Search Results (61)

Industrial image anomaly detection plays a critical role in intelligent manufacturing by automatically identifying defective products through visual inspection. While unsupervised approaches eliminate dependency on annotated anomaly samples, current teacher–student framework-based methods still face two fundamental limitations: insufficient discriminative capability for structural anomalies and suboptimal anomaly feature decoupling efficiency. To address these challenges, we propose a Synthetic-Anomaly Contrastive Distillation (SACD) framework for industrial anomaly detection. SACD comprises two pivotal components: (1) a reverse distillation (RD) paradigm whereby a pre-trained teacher network extracts hierarchically structured representations, subsequently guiding the student network with inverse architectural configuration to establish hierarchical feature alignment; (2) a group of feature calibration (FeaCali) modules designed to refine the student’s outputs by eliminating anomalous feature responses. During training, SACD adopts a dual-branch strategy, where one branch encodes multi-scale features from defect-free images, while a Siamese anomaly branch processes synthetically corrupted counterparts. FeaCali modules are trained to strip out a student’s anomalous patterns in anomaly branches, enhancing the student network’s exclusive modeling of normal patterns. We construct a dual-objective optimization integrating cross-model distillation loss and intra-model contrastive loss to train SACD for feature alignment and discrepancy amplification. At the inference stage, pixel-wise anomaly scores are computed through multi-layer feature discrepancies between the teacher’s representations and the student’s refined outputs. Comprehensive evaluations on the MVTec AD and BTAD benchmark demonstrate that our method is effective and superior to current knowledge distillation-based approaches. Full article

This study examines the critical role of institutional quality in driving corporate adaptation to climate change within the EU-27 member states from 2006 to 2023. It aims to investigate how governance factors—control of corruption, government effectiveness, rule of law, and regulatory quality—influence business strategies for environmental resilience and sustainability, focusing on environmental investments and industrial production. Employing fixed and random effects regression models on a balanced panel dataset, we analyze two dependent variables: environmental protection investment corporations (EPIC), measuring investments in pollution prevention and environmental degradation reduction, and industrial production (IP), reflecting output in mining, manufacturing, and utilities. A composite institutional quality index, derived through principal component analysis (PCA) from the four governance indicators, captures their collective impact, reducing multicollinearity and enhancing analytical robustness. Control variables, including final energy consumption, environmental tax revenues, expenditure on environmental protection, and a Paris Agreement dummy, are incorporated to test the institutional quality effect. Results demonstrate that higher institutional quality significantly enhances EPIC, particularly in countries with greater environmental tax revenues, indicating that robust governance and fiscal policies incentivize sustainable corporate investments. Conversely, the effect on IP is less consistent, with higher fossil energy consumption and lower environmental tax revenues driving production, suggesting a reliance on high-polluting industries. The Paris Agreement positively influences IP, reflecting stronger climate-focused industrial strategies post-2015. These findings underscore the pivotal interplay between institutional quality and environmental fiscal policies in fostering corporate adaptation to climate change. Over the long term, strong governance is essential for aligning business practices with sustainability goals, reducing environmental degradation, and mitigating climate risks across the EU. This study highlights the need for cohesive policies to support green investments and transition industries toward renewable energy sources, addressing disparities in environmental performance among EU member states. Full article

The rapid proliferation of Large Language Models (LLMs) across industries such as healthcare, finance, and legal services has revolutionized modern applications. However, their increasing adoption exposes critical vulnerabilities, particularly through adversarial prompt attacks that compromise LLM security. These prompt-based attacks exploit weaknesses in LLMs to manipulate outputs, leading to breaches of confidentiality, corruption of integrity, and disruption of availability. Despite their significance, existing research lacks a comprehensive framework to systematically understand and mitigate these threats. This paper addresses this gap by introducing a taxonomy of prompt attacks based on the Confidentiality, Integrity, and Availability (CIA) triad, an important cornerstone of cybersecurity. This structured taxonomy lays the foundation for a unique framework of prompt security engineering, which is essential for identifying risks, understanding their mechanisms, and devising targeted security protocols. By bridging this critical knowledge gap, the present study provides actionable insights that can enhance the resilience of LLM to ensure their secure deployment in high-stakes and real-world environments. Full article

