香港六和大全

讲座信息

08.11 | What are language agents and why should we study them?

2023-08-05

A recent focus in AI is autonomous agents, which, usually powered by large language models (LLMs), can follow language instructions to autonomously carry out complex tasks in real-world environments. However, the concept of agent has been introduced into AI since its dawn, so what's different this time around? I argue that the most fundamental change is the capability of using language. Contemporary AI agents use language as a vehicle for both communication and thought, which enables them to do

全文》
08.10 | 数字健康技术新进展 New Advances in Digital Health for Improving Healthcare

2023-08-03

近年来，现代化"数字健康"在数据科学领域取得了长足进步，为改善患者护理提供了许多机会。本讲座将全景式介绍现代医疗健康技术的发展现状与前景，并列举牛津大学在开发改善患者生活的医疗健康干预措施方面的研究实例。

全文》
08.01 | Open-World Streaming Algorithms: Paradigm, Challenge, and Prospect 开放世界流处理算法：范式、挑战与展望

2023-07-26

Machine learning (ML) strives to build mathematical models that uncover and mimic the hidden decision-making process of human minds in a data-driven fashion. Despite its popularity and successes in many industrial and scientific endeavors, a fundamental difference lying between the learning environments of ML models and human-beings hinders the widespread application of ML in real tasks. To wit, we humans live in an evolving world so must keep adjusting and reshaping our mental process to accomm

全文》
07.18 | Object-Centric Perception for Robot Manipulation

2023-07-11

One of the core problems in robot manipulation is enabling robots to manipulate objects and use objects to perform tasks. Compared to end-to-end approaches that take sensory input and output control commands, I believe enabling robots to understand objects from sensory input and conditioning planning and control on top of object perception can achieve generalizable robot manipulation. In this talk, I will highlight our efforts on object-centric perception for robot manipulation. I will start wit

全文》
07.14 | Towards Regulated Secure and Privacy-Preserving Computing Platforms

2023-07-03

Computing is undergoing a significant shift. First, the explosive growth of the Internet of Things (IoT) enables users to interact with computing systems and physical environments in novel ways through perceptual interfaces (e.g., microphones and cameras). Second, machine learning algorithms collect huge amounts of data and make critical decisions on new computing systems. While these trends bring unprecedented functionality, they also drastically increase the number of untrusted algorithms, imp

全文》
07.13 | 打造互联网、无线网、和区块链仿真器

2023-06-30

为了给网络和网络安全提供一个学习、科研、测试和示范平台，香港六和大全开发了一个开源的互联网仿真器。用它可以创建一个能在个人或多个计算机上运行的微型互联网。它麻雀虽小但五脏俱全，具有互联网的所有基本要素。在该仿真器上香港六和大全可以部署很多系统和应用，例如 DNS系统（包括根和顶级域名服务器），各种路由协议，暗网，僵尸网络，蠕虫，等等。也可以重现很多互联网上有名的攻击。香港六和大全还在仿真器上部署了以太坊系统（Ethereum），实现了区块链仿真器。目前正在实现对无线 ad hoc 网络的仿真。该仿真器自2021年发布以来已被多个香港六和大全在科研和教育上使用。我将介绍该仿真器的设计理念，并演示它的一些功能以及应用。这是一个开源软件，希望感兴趣的老师同学们加入香港六和大全，在上面打造更多有用的网络系统。

全文》
07.05 | Emerging wearable technology and its applications

2023-06-29

From smart watches and fitness trackers to smart textiles, from Google Glasses to Apple Vision Pro, wearable technology is developing rapidly and is gaining popularity in the fields of healthcare, fitness and wellbeing, manufacturing, entertainment, etc. In this talk, we will present several wearable devices and applications developed by Aalto University Wearable Systems Lab, including wearable cognitive assistance for assembly and maintenance assistance, virtual stroke rehabilitation, and weara

全文》
06.29 | Turning information into knowledge: challenges & solutions

2023-06-20

Extracting structured knowledge from unstructured information such as text, images and speech is a long-standing problem and is useful in a variety of domains and applications. This problem comes with many challenges: label scarcity, multimodality and data biases, to name a few. In this talk, I will discuss some of these challenges and present our recent works that address them for the tasks of event and relation extraction.

