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
Unlocking MnO<sub>2</sub> electrolysis kinetics via dynamic chromium doping for efficient and stable zinc-manganese flow batteries
No full textZinc‑manganese redox flow batteries are attractive for energy storage due to their high energy density and low cost. However, the formation of poor conductivity and inactive MnO2 limits its reversibility at high areal capacity, leading to sluggish reaction kinetics and poor stability. In this study, the introduction of Cr3+ into the electrolyte for the dynamic doping process increases the oxygen vacancies in MnO2, generates more active electronic states and promotes the charge transfer kinetics, further optimizing the deposition and dissolution of MnO2. Therefore, the designed zinc‑manganese redox flow battery exhibits a discharge voltage of 1.98 V at 20 mA cm−2, along with a cycle life over 1800 cycles. Furthermore, the battery demonstrates a high areal capacity of 60 mAh cm−2, corresponding to an areal energy density of 105 mWh cm−2. This work effectively alleviates the issue of sluggish MnO2 electrolysis reaction kinetics, highlighting its potential in grid-scale energy storage applications.</p
Sub-Milliscale-Resolution Bimodal Tactile Sensor Array with Human-Skin-Like Graphesthesia Sensation
No full textMultimodal sensory integration advances the development of embodied intelligent systems and robotics with human-skin-like tactile perception. However, the difficulties in simultaneously achieving high resolution and multimodality hamper the exquisite tactile perception to differentiate information through touching. In this study, we report a sub-milliscale-resolution bimodal tactile sensor array consisting of a piezoelectric sensor array for mapping pressure magnitude distribution and a triboelectric sensor array for contact height detection, enabling the calculation of Young's modulus distribution. As compared to existing studies, the bimodal tactile sensor array achieved sub-milliscale spatial resolution of 700 µm and relatively high sensor density of 226 pixels/cm2, demonstrating fine-grained multimodal perception. By combining the pressure mapping information from the piezoelectric sensor array and contact height information from the triboelectric sensor array, with a rapid response time of 50 ms, the Young's modulus distribution can be revealed. Furthermore, the tactile sensor array can achieve human-skin-like graphesthesia sensation and reconstruct the softness-encrypted pattern with the assistance of deep learning algorithms, providing a paradigm-shift strategy of sub-milliscale-resolution tactile perception toward embodied intelligence and robotics.</p
Secure State Estimation against DoS Attack over SINR-based Channels: A Stackelberg Game Approach
No full textThis paper investigates defense strategies against Denial-of-Service (DoS) attacks in remote state estimation. When DoS attacks disrupt sensor communication, the signal-to-interference-and-noise ratio (SINR) decreases, thereby reducing the probability of successful state estimate transmission to the remote estimator and ultimately degrading estimation accuracy. The interaction between the sensor and attacker is modeled as a non-zero-sum Stackelberg game, where the smart sensor first determines a communication channel, and the attacker then chooses its energy allocation. The equilibrium of the Stackelberg game is investigated in two scenarios. In the static game scenario, each player aims to maximize the one-step objective, with equilibrium strategies derived via convex optimization. In the dynamic game scenario, each player seeks to maximize the long-term accumulated reward. A reinforcement learning algorithm is proposed to attain the equilibrium, and the convergence of optimal stationary strategies is proven. Additionally, structural properties of the optimal solutions are also analyzed. Finally, numerical simulations are provided to validate the feasibility and effectiveness of the proposed methods.</p
History and Philosophy of the Social Sciences
No full textThis book argues for a new direction of research, namely ‘History and Philosophy of the Social Sciences’ (HPSS). By adding a historical approach to the philosophy of the social sciences, the editor and authors show how fruitful HPSS is and simultaneously sheds new light on many central issues concerning the nature, development, and methodologies of the social sciences. This book revisits some classic works on the nature and methodology of the social sciences as well as offering a careful historical examination of the origins and development of a selection of social sciences and their distinctive methodologies. By reassessing the important debates in the philosophy of the social sciences as well as the methodological disputes between social scientists, the book provides a valuable resource to researchers and advanced students, but also to general readers.<br/
AI and Deep Learning for Terahertz Ultra-Massive MIMO: From Model-Driven Approaches to Foundation Models