Digital trade, as an emerging and transformative trade model in the digital era, has significantly altered global trade methods, products, services, and regulatory frameworks. This study investigates the impact mechanism of innovation capability on the sustainability of the ASEAN’s digital trade, emphasizing how technological advancements contribute to sustainable economic growth and digital resilience. Utilizing panel data from nine ASEAN countries between 2007 and 2021, this research explores how innovation capability fosters digital trade development by reducing the digital divide and promoting equitable access to digital markets. Findings highlight the substantial disparities in digital trade and innovation capacity across the ASEAN, with innovation capability playing a pivotal role in driving trade practices. This study reveals that digital readiness mediates the relationship between innovation capability and digital trade, while the RCA index serves as a moderating factor enhancing digital trade competitiveness. Furthermore, this study underscores that effective governance, regulatory quality, foreign direct investment (FDI), and a balanced wage–output ratio in the digital industry positively influence digital trade, whereas corruption and inadequate discourse power hinder it. The findings provide valuable policy recommendations for ASEAN countries to develop sustainable digital trade policies, strengthen innovation ecosystems, and bridge the digital divide, thereby contributing to the broader agenda of sustainable development. Full article

This paper deals with developing a comparative study of the principal time-domain system identification methods suitable for performing an experimental modal analysis of structural systems. To this end, this work focuses first on analyzing and reviewing the mathematical background concerning the analytical methods and the computational algorithms of interest for this study. The methods considered in the paper are referred to as the AutoRegressive eXogenous (ARX) method, the State-Space ESTimation (SSEST) method, the Numerical Algorithm for Subspace State-Space System Identification (N4SID), the Eigensystem Realization Algorithm (ERA) combined with the Observer/Kalman Filter Identification (OKID) method, and the Transfer Function ESTimation (TFEST) method. Starting from the identified models estimated through the methodologies reported in the paper, a set of second-order configuration-space dynamical models of the structural system of interest can also be determined by employing an estimation method for the Mass, Stiffness, and Damping (MSD) matrices. Furthermore, in practical applications, the correct estimation of the damping matrix is severely hampered by noise that corrupts the input and output measurements. To address this problem, in this paper, the identification of the damping matrix is improved by employing the Proportional Damping Coefficient (PDC) identification method, which is based on the use of the identified set of natural frequencies and damping ratios found for the case study analyzed in the paper. This work also revisits the critical aspects and pitfalls related to using the Model Order Reduction (MOR) approach combined with the Balanced Truncation Method (BTM) to reduce the dimensions of the identified state-space models. Finally, this work analyzes the performance of all the fundamental system identification methods mentioned before when applied to the experimental modal analysis of flexible structures. This is achieved by carrying out an experimental campaign based on the use of a vibrating test rig, which serves as a demonstrative example of a typical structural system. The complete set of experimental results found in this investigation is reported in the appendix of the paper. Full article

Heterogeneous federated learning, as an innovative variant of federated learning, aims to break through the constraints of vanilla federated learning on the consistency of model architectures to better accommodate the heterogeneity in mobile computing scenarios. It introduces heterogeneous and personalized local models, which effectively accommodates the heterogeneous data distributions and hardware resource constraints of individual clients, and thus improves computation and communication efficiency. However, it poses a challenge to model ownership protection, as watermarks embedded in the global model are corrupted to varying degrees when they are migrated to a user’s heterogeneous model and cannot continue to provide complete ownership protection in the local models. To tackle these issues, we propose a dynamic black-box model watermarking method for heterogeneous federated learning, PWFed. Specifically, we design an innovative dynamic watermark generation method which is based on generative adversarial network technology and is capable of generating watermark samples that are virtually indistinguishable from the original carriers. This approach effectively solves the limitation of the traditional black-box watermarking technique, which only considers static watermarks, and makes the generated watermarks significantly improved in terms of stealthiness and difficult to detect by potential model thieves, thus enhancing the robustness of the watermarks. In addition, we design two watermark embedding strategies with different granularities in the heterogeneous federated learning environment. During the watermark extraction and validation phase, PWFed accesses watermark samples claiming ownership of the model through an API interface and analyzes the differences between their output and the expected labels. Our experimental results show that PWFed achieves a 99.9% watermark verification rate with only a 0.1–4.8% sacrifice of main task accuracy on the CIFAR10 dataset. Full article

