Hong Kong University of Science and Technology
Hong Kong University of Science and Technology Institutional RepositoryNot a member yet
162821 research outputs found
Sort by
From delocalized orbitals to Lewis structures: Trade-offs in the imperfect art of chemical bonding analysis
No full textOrbital-based bonding analysis has become an essential component of structure and bonding studies in chemistry. However, the proliferation of bonding analysis tools has led to an overwhelming variety of results presented in many application studies, often lacking a clear framework for interpretation. In this work, we critically examine the desirable features of these analysis tools and explore the inherent limitations that prevent a single tool from satisfying all criteria. By discussing the trade-offs commonly encountered in bonding analysis, we aim to provide insights that will guide the future development and selection of more effective analysis tools.</p
How can we trust TROPOMI based methane emissions estimation: calculating emissions over unidentified source regions
No full textWe propose a novel method for computing the effects of TROPOMI observational uncertainties on emissions calculation arising from the nonlinearity of the gradient terms and non-biased filtering in space and time. Application using TROPOMI XCH4 data in clean areas of Western China with long-term WMO background observations quantifies a minimum detectable emission threshold of 0.3 µgm-2 s-1, lower than existing community thresholds using TROPOMI. By combining threshold-based and stochastic approaches that incorporates pixel-by-pixel and day-by-day XCH4 uncertainties, we identify and filter physically unrealistic emission values in both space and time. The resulting emissions reveal both missing sources and emission bias caused by the nonlinearity of the gradient term. Validation was performed by applying the method to the Permian Basin, where comparisons with airborne observations demonstrate the method's ability to align with independent datasets. The importance and implications of our results are related to this being a new methodology for methane emissions estimate from TROPOMI which enables precise identification of emission sources and improved handling of observational noise, offering a more accurate framework for methane emission monitoring across diverse regions using existing satellite platforms. Our results yield a non-negative emissions dataset using an objective selection and filtration method, which includes a lower minimum emissions threshold on all grids and reduction of false positives. Finally, the new approach can be adopted to other satellite platforms to provide a more robust and reliable quantification of emissions under data uncertainty that moves beyond traditional plume identification and background subtraction.</p
MCFL: Multimodal Collaborative Fusion Learning for Hashimoto’s Thyroiditis Recognition
No full textUltrasound imaging and biochemical examinations are the primary methods for diagnosing Hashimoto’s thyroiditis (HT). However, neither of them is sufficient to accurately diagnose HT alone. Most existing multimodal models for HT diagnosis focus primarily on extracting and concatenating features from different modalities, which are ineffective due to the dimensional imbalance of the features between the textual and image data. To address this issue, we propose a novel Multimodal Collaborative Fusion Learning (MCFL) approach, which can enhance and recalibrate the biochemical indicators using ultrasound images, effectively improving the significance and specificity of biochemical indicators for the diagnosis of HT. Specifically, MCFL first constructs a novel INNet to convert the image-level characteristics of the HT ultrasound image into two numerical indicators, i.e., the Local prominent inflammatory (Lpi) and the Global diffuse lesion (Gdl), unifying image data and textual data into a single representation space. Then, a decision tree-based optimization strategy is employed to supervise the training of INNet, interactively recalibrating biochemical indicators with the guidance of the two numerical indicators mentioned above and obtaining a more accurate feature representation of HT. Finally, based on the deep Q-learning framework, a reward mechanism is established to guide the HT diagnostic process, in which the experience replay mechanism and the (Formula presented) strategy are utilized collaboratively to improve the accuracy and robustness of the model. Extensive experiments are conducted on a multimodal dataset from multiple medical centers, and the results demonstrate that MCFL achieves state-of-the-art performance, setting a new benchmark.</p
Resilient Control and Estimation of Networked Control Systems
No full textWith advances in multiple access technologies and wireless communications, networked control systems (NCSs) play a pivotal role in the modern automation and control industry, enabling seamless communication and control across remotely distributed systems. However, due to the openness of wireless networks, the integration of networked components also introduces unprecedented security challenges to the physical plant. This chapter provides a comprehensive examination of the state-of-the-art techniques, methodologies, and advancements in ensuring the security of networked control systems. By exploring a range of security threats, vulnerabilities, and corresponding countermeasures, this chapter aims to provide insights into the current landscape of NCS security, addressing critical issues such as confidentiality, integrity, availability, and resilience. Understanding and addressing these challenges is crucial for safeguarding the integrity and reliability of networked control systems in the face of evolving cyber threats.</p
Directing environmental innovation toward radical clean technologies for sustainable transitions: Market-based vs. command-and-control policies