No full textThis study explored the transformative potential of artificial intelligence (AI) in addressing the challenges posed by terahertz ultra-massive multiple-input multiple-output (UM-MIMO) systems. It begins by outlining the characteristics of terahertz UM-MIMO systems and identifies three primary challenges for transceiver design: computational complexity, modeling difficulty, and measurement limitations. The study posits that AI provides a promising solution to these challenges. Three systematic research roadmaps are proposed for developing AI algorithms tailored to terahertz UM-MIMO systems. The first roadmap, model-driven deep learning (DL), emphasizes the importance of leveraging available domain knowledge and advocates the adoption of AI only to enhance bottleneck modules within an established signal processing or optimization framework. Four essential steps are discussed: algorithmic frameworks, basis algorithms, loss function design, and neural architecture design. The second roadmap presents channel state information (CSI) foundation models, aimed at unifying the design of different transceiver modules by focusing on their shared foundation, that is, the wireless channel. The training of a single compact foundation model is proposed to estimate the score function of wireless channels, which serve as a versatile prior for designing a wide variety of transceiver modules. Four essential steps are outlined: general frameworks, conditioning, site-specific adaptation, and the joint design of CSI foundation models and model-driven DL. The third roadmap aims to explore potential directions for applying pretrained large language models (LLMs) to terahertz UM-MIMO systems. Several application scenarios are envisioned, including LLM-based estimation, optimization, search, network management, and protocol understanding. Finally, the study highlights open problems and future research directions.</p
Channel-adaptive generative reconstruction and fusion for multi-sensor graph features in few-shot fault diagnosis
No full textRecently, multi-sensor feature fusion has been proven to be an effective strategy for improving the accuracy of few-shot fault diagnosis. However, existing fault diagnosis models based on multi-sensor feature fusion often overlook significant inter-channel discrepancies and struggle to mitigate noise pollution inherent in multi-source signals. To address these limitations, this paper proposes a channel-adaptive generative reconstruction and fusion framework that integrates a contrastive variational graph autoencoder feature fusion (CogFusion) module for robust few-shot fault representation learning. The CogFusion module leverages the generative capability of a contrastive variational graph autoencoder (CGE) to reconstruct noise-suppressed node features while explicitly modeling latent distributions of multi-sensor signals. By incorporating a multi-channel parallel graph contrastive learning strategy, CogFusion enhances discriminative feature separation by contrasting topological structures of positive and negative sample pairs, effectively isolating fault-related patterns from noisy embeddings. To adaptively fuse multi-channel information, a channel discrepancy-guided weighting mechanism dynamically prioritizes high-credibility sensor features, mitigating the impact of low-quality data. To further enhance feature learning in few-shot diagnosis, a dual-scale topological Transformer (DSTT) model is introduced to deeply mine the reconstructed multi-channel topological graph, enabling high-precision few-shot fault diagnosis. Experimental results on the axial flow pump and HUSTgearbox datasets demonstrate that the proposed method outperforms both single-channel and existing multi-sensor feature fusion methods, highlighting its superiority in feature fusion and cross-channel information integration.</p
Stabilized explicit material point method for fluid flow and fluid-structure interaction simulations using dual high-order B-spline volume averaging
No full textTraditional explicit Material Point Methods (MPM) for weakly compressible fluids often suffer from volumetric locking, cell-crossing instability, and excessive energy dissipation, particularly in fluid-structure interaction (FSI) scenarios. This study presents a stabilized explicit MPM framework that integrates dual high-order B-spline volume averaging to address these challenges. The proposed dual averaging technique simultaneously smooths deformation gradients and pressure fields using cubic B-spline basis functions to eliminate cell-crossing errors and reduce volumetric locking. A blended APIC/FLIP mapping scheme is developed to enhance energy conservation and stability at coarse grid resolutions. The framework is further enhanced by seamlessly integrating various complementary techniques such as δ-correction, pressure smoothing, and specialized boundary handling for more robust and effective modeling of free-surface and FSI problems. The framework is rigorously validated through benchmark cases, including 1D elastic wave propagation, Poiseuille flow, lid-driven cavity flow, water sloshing, dam break, and water impact on elastic obstacles. The simulation results demonstrate a remarkable reduction in pressure oscillations and improved particle distribution uniformity compared to prior MPM approaches. The proposed method establishes a robust and efficient tool for large-deformation FSI problems and bridges gaps in accuracy and stability for industrial-scale applications.</p