Phage display has been widely used to identify peptides binding to a variety of biological targets. In the current work, we planned to select novel peptides targeting CD4 through screening of a commercial phage display library (New England Biolabs Ph.D.^TM-7). After three rounds of biopanning, 57 phage clones were Sanger-sequenced. These clones represented 30 unique peptide sequences, which were subjected to phage ELISA, resulting in the identification of two potential target binders. Following peptide synthesis, downstream characterization was conducted using fluorescence plate-based assay, flow cytometry, SPR, and confocal microscopy. The results revealed that neither of the peptides identified in the Sanger-based phage display selection exhibited specific binding toward CD4. The naïve library and the phage pool recovered from the third round of biopanning were then subjected to next-generation sequencing (NGS). The results of NGS indicated corruption of the selection output by a phage already known as a fast-propagating clone whose target-unrelated enrichment can shed light on the misidentification of target-binding peptides through phage display. This work provides an in-depth insight into some of the challenges encountered in peptide phage display selection. Furthermore, our data highlight that NGS, by exploring a broader sequence space and providing a more precise picture of the composition of biopanning output, can be used to refine the selection protocol and avoid misleading the process of ligand identification. We hope that these findings can describe some of the complexities of phage display selection and offer help to fellow researchers who have faced similar situations. Full article

This paper analyzes anti-corruption research and the body of knowledge produced by researchers in Southeast Europe (SEE). It compares it to the extant global anti-corruption literature. We consider all available scientific outputs from the SEE region, indexed by the Elsevier Scopus bibliometric database, and employ the Elsevier SciVal software solution to analyze the productivity and impact of the regional anti-corruption research. We also consider the influence of international cooperation on regional scientific productivity and its impact. The bibliometric comparison of the global and regional research and the authors’ qualitative assessment of these differences are performed. We also identify the implications and future research priorities. Full article

Image inpainting infers the missing areas of a corrupted image according to the information of the undamaged part. Many existing image inpainting methods can generate plausible inpainted results from damaged images with the fast-developed deep-learning technology. However, they still suffer from over-smoothed textures or textural distortion in the cases of complex textural details or large damaged areas. To restore textures at a fine-grained level, we propose an image inpainting method based on a hierarchical VQ-VAE with a vector credibility mechanism. It first trains the hierarchical VQ-VAE with ground truth images to update two codebooks and to obtain two corresponding vector collections containing information on ground truth images. The two vector collections are fed to a decoder to generate the corresponding high-fidelity outputs. An encoder then is trained with the corresponding damaged image. It generates vector collections approximating the ground truth by the help of the prior knowledge provided by the codebooks. After that, the two vector collections pass through the decoder from the hierarchical VQ-VAE to produce the inpainted results. In addition, we apply a vector credibility mechanism to promote vector collections from damaged images and approximate vector collections from ground truth images. To further improve the inpainting result, we apply a refinement network, which uses residual blocks with different dilation rates to acquire both global information and local textural details. Extensive experiments conducted on several datasets demonstrate that our method outperforms the state-of-the-art ones. Full article

Point clouds obtained with 3D scanners in realistic scenes inevitably contain corruption, including noise and outliers. Traditional algorithms for cleaning point cloud corruption require the selection of appropriate parameters based on the characteristics of the scene, data, and algorithm, which means that their performance is highly dependent on the experience and adaptation of the algorithm itself to the application. Three-dimensional object recognition networks for real-world recognition tasks can take the raw point cloud as input and output the recognition results directly. Current 3D object recognition networks generally acquire uniform sampling points by farthest point sampling (FPS) to extract features. However, sampled defective points from FPS lower the recognition accuracy by affecting the aggregated global feature. To deal with this issue, we design a compensation module, named offset-adjustment (OA). It can adaptively adjust the coordinates of sampled defective points based on neighbors and improve local feature extraction to enhance network robustness. Furthermore, we employ the OA module to build an end-to-end network based on PointNet++ framework for robust point cloud recognition, named R-PointNet. Experiments show that R-PointNet reaches state-of-the-art performance by 92.5% of recognition accuracy on ModelNet40, and significantly outperforms previous networks by 3–7.7% on the corruption dataset ModelNet40-C for robustness benchmark. Full article

