Home
On Mon March 10, 2025

Speaker

Gyuyeong Kim

Title

Towards Network-Accelerated Computing Systems in the Era of Network Programmability

Abstract

Modern planet-scale online services require high-performance computing infrastructures, but the end of Dennard scaling and excessive coordination overhead make it challenging to augment computing resources efficiently. In this talk, I will introduce the concept of network programmability that provides new opportunities to transform the network into a computation-facilitating infrastructure. To show its potential impacts on the performance of computing systems, I will present examples of switch-based in-network acceleration, including in-network caching and in-network request cloning. Finally, I will briefly discuss the future directions of in-network acceleration, which include next-generation SmartNICs and eBPF/XDP.

Bio

Gyuyeong Kim is an Assistant Professor in the Department of Computer Engineering at Sungshin Women's University. He received his Ph.D. and B.S. in Computer Science from Korea University in 2020 and 2012, respectively. Before joining Sungshin Women's University, he was a Research Professor at Korea University. He works on broad topics in computer networking and systems. During undergraduate, he developed KLUE, a lecture evaluation service for Korea University.

Language

English

On Mon March 17, 2025

Speaker

Jun Han

Title

Sensing the Future: Unveiling the Benefits and Risks of Sensing in Cyber-Physical Security

Abstract

With the emergence of the Internet-of-Things (IoT) and Cyber-Physical Systems (CPS), we are witnessing a wealth of exciting applications that enable computational devices to interact with the physical world via an overwhelming number of sensors and actuators. However, such interactions pose new challenges to traditional approaches to security and privacy. In this talk, I will present how I utilize sensor data to provide security and privacy protections for IoT/CPS scenarios, and further introduce novel security threats arising from similar sensor data. Specifically, I will highlight some of our recent projects that leverage sensor data for attack and defense in various IoT settings. I will also introduce my future research directions such as identifying and defending against unforeseen security challenges from newer domains including smart homes, buildings, and vehicles.

Bio

Jun Han is an Associate Professor at KAIST of Computer Science, School of Computing at KAIST. He founded and directs the Cyber-Physical Systems and Security (CyPhy) Lab at KAIST. Prior to joining KAIST, he was at the National University of Singapore with an appointment in the Department of Computer Science, School of Computing, and Yonsei University with an appointment in the School of Electrical and Electronic Engineering. His research interest lies at the intersection of security and mobile/sensing systems and focuses on utilizing contextual information to solve security problems in the Internet-of-Things and Cyber-Physical Systems. He publishes at top-tier venues across various research communities spanning mobile computing, sensing systems, and security (including IEEE S&P, USENIX Security, ACM CCS, MobiSys, MobiCom, SenSys, Ubicomp, IPSN).

Language

English

On Mon March 24, 2025

Speaker

서영우

Title

My Perspective on Autonomy

Abstract

In the field of autonomous mobility, autonomy refers to a software stack designed to enable robots to navigate freely within human environments. Ensuring robust and reliable autonomy is essential for realizing the era of truly autonomous driving. In this talk, I will share my thoughts on autonomy: What it is, why it matters, the challenges it faces, and the milestones in its development.

Bio

Dr. Youngwoo Seo is a field-roboticist of building mobile robots including self-driving cars, drones, a high-speed transport – hyperloop, unmanned ground vehicles, etc. for more than two decades, and a seasoned executive with experience of managing diverse teams to deliver what matters. He currently serves as an Executive Vice President at the Land Systems Business Group, Hanwha Aerospace, where he oversees R&D efforts, among other responsibilities, for developing robotics and autonomous systems. Prior to joining Hanwha, Dr. Seo ran Atlas Robotics, Inc. to deliver a technology stack for autonomous mobility by shared autonomy, and led a team of engineers to develop perception stacks and mission-critical systems for Hyperloop One, to develop an autonomous flight stack for Autel Robotics, to deliver parts of the next-generation product at the Special Project Group of Apple, Inc., to deliver public demonstration of autonomous driving with GM-CMU Autonomous Driving Collaborative Research Lab. During his doctoral study, he was a member of the Tartan Racing team, the winning entry of the 2007 DARPA Urban Challenge, and worked on developing computational ways of augmenting cartographic resources and of assessing roadway status for reliable autonomous driving. While working as a research staff at the Robotics Institute of Carnegie Mellon University, he developed many machine learning algorithms and multi-agents systems to solve real-world problems. He earned a Ph.D. and a master’s degree in robotics from Carnegie Mellon University, and a master’s degree in computer science from Seoul National University.

