Protected: More is Less: Dynamic Sparse Processing in the Era of Sustainable AI – Copy
There is no excerpt because this is a protected post.
TechTalk
TechTalk – Building Multi-dimensional Parallel Training Systems for Large AI Models
June 24, 2025 (Tuesday) 4:30-5:30pm
Achieving fairness in resource allocation can be modeled as a graph-based optimization problem, with many efficient algorithms available. This talk explores the connection between market equilibrium and graph density decomposition, showing how fast convergence can be achieved in large-scale systems. We present a unified framework linking hypergraph density decomposition and Fisher market equilibrium through locally verifiable optimality conditions. This symmetry allows repurposing algorithms between domains, significantly accelerating convergence.
We focus on iterative gradient-based methods, including the iterative proportional response process and its momentum-enhanced extensions. Our novel exponential momentum approach refines traditional techniques, delivering near-optimal solutions in distributed settings. Empirical results show these methods outperform existing algorithms, achieving speedups by several orders of magnitude in large-scale graphs.
By integrating graph theory, market dynamics, and optimization, this talk offers new insights into efficient computation to achieve fairness in networked systems. These methods deepen our understanding of algorithmic principles and open new applications in algorithm design, social networks, and economic modeling.
More is Less: Dynamic Sparse Processing in the Era of Sustainable AI
June 19, 2025 (Thursday) 4:30-5:30pm
With modern AI seemingly transforming all aspects of our modern society like magic, it is easy to forget the impacts such modern technology marvel is causing to our fragile environment. From massive industry-scale training of large neural networks of epic sizes, to the proliferation of folding AI inference in our everyday activities, AI applications are rapidly increasing the stress to our global energy and water infrastructure. The coming era of AI demands not only the smartest AI models, but also a new generation of sustainable AI mindset that rethinks the when and how to apply AI intelligence.
In this talk, Professor So will discuss one angle that addresses the “how” question with intelligent algorithm-architecture co-designed systems that reduce both energy consumption and computing latency using dynamic sparse processing. Sharing the same “more is less” principle, a series of works from a dynamic sparse processing system for event cameras to token-steering in modern diffusion models will be discussed. Together, these works illustrate that by doing a little bit more upfront intelligently, it is possible to drastically reduce the amount of work necessary to perform the same AI inference operation during run time without affecting the accuracy of a model. The results are solutions that not only are fast, but they are also orders of magnitude more energy-efficient than typical GPU-accelerated systems.
HKAE TechTalk – Transforming Wastewater into Green Hydrogen: An Option to Produce the Fuel of the Future in HK?
June 17, 2025 (Tuesday) 4:00-5:00pm
Environmental Analytical Microbiology integrates advanced sequencing, absolute quantification, and standardization to address microbial pollution and antibiotic resistance. This talk traces microbial ecology’s evolution from microscopy to metagenomics, highlighting innovations like cellular internal standards for cross-study comparisons. Case studies demonstrate applications in wastewater treatment, anaerobic digestion, and microbial risk assessment, emphasizing quantification of pathogens and antibiotic resistance genes (ARGs). The global resistome analysis reveals anthropogenic impacts on ARG distribution, while frameworks prioritize high-risk genes. Future efforts focus on the Environment–Microbiome–Health axis, advocating for microbiome-informed urban design to enhance public health. By merging cutting-edge technologies with interdisciplinary collaboration, this field aims to mitigate environmental risks and promote sustainable co-evolution with microbial ecosystems.
Young Scholar TechTalk – From Grains to Landslides: An Efficient Particle Method for Simulating Large Deformation of Geomaterials
May 13 2024 (Tuesday) 4:30-5:30pm
Mr Shuaihao Zhang is currently a fourth-year Ph.D. candidate in the Department of Civil Engineering at the University of Hong Kong, supervised by Professor Sérgio D.N. Lourenço. He previously obtained his B.Eng. and M.Eng. degrees from Central South University. During his Ph.D. studies, he spent one year as a visiting researcher at the School of Engineering and Design, Technical University of Munich, under the supervision of PD Dr. Xiangyu Hu. His research focuses on improving the stability and accuracy of Smoothed Particle Hydrodynamics (SPH) algorithms and applying them to simulate large deformation problems in geotechnical engineering.