In environments characterised by noise or the absence of audio signals, visual cues, notably facial and lip movements, serve as valuable substitutes for missing or corrupted speech signals. In these scenarios, speech reconstruction can potentially generate speech from visual data. Recent advancements in this domain have predominantly relied on end-to-end deep learning models, like Convolutional Neural Networks (CNN) or Generative Adversarial Networks (GAN). However, these models are encumbered by their intricate and opaque architectures, coupled with their lack of speaker independence. Consequently, achieving multi-speaker speech reconstruction without supplementary information is challenging. This research introduces an innovative Gabor-based speech reconstruction system tailored for lightweight and efficient multi-speaker speech restoration. Using our Gabor feature extraction technique, we propose two novel models: GaborCNN2Speech and GaborFea2Speech. These models employ a rapid Gabor feature extraction method to derive lowdimensional mouth region features, encompassing filtered Gabor mouth images and low-dimensional Gabor features as visual inputs. An encoded spectrogram serves as the audio target, and a Long Short-Term Memory (LSTM)-based model is harnessed to generate coherent speech output. Through comprehensive experiments conducted on the GRID corpus, our proposed Gabor-based models have showcased superior performance in sentence and vocabulary reconstruction when compared to traditional end-to-end CNN models. These models stand out for their lightweight design and rapid processing capabilities. Notably, the GaborFea2Speech model presented in this study achieves robust multi-speaker speech reconstruction without necessitating supplementary information, thereby marking a significant milestone in the field of speech reconstruction. Full article

Both executive corruption and corporate innovation are important factors affecting corporate development. This paper explores the impact of executive corruption on corporate innovation and examines the mechanism of their effects from the perspective of financing constraints. It is found that executive corruption significantly inhibits corporate innovation in general. In addition, financing constraints act as a mediator between executive corruption and corporate innovation, i.e., executive corruption exacerbates the financing constraints faced by firms and affects the access to and allocation of corporate resources, thus leading to a decrease in corporate innovation inputs and outputs. Further, the inhibitory effect of executive corruption on firm innovation is more pronounced in firms with low quality internal controls, strong professional background of executives, low quality external audit, low shareholding of institutional investors, strong political affiliation, and state-owned enterprises. Full article

In recent years, the study of corruption has become one of the most prominent in the social sciences. If there is corruption, however, it is because something has been corrupted; something pure has been sullied. This pure element serves mainly as a normative reference: It may never have constituted a social and political reality. However, the purpose of this article is to try to define what its components might be. In this way, theoretical considerations can be used to provide a more solid basis for the fight against corruption. The position of this paper is that the opposite of corruption should be explicitly defined without the use of abstract categories such as good governance or integrity. The paper will begin with a discussion of the concept of “non-corrupt government” and then proceed to a theoretical analysis of the main issues involved. It will conclude with some practical remarks on how to build, in the most parsimonious way, the benchmark of quality that corruption undermines. The contention is that a “non-corrupt government” is based on four principles: (1) equality (input side), (2) reasonableness (input side), (3) impartiality and professionalism of the administration (output side), and (4) accountability of the office (output side). Full article

Parallel natural language processing systems were previously successfully tested on the tasks of part-of-speech tagging and authorship attribution through mini-language modeling, for which they achieved significantly better results than independent methods in the cases of seven European languages. The aim of this paper is to present the advantages of using composite language models in the processing and evaluation of texts written in arbitrary highly inflective and morphology-rich natural language, particularly Serbian. A perplexity-based dataset, the main asset for the methodology assessment, was created using a series of generative pre-trained transformers trained on different representations of the Serbian language corpus and a set of sentences classified into three groups (expert translations, corrupted translations, and machine translations). The paper describes a comparative analysis of calculated perplexities in order to measure the classification capability of different models on two binary classification tasks. In the course of the experiment, we tested three standalone language models (baseline) and two composite language models (which are based on perplexities outputted by all three standalone models). The presented results single out a complex stacked classifier using a multitude of features extracted from perplexity vectors as the optimal architecture of composite language models for both tasks. Full article

This study examined the determinants of fiscal space within the Sub-Saharan Africa (SSA) region, utilising a panel of 33 countries from 2005 to 2021. The paper applied the panel threshold, difference, and system generalised method of moments (GMM) regression techniques. The empirical results found evidence of constrained fiscal space and poor governance in Central, Western, and Eastern Africa. The results further unveiled that an enhancement in governance indicators beyond −0.23 for the governance index, −0.15 for control of corruption, −0.98 for the rule of law, −0.37 for regulatory quality, −0.15 for voice and accountability, +0.36 for political stability, and −0.61 for government effectiveness, respectively, increase fiscal space. Moreover, the study concluded that the output gap, COVID-19, trade openness, and economic growth impact fiscal space availability in Central, Western, Southern, and Eastern Africa. The paper investigated whether the COVID-19 pandemic and governance quality significantly influenced fiscal space within SSA. We strongly recommend enhancement in all facets of governance through comprehensive restructuring of governance policies across all SSA countries. Another key recommendation is fostering trade openness to expand tax revenue generation and broaden the tax base, thereby providing the continent with greater fiscal space and improved resilience to unforeseen shocks. Full article