全文》
06.16 | Connected Vehicles and Edge Intelligence for Cooperative Autonomous Driving

2023-06-13

Connected autonomous vehicle (CAV) holds great potentials for future transport, including significantly improved road safety and efficiency. 3GPP 5G has specified support for advanced CAV driving use cases. However, the harsh vehicle communication environment and stringent CAV application requirements present many technical challenges. Cooperative CAV will be critical to tackle these challenges and fully realise the potentials of CAV, by which computing and sensing resources and driving informat

全文》
06.13 | Implicit Functions for 3D Reconstruction and Neural Rendering

2023-06-12

Implicit functions have served as a widely used representation for both 2D and 3D objects in computer graphics, playing a significant role in the early development of the field. In recent years, with the advent of 3D deep learning and the rapid growth of neural rendering techniques, the implicit representation of 3D shapes has once again gained substantial attention. In this talk, I will highlight several recent research projects conducted by myself and my collaborators in the areas of 3D recons

全文》
06.13 | Efficient Zero-Knowledge Proofs: Theory and Practice.

2023-06-06

In this talk, we discuss a cryptographic tool named zero-knowledge proof from both theory and application perspectives. In theory, we present Libra, the first zero-knowledge protocol with optimal prover time, fast verifier time, and succinct proof size. Libra also has excellent concrete efficiency in practice. In application, we present the first solution for building trustless and permissionless cross-chain bridges in blockchains using zero-knowledge proof. In addition, we discuss how to apply

全文》
06.09 | 上海技术论坛人形机器人系列机器人大模型的机遇与挑战：从AIGC到AIGA?

2023-06-05

全文》
06.09 | Prototype Pollution and Beyond: An Existential, Emerging Threat to the World Wide Web

2023-05-30

Prototype pollution is a relatively-new type of vulnerability specific to prototype-based languages, such as JavaScript, which allows an adversary to pollute a base object’s property, leading to further consequences such as Cross-site Scripting (XSS) and session fixation. In this talk, I am presenting our research works in the past five years, which detect and exploit not only prototype pollution vulnerabilities but also other related JavaScript vulnerabilities across server- and client-side app

全文》
06.06 | 【学院讲座】The Hidden Dangers of Generative Models: Uncovering the Risks and Mitigating the Harm

2023-05-30

The Hidden Dangers of Generative Models: Uncovering the Risks and Mitigating the HarmAbstract: Generative models, such as GANs, VAEs, and language models, have revolutionized the field of AI and transformed our ability to generate high-quality images, videos, and text. However, these powerful tools also come with hidden dangers that can have serious consequences for individuals and society as a whole. In this keynote talk, we will explore the risks associated with generative models and the poten

全文》
06.03 | 深度图学习：数据、方法和应用

2023-05-30

图数据（Graph-structured Data）广泛存在于现实世界中如社交网络数据、交通网络数据等。图数据的非欧特性使得对其有效表征尤为重要。然而，当前对图数据的表示方法受限于传统图数据的二元关系（节点-关系-节点）表达形式，对多元关系及隐式关系的捕捉能力不足。深度图学习方法能够有效增强数据质量、挖掘隐式关系，优化图数据表示，增强图学习方法表达能力，对如群体异常检测、疫情防控等现实应用有着十分重要的研究意义。报告围绕深度图学习的基本概念及分类展开，针对数据、方法和应用三个层面分别探讨深度图学习的相关工作。具体地，在社交网络分析方面，融合社交属性等对学术合作团队建模、识别、优化等全生命周期给出了一系列解决方案，研究成果对组建高水平学术团队、助力团队良性发展及跨学科团队建设给予量化分析与支撑；在异常检测方面，针对金融、医疗、国家安全等重要领域中不断升级的异常行为及异常模式展开研究，融合多源数据在约束查询条件的情况下识别群体异常，为群体异常检测提出了开创性方法，研究成果可应用于优化问答系统搜索结果，快速识别社交网络中虚假评论，有效阻断假新闻传播；在疫情防控方面，融合多因素提出城市风险评估

全文》
06.03 | Travel Intention Enhanced Out-of-Town Mobility Prediction for Social Network Users

2023-05-30

It is becoming a hot research topic to predict where a user would visit in a neighboring city based on check-in data collected from location-based mobile Apps. However, as most users rarely travel out of hometown, there is a high skew of the quantity of check-in data between hometown and neighboring cities. Suffering from the severe sparsity of user mobility data, existing studies do not perform well as they can hardly maintain users' intrinsic preference and meanwhile adapt to their interest dr

全文》