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
Multiscale modeling of coupled thermo-hydro-mechanical- chemical behavior in hydrate-bearing sediment
No full textThis study presents a hybrid continuum-discrete multiscale computational framework that integrates the material point method (MPM) and the discrete element method (DEM) to model fully coupled thermo-hydro-mechanical-chemical (THMC) behavior of hydrate-bearing sediments (HBS). Key innovation of the framework lies in its direct use of DEM to model microscale mechanisms, such as hydrate bond degradation, particle rearrangement, and pore evolution, thereby bypassing the need for conventional elastoplastic constitutive models to define effective stress. We show that a simple hydrate saturation-dependent contact model within the DEM can effectively reproduce characteristic shear and volumetric responses of HBS under various hydrate saturation and confining stresses. By embedding a DEM-based representative volume element (RVE) at each material point in the MPM grid, microscale mechanical behaviors are seamlessly homogenized to inform large-deformation macroscale multiphysics processes. Numerical simulations of biaxial compression and indenter penetration demonstrate the framework's capability to capture critical phenomena, including shear band formation, shear-induced dilation, and the generation of negative excess pore pressure that drives localized hydrate dissociation. The results further reveal that while higher hydrate saturation enhances shear strength, it also promotes brittle failure and intensified dissociation. Conversely, increased confining stress suppresses volumetric dilation and stabilizes the sediment by mitigating the development of negative pore pressure. This multiscale approach provides a powerful new tool for elucidating complex THMC interactions in HBS, with important implications for assessing hydrate-related geohazards and optimizing gas extraction strategies.</p
2D perovskites featuring intralayer bidentate ligands
No full textMaking efficient and stable metal halide perovskites typically involves challenging trade-offs between structural integrity and performance. Now, a series of two-dimensional perovskites featuring intralayer bidentate coordination ligands has been developed, providing an extendable molecular approach to strengthen the structure and modulate the performance of these hybrid materials and their analogues.</p
Mitigating Server-Side Communication Bottlenecks in Distributed Learning With Round-Robin Participant Coordination
No full textDeep neural networks are increasingly trained in a distributed manner—in either clusters or with federated devices, where the participants jointly refine the global model with their gradients calculated locally. More often than not, those gradients are collected to a central server in a synchronous manner to avoid the negative impact of stale updates. However, when all the participants communicate their gradients to the server in such a uniform pace, the network on the server side—under intense contention—often becomes a performance bottleneck. To address this problem, for the cluster environment we propose the Round-Robin Synchronous Parallel (R2SP) scheme, which coordinates the participants to make updates in an evenly-gapped, round-robin manner. This way, we can minimize the network contention with a minimum cost of the update quality; we also propose to incorporate adaptive batch sizing in R2SP to address the hardware heterogeneity among workers. Moreover, for the federated learning (FL) scenarios, we note that it is necessary yet challenging to apply the insight of R2SP to mitigate the network bottleneck in the FL server, given that there are a huge number of participants with unstable resources and inconsistent data distributions. To tackle those challenges, we further propose FL-R2SP, which extends the coordination units from individual participants to participant groups—with the resource instability and data heterogeneity tackled within each group. We have implemented R2SP and FL-R2SP respectively with TensorFlow and PyTorch, and extensive EC2 experiments show that R2SP and FL-R2SP can respectively speed up model convergence for clustered and federated scenarios by over 20%.</p
Direct Laser Writing of Functional Materials for Wearable Human-Machine Interaction: Mechanisms, Materials, and Applications
No full textDirect laser writing (DLW) uses tightly confined laser beams to trigger photothermal, photochemical, and ultrafast photophysical processes. It enables mask-free, sub-micrometer patterning of functional materials on diverse substrates under ambient conditions. Wide material compatibility, fine feature control, and low heat input make DLW attractive for wearable human-machine interaction (HMI). These systems are soft, skin-conformal devices that monitor physiological and environmental signals in real time. This review surveys DLW-derived materials and architectures for wearable HMI devices. It provides a system-level perspective and highlights manufacturing-structure-property-performance relationships that guide the design of DLW-enabled wearables. We first outline the main laser-material interaction mechanisms, then classify DLW-derived materials into six families: carbon-based frameworks, metals and alloys, metal oxides, polymers and composites, 2D materials, and hybrids. For each family, we link manufacturing routes, microstructures, key properties, and device