Seminar

Zoom Link: https://hku.zoom.us/j/92852691758?pwd=ApSXYeoYJk3Y5MLcS33duwNe4ZyTnM.1 Abstract The rapid growth of Internet-of-Things (IoT) deployments, combined with increasing adoption of edge computing, has created substantial challenges in energy efficiency, real-time intelligence, and collaborative resource management. IoT systems—ranging from smart buildings to cellular base stations—exhibit highly dynamic, heterogeneous, and energy-intensive behaviors that require new data-driven methods for prediction and optimization. […]

Zoom Link:  https://hku.zoom.us/j/91408893076?pwd=5rgw1jrHuqg5lKfbIba5O5OsJGZbdf.1 Abstract Marine vehicles play an essential role in modern ocean operations, where reliable motion control is critical for safe, precise, and efficient task execution. This talk presents a direct data-driven control framework that synthesizes controllers directly from input–state–output data, thereby bypassing the need for complex hydrodynamic modeling and system identification. We address […]

Zoom Link: https://hku.zoom.us/j/8157366378?omn=98994435560 Abstract The poor mechanical durability of perovskite films due to the severe intrinsic strain, and the brittle nature of the flexible ITO electrode hinder foldable perovskite solar cells (F-PSCs) realization. In this talk, the strategy of region-dependent microscopic and macroscopic strain suppression is demonstrated to achieve efficient F-PSCs on silver nanowires (AgNWs) […]

Zoom Link:  https://hku.zoom.us/j/97347520963?pwd=ohdjCe9kx6axOTFn2m2M9gsVojb2kG.1 Abstract Federated Learning (FL) has rapidly advanced as a foundational paradigm for realizing privacy-preserving intelligence in edge networks. However, its real-world deployment is fundamentally challenged by two dimensions: spatial scalability across a large, heterogeneous population of devices, and temporal robustness over long-lived, evolving learning processes. While blockchain technology offers inherent benefits like […]

Zoom Link:  https://hku.zoom.us/j/97338207102 Abstract Electrified high-speed railways are emerging as major and spatially distributed electricity consumers in modern power systems, and their traction demand is tightly coupled with train dynamics and timetable scheduling. With the increasing exploitation of renewable resource endowments along railway corridors, electrified railways are evolving from pure loads into potential flexibility and […]

Zoom Link:  https://hku.zoom.us/j/96462551842?pwd=MFNEYU1qcmZzeE9Rby9aRVZLQ0RZdz09 Abstract To address the computational challenges of AC Unit Commitment (AC-UC), this paper proposes a hybrid iterative framework that decomposes the MINLP model into a linearized MILP master problem and an exact AC feasibility check. The approach integrates Taylor-expansion-based linearization with a novel switching strategy that coordinates global updates for initial geometric […]

Zoom Link: https://hku.zoom.us/j/97793742616?pwd=YIyYlokhzOsap3IvbsbwmfaHVHFoin.1 Abstract As Generative AI shifts toward handling massive context windows, the quadratic complexity of Transformer architecture has become a significant bottleneck. State Space Models (SSMs), particularly Mamba, have emerged as a promising solution, offering linear-time scaling and superior efficiency. However, the unique computational duality of SSMs—requiring both memory-intensive projections and agile, input-dependent […]

Zoom Link: https://hku.zoom.us/j/98499142544?pwd=zvVs3BqWzIzCA071Dqq2rYW7vIAqj7.1 Abstract The advent of GPT-4o highlights the immense potential of Multimodal Large Language Models (MLLMs) with native visual generation capabilities. These unified models offer precise control in multimodal interactions, enabling exceptional fluency in tasks such as multi-turn image editing and visual in-context learning. However, a fundamental dilemma remains in the choice of […]

Zoom Link: https://hku.zoom.us/j/91953486299?pwd=lEW6N6JKhmAHLoS38VohufbEoJyJtw.1 Abstract This seminar examines how modern image generation models can be redesigned to achieve both high fidelity and high efficiency under flexible inference budgets. I will begin with a brief overview of generative modeling, focusing on diffusion models and related flow-based formulations. While these approaches deliver strong stability and visual quality, they […]

Abstract This seminar provides a systematic showcase of our end-to-end energy-efficient solutions in AI accelerator design, spanning from the foundational compute unit to specialized architectures and large model deployment on edge devices. We begin by introducing a novel Reconfigurable Processing Element (PE), the core compute unit designed to support multiple floating-point (e.g., BF16, FP16) and […]

Zoom Link: https://hku.zoom.us/j/6983793721?pwd=UjwaYARlhjvhUf7DDp3bMjbJnlhVbr.1 Abstract Structured illumination refers to a class of optical imaging methods in which the sample is illuminated with a controlled spatial light pattern rather than uniform illumination. By altering the illumination pattern, it can facilitate the extraction of spatial or temporal information from the sample, and may improve imaging performance such as […]