No full textEnvironmental regulations present firms with two innovation pathways: developing abatement technologies that mitigate pollution from existing processes or investing in clean technologies that do not produce pollution by design. Achieving disruptive sustainability requires radical clean technologies—fundamental departures from established trajectories that enable system-level shifts. However, pursuing radical clean innovation demands substantial financial investment, organizational restructuring, and operational flexibility. While existing research examines how policy instruments affect the rate of environmental innovation, we lack an understanding of how they shape direction—whether they steer firms toward clean rather than abatement technologies, and whether they foster more radical clean innovation. We argue that market-based instruments provide long-term incentives and technological flexibility that enable transformative innovation, while generating market-size effects that shrink demand for polluting technologies and expand markets for clean alternatives. This creates strategic incentives for radical innovation. In contrast, command-and-control policies impose fixed compliance thresholds without altering market structures or generating continuous pressure for innovation. We test these hypotheses using environmental patenting data from manufacturing firms across 17 EU countries from 1991 to 2020. Results demonstrate that market-based policies significantly enhance clean technology innovation and increase technological radicalness, while command-and-control policies show no significant effect on environmental innovation. These findings contribute to research on disruptive sustainability and policy-induced innovation by demonstrating how market-based instruments catalyze radical clean technologies and support sustainability transitions.</p
An effective implementation of high-order compact gas-kinetic scheme on structured meshes for compressible flows
No full textA novel fifth-order compact gas-kinetic scheme is developed for high-resolution simulation of compressible flows on structured meshes. Its accuracy relies on a new multidimensional fifth-order compact reconstruction that uses line-averaged derivatives to introduce additional degrees of freedom, enabling a compact stencil with superior resolution. For non-orthogonal meshes, reconstruction is performed on a standard reference cell in a transformed computational space. This approach provides a unified polynomial form, significantly reducing memory usage and computational cost while simplifying implementation compared to direct multi-dimensional or dimension-by-dimension methods. A nonlinear adaptive method ensures high accuracy and robustness by smoothly transitioning from the high-order linear scheme in smooth regions to a second-order scheme at discontinuities. The method is implemented with multi-GPU parallelization using CUDA and MPI for large-scale applications. Comprehensive numerical tests, from subsonic to supersonic turbulence, validate the scheme’s high accuracy, resolution and excellent robustness.</p
Targeting hypersialylation via lectin-directed protein aggregation therapy (LPAT) for anti-metastasis applications
No full textHere, we report the development of lectin-directed protein aggregation therapy (LPAT), which combines the strong glycan-targeting capabilities of multivalent lectins with the aggregating propensities of bacterial microcompartment proteins. The design aims to create a system sensitive enough to elicit cell-specific aggregation towards invasive, metastatic tumor cells, while being nontoxic to normal tissues. LPAT agents were screened against a panel of 6 breast cancer cell lines, with the most potent agent showing preferential anti-adhesive and anti-invasive activity against the hypersialylated/MMP9 overexpressing MDA-MB-231 cell line. Furthermore, LPAT agents did not exhibit any propensity for hemagglutination, a principal disadvantage of lectin-based targeting systems. Subsequent studies using a metastatic mouse model showed that LPAT agents could prevent the formation of experimental lung metastases caused by the highly metastatic MDA-MB-231-LM2 isoform cell line. Overall, this work has laid the foundation for a potential glycan-targeting therapy aimed at preventing the onset and progression of metastatic tumors in a safe and selective manner.</p
The effect of glucagon-like peptide 1 (GLP-1) receptor agonists on cognition: A systematic review of systematic reviews and meta-analyses
No full textBackground Results from extant studies indicate that type 2 diabetes mellitus (T2DM) is associated with decreased cognitive function and increased risk for dementia, notably Alzheimer's Disease (AD). Herein, we aimed to evaluate the effect of glucacon-like peptide-1 receptor agonists (GLP-1 RAs) on cognitive measures in persons with cognitive impairment and T2DM from available systematic reviews and meta-analyses. Methods A literature search was conducted using Web of Science and MEDLINE from inception to July 15, 2025. Article screening and data extraction were conducted by three reviewers (H.B., C.D., S.L.). Primary research studies examining race, sex, children, animals and neurodegenerative diseases other than AD were excluded. Results Nine systematic reviews and meta-analyses examining the effect of GLP-1 RAs on cognitive function in adults with T2DM were included. Evidence suggests that GLP-1 RAs are associated with a reduction in overall cognitive decline in adults with T2DM and dementia/AD. Replicated findings from meta-analyses indicated that GLP-1 RAs improved performance on cognitive assessments (total learning; p = 0.039; p < 0.00001). Some meta-analyses observed a change in cognitive measures but lacked sufficient statistical significance (p > 0.05). Conclusions GLP-1 RAs positively affect cognitive function in AD patients with T2DM. However, their efficacy on disparate cognitive domains requires further replication in larger scale controlled clinical trials.</p