFThis article deals with the recursive quadratic filtering problem for a class of linear discrete-time systems with the random access protocol (RAP) and non-Gaussian noises (NGNs). In order to mitigate undesirable data collisions, the RAP scheduling, used in conjunction with the zero-order hold strategy (ZOHS), is exploited in the sensor-to-filter channel. This coordination of the transmission order of sensors is characterized by a set of independent and identically distributed random variables. The objective of this article is to design a RAP-based quadratic filtering algorithm within the minimum-variance framework. The addressed system is first transformed into an enhanced system, which offers more information about the RAP and NGNs, by assembling the augmented states (including the original state and the latest measurement) and their second-order Kronecker power. With the assistance of two difference equations, an upper bound on the filtering error covariance (FEC) is, then, established, and the gain parameter is subsequently designed by minimizing this upper bound. To address challenges from the RAP scheduling with ZOHS, a matrix decomposition technique is employed. The effectiveness of the proposed RAP-based quadratic filtering algorithm is ultimately confirmed by various simulation results.10.13039/501100001809-National Natural Science Foundation of China (Grant Number: 62473235 and U21A2019);
10.13039/501100001809-Royal Society of U.K.;
Alexander von Humboldt Foundation of Germany
Privacy-Preserving Distributed Energy Management for Battery Energy Storage Systems over Time-Varying Networks
Get PDFThis article addresses the privacy-preserving energy management problem of battery energy storage systems (BESSs). An autonomous privacy-preserving distributed optimization (APPDO) scheme is developed over time-varying networks with the aim of regulating the power output of local BESS to fulfill the total load demand at the minimum economic cost under battery capacity constraints without privacy leakage. To this end, a linearly convergent distributed algorithm is proposed by combining the gradient descent algorithm with leaderless and leader-following consensus schemes. This algorithm is applicable to both islanded and grid-connected modes of BESSs. Furthermore, a novel privacy-preserving approach is constructed by injecting well-designed perturbation sequences into the data exchanged between neighboring nodes, making it effective against malicious eavesdroppers. Furthermore, a comprehensive analysis framework is established to evaluate the convergence, optimality, and privacy-preserving performance of the APPDO algorithm. Finally, numerical studies are conducted to demonstrate the effectiveness of the developed APPDO scheme.10.13039/501100001809-National Natural Science Foundation of China (Grant Number: 62303210 and 62188101);
Shenzhen Science and Technology Program (Grant Number: JCYJ20241202125309014 and KQTD20221101093557010);
10.13039/501100012245-Science and Technology Planning Project of Guangdong Province (Grant Number: 2024B1212010002);
Future Resilient Systems at the Singapore-ETH Centre;
10.13039/501100000266-Engineering and Physical Sciences Research Council;
Royal Society of the U.K.;
Alexander von Humboldt Foundation of Germany
EXPRESS: Sad, Angry and Fearful Facial Expressions Interfere with Perception of Causal Outcomes
Get PDFData Accessibility Statement: The data and materials from the present experiment are publicly available athttps://doi.org/10.6084/m9.figshare.28162817.v1 . Data, were analyzed using SPSS, version25 without pre-registration but the primary theories regarding statistical learning are derivedfrom existing associative theory regarding contingency learning (Murphy et al., 2017).Facial expressions convey a speaker's emotional state, facilitating the prediction and interpretation of their thoughts and behaviours. Interactive feedback during social interactions provides statistical evidence, for the basis of a causal percept which allows understanding of conversations. We aimed to determine whether emotional expression affects sensitivity to contingent relationships and whether this sensitivity is guided by the statistical evidence for causality. In Experiments 1-3, we tested happy and sad facial expressions and non-emotional control stimuli (e.g., shapes) and varied contingent emotional expressions (negative, zero, and positive contingency) as well as outcome frequency (low, moderate, and high). Participants’ judgements of contingency were based on a probabilistic learning process rather than simple pairing or prior knowledge and they perceived a weaker sense of causality with sad faces than either happy faces or non-emotional control