The dosage of deception: How frequency and type influence trust evaluations
No full textLeading negotiation scholars have recommended that individuals never lie to their counterpart. This advice is based on negotiations research that has examined the interpersonal costs of deception through studies where a target is categorized as being deceptive or honest without consideration of the relative frequency of the deception. For example, prior work has broadly categorized individuals who lie once in a single-issue negotiation and individuals who lie once in a five-issue negotiation as liars. Consequently, it is hard to disentangle how many of the theoretical and prescriptive claims pertain to using deception sparingly, frequently, or only being deceptive. Across five preregistered studies (N = 4003), I examine contexts where individuals negotiate over multiple issues and disentangle the effects of being sparingly, mostly, or exclusively deceptive. Examining diverse deception strategies (e.g., lies by commission, dodging, paltering, deflection), I find that the economic and interpersonal consequences of deception are significantly different depending on the relative frequency with which individuals use it, underscoring the need to not only understand the effects of deception, but also the dosage. Although individuals punish deception, they also reward honesty, and are forgiving of counterparts who use deception sparingly. Combined, these findings deepen our understanding of deception and trust and advance our theoretical and prescriptive understanding of negotiations.</p
Synergizing AIE Luminogens and hydrogel matrices: Advancing precision medicine through smart material design
No full textThe integration of aggregation-induced emission (AIE) luminogens (AIEgens) with hydrogel materials has significantly expanded their applications in precision medicine, fostering groundbreaking innovations in biomedicine. AIEgens exhibit exceptional photostability, a high signal-to-noise ratio, and stimuli-responsive properties making them uniquely suited for constructing theranostic platforms for image-guided diagnosis and therapy. Meanwhile, hydrogels provide an ideal matrix for precision diagnostics and therapeutics due to their outstanding biocompatibility, tunable physicochemical properties, and capacity to serve as carriers for drugs and molecular probes. Given the immense potential of AIEgen-hydrogel composite systems in precision medicine, this review systematically summarizes recent advances in their applications, including cancer-targeted therapy, biofluorescent probes, antibacterial treatment, biomimetic cell culture, biomarker detection, and drug delivery/release monitoring. Furthermore, we critically analyze the key challenges hindering their widespread adoption and discuss future research directions. Together, AIEgen-integrated hydrogel systems represent a transformative approach in precision medicine, driving significant progress through their versatile biomedical functionalities.</p
Inflammation-targeted nanoplatform: NIR-II imaging-guided encephalitis suppressing by dual antioxidant-ferroptosis action
No full textEncephalitis, a life-threatening neurological disorder with high mortality and debilitating long-term sequelae, remains a formidable clinical challenge due to limited therapeutic strategies targeting its underlying pathological mechanisms. Current interventions, constrained by blood-brain barrier (BBB) impermeability and a focus on symptomatic relief, fail to mitigate ferroptosis and reactive oxygen species (ROS)-mediated neurotoxicity—central drivers of disease progression. Here, we present CeO2/3TT@NP-RVG, a multifunctional nanomaterial engineered for integrated diagnosis and treatment of encephalitis. The nanoplatform combines ROS-scavenging cerium oxide (CeO2) with photothermal NIR–II–emissive (3TT) nanoparticles, enabling real-time fluorescence imaging of encephalitis with deep-tissue resolution. Functionalization with rabies virus glycoprotein-derived RVG peptide ensures BBB penetration and neuron-targeted delivery. In LPS-induced encephalitis models, CeO2/3TT@NP-RVG demonstrated dual therapeutic efficacy: alleviating oxidative stress by neutralizing ROS, suppressing pro-inflammatory cytokines (TNF-α, IFN-β), and inhibiting ferroptosis via ubiquitination-mediated downregulation of the POR-ACSL4-LPCAT3 pathway, thereby reducing polyunsaturated fatty acid peroxidation. These synergistic actions significantly improved survival rates and mitigated neuroinflammation. Our findings highlight CeO2/3TT@NP-RVG as a pioneering theranostic platform that bridges molecular mechanism-based therapy with precision imaging, offering a transformative strategy for encephalitis and related neurological disorders.</p