performance. Next, we map these materials onto five building blocks of wearable platforms: sensors, energy modules, communication components, feedback interfaces, and fully integrated systems. We cover mechanical, pressure, strain, humidity, temperature, gas, optical, biochemical, and multimodal sensors. Performance is benchmarked in terms of sensitivity, detection limit and range, response and recovery time, and stability under realistic deformation and wear. Finally, we discuss remaining challenges and future directions. Key issues include throughput and scalability, mechanical durability, biocompatibility, and standards for long-term reliability. We highlight emerging strategies such as multi-beam and roll-to-roll DLW, hybrid workflows with printing and 3D structuring, sustainable precursors and advanced nanocomposites, and AI-guided design and process control. Together, these directions aim to accelerate DLW-enabled wearables from laboratory prototypes to everyday wearable HMI systems
Towards Trustworthy Dialogue Systems With Advanced Out-of-Scope Intent Detection Model
No full textThe growing demand for trustworthy dialogue systems emphasizes the need for consistently accurate responses to user inputs. The first step in developing a trustworthy dialogue system is detecting user inputs with the out-of-scope (OOS) intent. Advanced research on OOS intent detection enhances effectiveness by using data augmentation to generate numerous artificial OOS samples from a limited set of true OOS data, modelling its distribution for training. However, data augmentation presents challenges, including higher costs, increased time, and a greater risk of overfitting. Additionally, current studies treat the OOS intent as a homogeneous category equivalent to known intents within a classification framework, overlooking the inherent diversity of OOS intents. To tackle these challenges, we introduce a novel method called Anchor-Integrated Dynamic Out-of-scope Intent Learning (AIDOIL), which integrates the selected anchor to represent the OOS intent adapting to diverse inputs dynamically. The intent representations transform the global classification problem into a matching task that determines if a user input aligns with each intent. This eliminates the necessity to augment OOS data and accommodate the diversity of OOS intents through dynamic representation learning. We conducted extensive experiments on three public dialogue datasets, demonstrating that AIDOIL achieves an average 7.21% improvement in OOS detection accuracy, while maintaining an acceptable increase in training time.</p
Simulation Study on the Scalability of Channel-All-Around Reconfigurable Field-Effect Transistors With Gate-Controlled Polarity
No full textConventional complementary metal oxide semiconductor (CMOS) devices rely heavily on doping, which increasingly limits scalability due to process constraints and performance degradations at advanced technology nodes. To overcome the drawbacks associated with doping, reconfigurable field-effect transistors (RFETs) that employ ferroelectric gate dielectrics with non-volatile programmability have emerged as a promising alternative for gate-controlled polarity modulation. Nevertheless, most reported RFETs adopt planar device geometries, raising concerns regarding their scalability at deeply scaled nodes. This work proposes a channel-all-around (CAA) RFETs architecture featuring gate-tunable polarity, based on an undoped WSe2 channel and an AlScN gate dielectric. Using calibrated TCAD simulations, we show that vertically stackable CAA structures, combined with intrinsically ambipolar WSe2, have significantly enhanced the scalability of RFETs for logic applications down to N0.5 technology node. Furthermore, the extracted device characteristics are implemented in a Verilog-A model for circuit-level simulations. The CAA-RFET-based complementary inverters exhibit robust noise margins, high voltage gains, and stable operation voltages at supply voltages down to 0.2 V. The reconfigurable CMOS logic gates with topologies identical to conventional CMOS designs, confirm the extreme scalability, and circuit-level viability of 2D CAA-RFETs for ultra-compact and energy-efficient programmable logic.</p
The Simple Essence of Boolean-Algebraic Subtyping: Semantic Soundness for Algebraic Union, Intersection, Negation, and Equi-recursive Types
No full textBoolean-algebraic subtyping (BAS) is a powerful subtyping approach introduced in 2022 as the "secret sauce" enabling backtracking-free principal type inference in the MLstruct research language, a structurally-typed functional programming language with tagged records, tag and record subtyping, and tag-based pattern matching. By supporting distributive intersection, union, negation, and equi-recursive types, MLstruct can express powerful programming patterns, such as subtyped extensible variants, without needing row variables. But the use of atypical subtyping rules that violate some interpretations of intersection and union types, the mutual distributivity between these types, and the complexity of coinductive reasoning for equi-recursive types have collectively made the study of BAS difficult. The syntactic soundness proofs provided in the original work are dauntingly complicated and long-winded, obscuring the intuitions behind the correctness of BAS. In this paper, we distill the simple essence of Boolean-algebraic subtyping: we discover that BAS can be understood through five families of