stimuli. Finally, in Experiment 4, we tested threat-related angry and fearful faces alongside happy faces. The results showed that participants could learn the statistical contingent relationships with faces but still perceived a weaker sense of causality with angry, and fearful faces compared to happy faces. Overall, the results suggest that learning was guided by statistical evidence, but aversive expressions (those with negative valence) were less effective. We discuss this result in relation to the stimulus properties (i.e., salience) of faces, the content of emotive expressions and how these impact learning.This work was supported by the Scientific and Technological Research Council of Turkiye(TUBITAK) [Grant Number: 1059B192202594] to Rahmi Saylik
Not Everyone Feels the Same: Engagement Profiles and User Reactions to Reddit’s Great Ban
Get PDFConference paper presented at the Forty-Sixth International Conference on Information Systems, Nashville, TN, USA, 11-14 December 2025.Online platforms increasingly function as digital societies, where top-down moderation interventions like subreddit bans aim to regulate user behavior. However, user responses vary widely, and prior research offers mixed evidence of effectiveness. Guided by theories of psychological reactance and rationalization, and drawing on the Social Media Engagement Behavior framework, this study examines how pre-ban behavioral engagement metrics explain changes in toxicity following subreddit bans. Using a dataset of 1,798 disruptive users across 15 banned subreddits, we analyze pre–post toxicity changes via Google’s Perspective API and multiple regression. We conceptualize engagement on Reddit across three levels: platform, subreddit, and activity. Findings reveal that users with higher pre-ban engagement intensity and greater behavioral consistency tend to increase toxicity, while those with exclusive subreddit focus are more likely to reduce it. These results demonstrate the explanatory value of engagement profiles and offer implications for more targeted, data-driven moderation strategies in online communities
Application of artificial intelligence in 3D printing
Get PDFThis paper explored the integration of AI with 3D printing technology to address two key challenges: predicting surface roughness and detecting printing defects. A full factorial design of experiments was conducted using a Bambu Lab P1S printer and PLA material, varying three process parameters: temperature, layer height, and filling density, across 27 combinations. Surface roughness (Ra) was measured with a TR200 profilometer, and the data was used to train and evaluate five machine learning models: ANN, SVM, DT, RF, and KNN. Among these, SVM demonstrated the best generalization performance, while ANN achieved the highest R² on the training set. Meanwhile, a custom CNN was trained on a publicly available dataset containing 1,912 labelled defect images to classify five common FDM defects. The CNN achieved an accuracy of 98.6% along with excellent ROC performance, confirming its reliability for real-world defect detection. The results validate the effectiveness of AI-driven approaches for additive manufacturing process optimization and quality assurance. The study demonstrates how machine learning and deep learning models can improve the intelligence, accuracy, and efficiency of FDM systems, thereby helping to achieve smart manufacturing and Industry 4.0 goals
Visual transformer with depthwise separable convolution projections for video-based human action recognition
Get PDFHuman action recognition is a task that utilizes algorithms to recognize human actions from videos. Transformer-based algorithms have attracted growing attention in recent years. However, transformer networks often suffer from slow convergence and require large amounts of training data, due to their inability to prioritize information from neighboring pixels. To address these issues, we propose a novel network architecture that combines a depthwise separable convolution layer with transformer modules. The proposed network has been evaluated on the medium-sized benchmark dataset UCF101 and the results have demonstrated that the proposed model converges quickly during training and achieves competitive performance compared with SOTA pure transformer network, while reducing approximately 7.4 million parameters.This work is supported by the Zhongyuan University of Technology-Brunel University London (ZUT-BUL) Joint Doctoral Training Programme. This work is funded by the ZUT/BRUNEL scholarship
Inter- and intra-bacterial strain diversity remains the “elephant in the (living) room”
Get PDFData availability:
No datasets were generated or analysed during the current study.Acinetobacter baumannii is an opportunistic Gram-negative bacterial pathogen responsible for severe nosocomial infections worldwide. Resistance to last-resort antibiotics causes A. baumannii to be ranked as a top priority for the research and development of new antibiotics by the WHO and an urgent threat to public health by the CDC. It is also a member of the ESKAPE group comprising the most problematic antibiotic-resistant nosocomial pathogens. Resistance towards desiccation, disinfectants, reactive oxygen species, and the host immune system helps A. baumannii thrive in hospital settings and infect individuals compromised by lines, tubes, and indwelling devices. A. baumannii displays extensive genomic heterogeneity, yet recent studies show that this level of plasticity is also prevalent in lab strains widely used to study A. baumannii biology. Successive subculturing of widely used strains and spontaneous genetic variations results in significantly altered genotypes and phenotypes, often not recognized by the scientific community. In addition, the current strain designation methods do not allow efficient communication about such differences. Even presumably identical strains from established culture collections have been found to demonstrate genetic heterogeneity. The “elephant in the (living) room” refers to the risk but also the potential of the bi-partite problem concerning the high diversity amongst A. baumannii isolates (inter-strain variability), and the universal issue of microevolution (intra-strain variability). This is generally ignored as it is not referenced adequately in scientific publications. We aim to raise awareness about the current issues and the problematic consequences generated by intra- and inter-strain diversity based on modern examples of A. baumannii isolates. Therefore, this review provides cases of broadly used A. baumannii strains and their genetic and phenotypic differences.C.V.D.H. is supported by the Flanders Institute for Biotechnology (VIB). A.V. is a recipient of a junior postdoctoral fellowship of the Research Foundation – Flanders (FWO; file number 1287223N). We gratefully acknowledge funding from the National Institutes of Health, grants AI138576 and AI150098 to M.S.T. R.R.M.C. and R.D. are supported by a Biotechnology and Biological Sciences Research Council New Investigator Award BB/V007823/1 and Medical Research Council Grant MR/Y001354/1. R.R.M.C. is supported by the Academy of Medical Sciences/the Wellcome Trust/the Government Department of Business, Energy and Industrial Strategy/the British Heart Foundation/Diabetes UK Springboard Award [SBF006\1040]. PNR is supported by grants I01BX001725 and IK6BX004470 from the Department of Veterans Affairs. XC is supported by grants FRM EQU202303016268 and ANR-20-CE12-0004. PV was supported by the National Biodiversity Future Centre, PNRR, Missione 4 Componente 2, ‘Dalla ricerca all’impresa’, Investimento 1.4, Project CN00000033. IE is supported by the research funding program Landes-Offensive zur Entwicklung Wissenschaftlich-ökonomischer Exzellenz (LOEWE) of the State of Hessen, Research Center for Translational Biodiversity Genomics (TBG)
Investigating the Impact of Musical Soundscapes on Well-being: A Qualitative Focus Group Study Using Arts-Based Methods
Get PDFThis study explores the impact of musical soundscapes on well-being through a qualitative inductive thematic analysis. Utilizing focus groups and participatory arts-based methods, participants engaged with meditative soundscapes periodically over a week, sharing their responses through text, voice, and visual imagery. These multisensory responses were cross-referenced with focus group transcripts to deepen the thematic analysis. The findings reveal diverse positive outcomes, including personal, psychological, physiological, and sociocultural benefits. Notable emergent themes include intersensory synchrony, embodied musical affects, stress relief, self-transcendence, communal connection, and integrated well-being. The study underscores the capacity of musical experiences, which transcend cultural boundaries, to enhance well-being across these dimensions. These preliminary results highlight the potential of cross-cultural musicality to foster holistic well-being, suggesting that musical soundscapes can act as a powerful medium for enhancing well-being.摘要
本研究通过质性归纳式主题分析, 探讨音乐声景对幸福感的影响。采用焦点小组与参与式艺术本位方法, 受试者在一周内分阶段体验冥想声景, 并通过文字、语音及视觉意象分享其反馈。这些多感官反馈与焦点小组转录文本进行交叉比对, 以深化主题分析。研究结果揭示了多样化的积极效益, 涵盖个人、心理、生理及社会文化多个层面。显著涌现的主题包括：多感官协同效应、具身化音乐影响、压力缓解、自我超越、社群联结及整合性幸福感。该研究强调了音乐体验超越文化界限、进而提升多维度幸福感的能力。这些初步结果凸显了跨文化音乐性促进整体幸福感的潜力, 表明音乐声景可作为提升幸福感的有效媒介。This project received funding from Beyond Beta
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