Young Scholar

Young Scholar TechTalk – Secure and High-performance AI Serving: Protecting AI Secretes, Accelerating AI Insights

September 19, 2023 (Tuesday) 4:30-5:30pm
Driven by the remarkable success of artificial intelligence (AI) and edge computing, the deployment of well-trained private AI models on third-party edge devices for mission-critical applications has become increasingly prevalent. Safeguarding these private models on untrusted devices, while simultaneously speeding up model serving (i.e., inference) through accelerators like GPUs, has escalated in urgency.
We introduce SOTER, a new AI serving system that, for the first time, achieves both high security and high performance. Harnessing the associativity property of AI operators, SOTER presents an innovative approach—transforming computationally expensive AI operators into parameter-morphed equivalents for secure execution on untrusted but fast GPUs, and losslessly restoring inference results within trusted execution environments (TEEs) in CPUs. Experimental results on six prevalent AI models in the three most popular categories show that, even with stronger model protection, SOTER achieves comparable performance with baselines while retaining the same high accuracy as insecure AI model inference.

Young Scholar TechTalk – Chip-scale Sensing: From Classical to Quantum Regime

Miniature optoelectronic sensors which have features of convenient, reliable, economic, ultra sensitive, and capable of real-time measurement are highly desirable nowadays. However, currently reported optical and electronic sensing devices are still hindered with complex optical components and bulky equipment. Hence, we hope to further minimize the volume of the sensing system and get rid of the dependence on complex, expensive and bulky sensing components. In particular, we demonstrate a micro-scale III-nitride chip that integrates a light emitter (LED) and a photodetector (PD) together, realizing the emission and detection of signals in a single miniature chip. Thus, we have applied the device into some classical sensing, such as pressure, salinity content and cell activities sensing. Additionally, we also conduct integration on the diamond based quantum sensing system, and demonstrate a compact chip architecture (sub ~mm3 volume) being capable of on-chip quantum sensing.

Young Scholar TechTalk – HOF2 – Interact with Device through Simple and Robust Hand-Over-Face Gesture

Mobile devices have been like an extended part of ourselves, but can we really operate a mobile device just as naturally as how we control our fingers or body? We present HOF2, a novel input modality that uses simple gestures over your face to interact with your device. Unlike other gesture-based modalities, HOF2 is highly robust and can avoid false triggering caused by many unconscious gestures like scratching or wiping, while is still easy, comfortable and natural to use. Moreover, HOF2 is highly available and can be implemented on any mobile phone/tablet/computer with a single camera and without remote servers. In this TechTalk, we will present a live demo on iOS/iPadOS demonstrating the performance of HOF2 scheme in practice and explore some real-life use cases such as virtual conferencing, selfie, or TV controller. We believe there are far more possibilities waiting to be explored with this novel interaction scheme.

Young Scholar TechTalk – High-throughput Cell Mechanics Characterization with Microfluidics

Cells can sense mechanical stimuli and convert them to biochemical signals for various specific cellular responses, such as stem cells differentiation, initiation of transcriptional programs, and cell migration. Cell mechanics focuses on the mechanical properties and behaviours of living cells and how cell mechanics relates to various cell functions. Currently, traditional cell mechanics measurement methods are cumbersome, low-throughput, and expensive to deploy. By exploiting microfluidic technology, Dr. Johnson Cui is investigating the cancer cell mechanics and developing an accurate, easy-to-use cell mechanics measurement platform for cell mechanics research and also for cancer diagnosis and therapeutics in the future.

Young Scholar TechTalk – Learning to Control and Coordinate Hybrid Traffic Through Robot Vehicles at Complex and Unsignalized Intersections

Intersections are essential road infrastructures for traffic in modern metropolises; however, they can also be the bottleneck of traffic flows due to traffic incidents or the absence of traffic coordination mechanisms such as traffic lights. Thus, various control and coordination mechanisms that are beyond traditional control methods have been proposed to improve the efficiency of intersection traffic. Amongst these methods, the control of foreseeable hybrid traffic that consists of human-driven vehicles (HVs) and robot vehicles (RVs) has recently emerged. We propose a decentralized reinforcement learning approach for the control and coordination of hybrid traffic at real-world, complex intersections–a topic that has not been previously explored. Comprehensive experiments are conducted to show the effectiveness of our approach. We show that using 5% RVs, we can prevent congestion formation inside the intersection under the actual traffic demand of 700 vehicles per hour. When there exist more than 50% RVs in traffic, our method starts to outperform traffic signals on the average waiting time of all vehicles at the intersection.

