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
MTSCCleav: a Multivariate Time Series Classification (MTSC)-based Method for Predicting Human Dicer Cleavage Sites
No full textMicroRNAs (miRNAs) are small non-coding RNAs (ncRNAs) that regulate gene expression at the post-transcriptional level, thereby playing essential roles in diverse biological processes. The biogenesis of miRNAs requires Dicer to cleave at specific sites on the precursor miRNAs (pre-miRNAs). Several machine learning approaches have been proposed to predict whether an input sequence contains a cleavage site. However, they rely heavily on complex feature engineering or opaque deep neural networks. It results in a lack of generalizability and a long running time. There is a need for an alternative modeling paradigm that is accurate, fast, and simple. We propose an approach to frame the task as a multivariate time series classification problem. Nine encoding methods have been proposed to convert the RNA sequence into a time series. The predicted secondary structure is also converted to a time series. We also leverage the probabilities of the base pairs in the predicted secondary structure. Computational experiments demonstrate that our proposed method can achieve better or comparable results in terms of using a simpler, more intuitive model and less computational time. It achieves 3.7X to 28.8X speedup. Through perturbation experiments, we found that regions close to the center of pre-miRNAs are essential for predicting human Dicer cleavage sites. By transforming the RNA sequence and its secondary structure information into a multivariate time series and utilizing simple, state-of-the-art time series classifiers, we achieved comparable or even superior performance in a simpler and faster manner.</p
Strategic Communications with Socializing Agents Under Unknown Public Health Threats
No full textProblem definition: This paper investigates how governments can design optimal public health policies to inform and guide the public amid uncertain health threats. To capture heterogeneity in social behavior, we introduce a class of socializing agents and examine how the government strategically combines two policy instruments—persuasive communication (messages) and physical or monetary penalties—to incentivize compliance with social restrictions. Methodology/results: We develop a game-theoretic model in which the government commits in advance to both messaging and penalty strategies. The optimal policy exhibits a nonmonotonic structure with respect to the pandemic severity, alternating between the use of messages and penalties. Messages are shown to be most effective when the severity of the pandemic is either mild or moderate to high. Interestingly, socializing agents can indirectly promote compliance among traditional agents because of negative externalities, and the government may reduce penalty levels as pandemic severity increases. Managerial implications: Our findings underscore the strategic value of coordinating messages and penalties as complementary tools in public health policy. When the divergence between individual and governmental incentives is small, costless messages— especially those delivering finely granulated information—can effectively influence public behavior. Notably, we identify a dual role for state-contingent penalties not only in enhancing compliance but also in signaling pandemic severity. Overall, by examining the interplay of multiple policy instruments across different dimensions, our results highlight the importance of behavioral heterogeneity and government credibility in shaping public health policies under competing societal objectives.</p
Moisture from US Corn Belt fuels more intense convective storms
No full textThe US Corn Belt is among the world’s most productive agricultural regions and a global hotspot for mesoscale convective systems (MCSs), which supply vital growing season rainfall but also drive hazardous flooding. While evapotranspiration (ET) from shallow groundwater, extensive croplands, and irrigation is known to influence regional precipitation, its role in fueling convective storms remains poorly understood. Here we use a high-resolution regional climate model coupled with an advanced water vapor tracer to track moisture from Corn Belt ET into individual convective storms. We find that the integrated groundwater-crop-irrigation interactions amplify MCS frequency by 24-35%, extend storm lifetime by up to 10%, and accelerate storm movement. Moisture from Corn Belt ET enhances the warm-moist inflows, sustaining convective cells and enhancing precipitation near the storm center, with more pronounced effects in stronger storms. These findings highlight the significant role of shallow groundwater and agricultural activities in intensifying convective storms, creating cascading hazards that threaten water and food security.</p
Thermo-Adaptive Biointerfacial Electrodes Enable Dynamic and Reusable ECG Monitoring
No full textUnstable skin–electrode interfaces hinder dynamic electrocardiogram (ECG) monitoring during exercise. Herein, a thermo-adaptive biointerfacial (TAB) electrode is developed using a physiological temperature-driven molecular–microstructural synergy (PTD-MMS) strategy, integrating a phase-change polymer, biomimetic trapezoidal microchannels, and a conductive Ag-fabric network to address this bottleneck. At body temperature, this design achieves strong, reversible adhesion (39.6 N m–1, 12.4 switching ratio) while retaining over 90% of initial strength after 25 reattachment cycles, enabling robust fixation yet gentle detachment. Biomimetic microchannels enable active sweat management with a 58.6 g m–2 h–1 water vapor transmission rate and 7.5 μL s–1 liquid removal to preserve stable skin contact. Outperforming conventional gel electrodes under movement, the TAB electrode delivers high-fidelity ECG signals (67.9 dB SNR) and maintained reliable monitoring during standing, walking, and football training. Notably, heart rate variability correlated with body mass index in the cohort, supporting potential for personalized health supervision, while this PTD-MMS approach establishes a targeted platform for adaptive bioelectronic interfaces in dynamic exercise monitoring.</p
MFSPart: A Generalized Partitioning Framework for Multi-FPGA Systems and Its Ensemble-Based Extension
No full textMulti-FPGA systems (MFSs) are essential for prototyping very large-scale integration (VLSI) designs, necessitating efficient partitioning techniques. Optimizing inter-FPGA communication—particularly minimizing hop count—is critical for improving performance and reducing latency. However, existing topology-aware partitioning methods primarily minimize driver–sink pair cut size under a strict one-hop constraint, which may not always yield feasible solutions. They also typically overlook practical constraints such as FPGA capacity limits and I/O resource optimization, leading to solutions with excessive pin usage and routing complexity. To address these limitations, we propose a unified formulation that jointly optimizes driver–sink pair cut size, connectivity, and mean hop count under maximum-hop and capacity constraints. Unlike prior frameworks that enforce hop-based constraints via propagation before coarsening—imposing significant limitations— we postpone propagation until after coarsening. Based on this, we develop MFSPart, a framework applicable to both fixed-node and non-fixed-node scenarios, featuring improved coarsening, initial partitioning, and refinement strategies. We further propose an enhanced version, MFSPart-Ensemble, which systematically combines high-quality structures from diverse solutions across uncoarsening stages to generate superior partitions. Experiments demonstrate that MFSPart achieves excellent performance across all three objectives with reduced runtime, while MFSPart-Ensemble further improves solution quality, especially for complex FPGA systems.</p
Event-Triggered State Estimation, Control and Learning
No full textEvent-triggered sampling and control provides an effective solution to actively acquire measurement information and selectively perform estimation or control updates for enhanced system performance. In this chapter, we provide a concise overview of event-triggered state estimation and control developed for networked systems with limited communication or computation resources, as well as the recent developments on event-triggered learning that enable active discovery of system dynamics with enhanced data efficiency. Key technical challenges, major developments, and potential future directions are also discussed.</p