HKAE TechTalk – Environmental Analytical Microbiology and its Applications
May 20, 2024 (Tuesday) 4:00-5:00pm
Environmental Analytical Microbiology integrates advanced sequencing, absolute quantification, and standardization to address microbial pollution and antibiotic resistance. This talk traces microbial ecology’s evolution from microscopy to metagenomics, highlighting innovations like cellular internal standards for cross-study comparisons. Case studies demonstrate applications in wastewater treatment, anaerobic digestion, and microbial risk assessment, emphasizing quantification of pathogens and antibiotic resistance genes (ARGs). The global resistome analysis reveals anthropogenic impacts on ARG distribution, while frameworks prioritize high-risk genes. Future efforts focus on the Environment–Microbiome–Health axis, advocating for microbiome-informed urban design to enhance public health. By merging cutting-edge technologies with interdisciplinary collaboration, this field aims to mitigate environmental risks and promote sustainable co-evolution with microbial ecosystems.
HKAE TechTalk – Interdisciplinary Energy Transition for Carbon Neutrality
May 16, 2024 (Friday) 4:00-5:00pm
Transitioning to renewable energy and efficient energy systems is paramount for achieving carbon neutrality, necessitating a multidisciplinary approach. Energy research must encompass a wide array of factors, including temporal aspects, geographical considerations, and data analysis. This discussion delves into the complexities of energy transitions, illustrating their interconnectedness with the global energy landscape. Challenges such as integrating renewable energy sources, managing demand fluctuations, and overcoming research barriers are explored. By forging connections between energy and various sectors, comprehensive and sustainable solutions can be developed. A focus on interdisciplinary research is crucial for advancing future energy endeavors and addressing broader environmental imperatives.
TechTalk – Multiscale Robotics and Automation for Advanced Healthcare Applications
April 10, 2025 (Thursday) 4:30-5:30pm
Advancements in biology and medicine necessitate enabling technologies for the manipulation and characterization of tissues, single cells, and even sub-cellular structures. Over the past decades, we have witnessed significant progress in the area of micro and medical robotics for biomedical applications such as single-cell manipulation and minimally invasive surgery. This talk will introduce a new search and detection method for automatically locating end-effector tips; vision-based contact detection algorithms, and microrobotic manipulation techniques. Furthermore, this talk will present microrobotic systems integrating these techniques for microinjection, along with the introduction of an inner actuated microrobot specially designed for minimally invasive medical procedures. The new technology has the potential to reshape cardiovascular drug testing and revolutionize the automated cryopreservation of reproductive cells in IVF clinics. Future research directions including augmented reality-assisted surgery and personalized healthcare will also be discussed.
TechTalk – Speeding Up Fairness: The Science of Fast Convergence in Markets and Graphs
April 3, 2025 (Thursday) 4:30-5:30pm
Achieving fairness in resource allocation can be modeled as a graph-based optimization problem, with many efficient algorithms available. This talk explores the connection between market equilibrium and graph density decomposition, showing how fast convergence can be achieved in large-scale systems. We present a unified framework linking hypergraph density decomposition and Fisher market equilibrium through locally verifiable optimality conditions. This symmetry allows repurposing algorithms between domains, significantly accelerating convergence.
We focus on iterative gradient-based methods, including the iterative proportional response process and its momentum-enhanced extensions. Our novel exponential momentum approach refines traditional techniques, delivering near-optimal solutions in distributed settings. Empirical results show these methods outperform existing algorithms, achieving speedups by several orders of magnitude in large-scale graphs.
By integrating graph theory, market dynamics, and optimization, this talk offers new insights into efficient computation to achieve fairness in networked systems. These methods deepen our understanding of algorithmic principles and open new applications in algorithm design, social networks, and economic modeling.
HKAE TechTalk – Nanotechnology for Three-Dimensional Biosensors, Terahertz Devices, and Meta-Devices
March 18, 2024 (Tuesday) 4:00-5:00pm
Recent advancements in nanotechnology have enabled the creation of diverse nanostructures, which are integral to high-performance biosensors, terahertz (THz) devices, and meta-devices. Nanoimprint technology, when combined with precise dry etching techniques, facilitates the rapid production of three-dimensional (3D) nanodevices with high uniformity and precisely controlled dimensions across large areas.
In this presentation, we will showcase several applications of these technologies:
1. 3D Biomimetic Platforms and Plasmonic Biosensors to control and monitor cells and biomolecules with exceptional sensitivity.
2. High-Frequency Terahertz (THz) Lenses and Antennas utilizing curved or meta surfaces to enhance functionality and performance.
3. Multi-Layered Meta Surfaces with Twist Angles to achieve chiral magic angles for advanced light manipulation.