Young Scholar TechTalk – Blowing Bubbles in Membranes for More Efficient Freshwater Production

Global scarcity and contamination of freshwater pose a significant threat to sustainable development. To address this crisis, reverse osmosis (RO) technology has been playing a pivotal role in desalination and water reuse for freshwater production. The effectiveness of the RO membrane filtration is highly dependent on its surface functional rejection layer. My research focuses on shaping this rejection layer to be a voids-bearing structure, resembling blowing bubbles within the layer. This technique will result in a thinner rejection layer with a larger surface area, favoring water transport. On this basis, shaping branch bubbles to resemble a tree or coral can potentially achieve an exponential increase in water filtration efficiency, resulting in faster production of freshwater with significantly lower energy consumption.

Young Scholar TechTalk – Understanding Rainfall-induced Slope Failures from an Integrated Perspective

Climate change increases the frequency and intensity of extreme rainfall events and magnifies the threat of rainfall-induced slope failure. The consequences of these failures can be dramatic and devastating if flow slides are triggered. While considerable efforts have been made in the past decades to understand the failure mechanisms and develop techniques to mitigate the hazards, the complexity of interplays of various factors causes it to remain an area of uncertainty and difficulty in geotechnical engineering. This talk will briefly review and discuss the main factors affecting rainfall-induced slope failures from a perspective integrating the geotechnical, hydrological, and climatological aspects. The two deadly landslides in Sau Mau Ping, Hong Kong, in June 1972 and August 1976, which caused 165 casualties, are revisited. We raise an intriguing question that has long been overlooked: why were the slopes able to withstand the 1972 rainfall but failed in the 1976 rainfall event, given that the rainfall intensity of the latter event was only half of the former. We explore the roles of geological and hydrological settings and the rainfall characteristics to look into the causes and mechanisms of these failures. Implications of the new findings for practice will also be discussed.

Young Scholar TechTalk – Flexible Learning of Quantum States with Generative Query Neural Networks

Deep neural networks are a powerful tool for the characterization of quantum states. Existing networks are typically trained with experimental data gathered from the specific quantum state that needs to be characterized. In this talk, Mr. Yan Zhu, from Department of Computer Science, will introduce a model of network that can be trained with classically simulated data from a fiducial set of states and measurements, and can later be used to characterize quantum states that share structural similarities with the states in the fiducial set. With little guidance of quantum physics, the network builds its own data-driven representation of quantum states, and then uses it to predict the outcome statistics of quantum measurements that have not been performed yet. The state representation produced by the network can also be used for tasks beyond the prediction of outcome statistics, including clustering of quantum states and identification of different phases of matter.

Young Scholar TechTalk – Cyber-Physical Spare Parts Intralogistics System for Aviation MRO

Spare parts management is a vital supporting function in aviation Maintenance, Repair,and Overhaul (MRO). Spare parts intralogistics (SPI), the operational perspective of spare parts management, significantly affects performance of MRO activities. This study proposes Cyber-Physical Spare Parts Intralogistics System (CPSPIS) to address the synchronization problems associated with the SPI business process and SPI resources. The proposed system applies Internet-of-Things technologies and unified representations to provide resources and operations traceability and visibility. Further, CPSPIS contributes several services with self-X abilities for real-time synchronization throughout the SPI process. In addition, CPSPIS develops applications and visualization tools for real-time cooperation between execution and decision-making. Finally, this study conducts a real-life case study in one of the largest aviation MRO organizations in Hong Kong, and discusses the quantitative and qualitative improvements of CPSPIS.

Young Scholar TechTalk – Subgraph Federated Learning with Missing Neighbor Generation

In computer science, a graph is a network modeling objects and their unique interactions. The graph learning model is a specialized machine learning model that learns on graphs. Similar to traditional machine learning models, a well-performed graph learning model can capture the global data distribution with sufficient and unbiased training data. However, in a distributed subgraph system, most data owners only possess small amounts of the data (small subgraphs) in their local systems and can have unpredictable biases.
In this talk, the speaker will introduce this novel yet realistic setting – subgraph federated learning, which aims to let distributed data owners collaboratively train a powerful and generalized graph learning model without directly sharing their subgraphs. Towards this setting, two major techniques are proposed by the research team. (1) FedSage, which trains a GraphSage model based on FedAvg to integrate node features, link structures, and task labels on multiple local subgraphs; (2) FedSage+, which trains a missing neighbor generator along FedSage to deal with missing links across local subgraphs. Empirical results and theoretical analysis of proposed models respectively demonstrate the effectiveness and prove the generalization ability.