characteristic Boolean homomorphisms defined on types in context. Two of these map to power sets of simpler objects; the rest map back to types, but under an unguarded coinductive assumptions context. Together, these homomorphisms let us prove rather directly that BAS is sound, in that it does not relate constructors of incompatible runtime shapes. These homomorphisms are characteristic in the sense that they are sufficient to capture the meaning of subyping: we prove that if an inequality holds between two types under all these homomorphisms, then subtyping holds between the two types in the original context. This directly suggests a new subtyping decision procedure for BAS, which avoids some inefficiencies in the original algorithm, although it still has exponential worst-case time complexity. We prove that the subtyping problem is in fact co-NP-hard even without recursive types. Finally, we discover that BAS is already powerful enough to encode the removal of a field from a type. This allows us to support extensible records through one new term form and one new typing rule, but, perhaps surprisingly, no changes to subtyping at all. Our new approach to the semantics of BAS sheds some light on the core of MLstruct’s type system. It could be adapted to other languages with algebraic flavors of subtyping, such as Scala 3 and Ceylon, making their design and verification more approachable. Tellingly, all our subtyping soundness proofs fit inside the main body of this paper, with only some administrative lemmas relegated to the appendix.</p
Mutual Information-Empowered Task- Oriented Communication: Principles, Applications and Challenges
No full textMutual information (MI)-based guidelines have recently proven to be effective for designing task-oriented communication systems, where the ultimate goal is to extract and transmit task-relevant information for downstream tasks. This paper provides a comprehensive overview of MI-empowered task-oriented communication, highlighting how MI-based methods can serve as a unifying design framework in various task-oriented communication scenarios. We begin with the roadmap of MI for designing task-oriented communication systems, and then introduce the roles and applications of MI to guide feature encoding, transmission optimization, and efficient training with two case studies. We further provide a deeper examination of the limitations and challenges associated with MI-based methods. Lastly, we highlight several open issues in MI-based task-oriented communication to inspire future research.</p
LEX v1.6.0: a new large-eddy simulation model in JAX with GPU acceleration and automatic differentiation
No full textLarge-eddy simulations (LES) are essential tools for studies on atmospheric turbulence and clouds and play critical roles in the development of turbulence and convection parameterizations. Current numerical weather models have approached kilometer-scale resolution as supercomputing facilities advance. However, this resolution range is in the so-called gray zone, where subgrid-scale (SGS) turbulence actively interacts with resolved motion and significantly influences the large-scale characteristics of simulated weather systems. Thus, a novel LES framework is required to enable the development of new SGS approaches for the gray zone. Here we used the Python library JAX to develop a new LES model. It is based on the generalized pseudo-incompressible equations formulated by Durran (2008). For a classic warm bubble case, the traditional Smagorinsky model fails to reproduce the correct structure evolution of the warm bubble, though it can modestly correct the rising speed in gray-zone resolution simulations. Utilizing the capability of JAX for automatic differentiation, we trained a deep learning-based SGS turbulence model for the same case. The trained deep learning SGS model, based on a simple autoencoder (AE), enables this physics-deep learning hybrid model to accurately simulate the expansion of the thermal bubble and the development of rotors surrounding the center of the bubble at a gray-zone resolution. The gray-zone simulation results are comparable to those of the benchmark LES resolution.</p
How Background Colour Shapes Digital Text-Information Processing: A fNIRS-Eye Tracking Study
No full textAs digitalization advances, colour in human-computer interaction now fundamentally shapes digital text processing beyond aesthetics. However, the background colour's impact on text information processing under controlled luminance conditions remains underexplored. This study employed a multimodal approach combining functional near-infrared spectroscopy (fNIRS), eye tracking, behavioural performance and subjective reports to systematically examine colour’s effects on text comprehension. Key findings demonstrated that green elicited higher activation in the left frontal pole area (FPA-L), right Broca’s area (Broca-R), and the left dorsolateral prefrontal cortex (DLPFC-L), corresponding to the highest accuracy. In contrast, red and blue reduced brain activation, arousal levels, accuracy, and prolonged average visit duration. Correlational analyses revealed critical underlying mechanisms, showing that colour difference negatively correlated with brain activation (DLPFC-L, Broca-R), pupil diameter, and subjective reports. These findings highlight the importance of colour in text information processing and provide actionable insights for optimizing colour choices in digital interfaces.</p