Language

Korean

On Mon March 31, 2025

Speaker

한선화

Title

과학 커뮤니케이터, 데이터 커뮤니케이터

Abstract

공학자의 가장 큰 허들로 일반인과의 커뮤니케이션을 꼽습니다. 오랜 시간 강연과 방송으로 얻은 경험을 바탕으로 과학 커뮤니케이터가 갖추어야 할 기본 자질을 이야기 하고, 데이터를 기반으로 소통하는 데이터 클리닉과 데이터 커뮤니케이터를 소개하고자 합니다.

Bio

한선화 박사는 KAIST 전산학과에서 석사(1989) 및 박사(1997) 학위를 취득했다. 1997년부터 한국과학기술정보연구원(KISTI)에서 근무하며 원장(2014~2017)을 비롯해 지식정보센터장, 정보기술개발단장, 정책연구실장, 선임연구부장, 첨단정보연구소장 등 주요 직책을 역임했다. 2023년부터 ㈜페블러스 데이터커뮤니케이터에서 활동하고 있다. 또한, 국가과학기술연구회 정책본부장(2018~2020), 공공데이터전략위원회 민간위원(2018~2021), 국가과학기술자문회의 자문위원(2013~2014), 국가과학기술심의회 심의위원(2011~2015) 등 국가 과학기술 정책 수립 및 연구 발전에 기여해왔다. 과학 커뮤니케이터로서도 활발히 활동하며 *KTV 과학톡* (2018~2020) 진행을 맡았고, TJB *생방송투데이* (2020~), *곽마더* (2022), *미래설계소* (2023) 등에 출연했다.

Language

Korean

On Mon April 07, 2025

Speaker

이동기

Title

초거대 AI Infra 구축 전략 실행을 위한 Full-Stack 솔루션

Abstract

오늘날 초거대 AI 모델과 서비스의 발전은 데이터센터 및 AI 인프라의 혁신을 필연적으로 요구하고 있으며, 더 규모있고(Watts) 고성능(Flops)의 AI 인프라를 확보하려는 기업과 국가 간 경쟁이 더욱 치열해지고 있습니다. 이러한 환경에서 AI 인프라는 단순한 IT 인프라를 넘어 국가 경쟁력을 결정짓는 핵심 요소로 자리 잡고 있으며, 이를 위해 효율적인 구축 전략과 최적화된 솔루션이 필수적입니다. 본 세미나에서는 국가적 차원의 과업으로서 AI 인프라 구축 전략을 기업의 관점에서 설계하고, 이를 실현하기 위해 필요한 핵심 솔루션과 연구 과제를 논의하고자 합니다. 특히, AI DC의 운영 최적화, GPU/NPU 등 AI 가속기 활용 방안, 대규모 분산 학습을 위한 S/W stack 등 기술적 이슈를 리뷰하며 초거대 AI 인프라의 미래를 대비하기 위한 방향성을 제시하는 것이 목적입니다.

Bio


Language

Korean

On Mon April 21, 2025

Speaker

Gitta Kutyniok

Title

Reliable and Sustainable AI: From Mathematical Foundations to the Future of AI Computing

Abstract

Artificial intelligence is currently leading to one breakthrough after the other, in industry, public life, and the sciences. However, major drawbacks are the lack of reliability of such methodologies in particular for critical infrastructure as well as the enormous energy consumption of current AI computing. In this talk we will first provide an introduction into this vibrant research area. Taking a mathematical viewpoint will then lead us to a profound understanding of the problems of reliability and sustainability, and we will survey some recent advances. Finally, we will reveal an intriguing connection of both problems to analog AI computing and spiking neural networks, showing the necessity to rethink current digital computing platforms.

Bio

Gitta Kutyniok currently has a Bavarian AI Chair for Mathematical Foundations of Artificial Intelligence at the Ludwig-Maximilians-Universität München. Before, she held visiting positions at Princeton University, Stanford University, and Yale University, was a full professor at the Universität Osnabrück, and held an Einstein Chair at the Technische Universität Berlin. She received various awards for her research such as the von Kaven Prize (DFG) in 2007, was elected as a member of the Berlin-Brandenburg Academy of Sciences and Humanities in 2017 and of the European Academy of Sciences in 2022, and became a SIAM Fellow in 2019 and an IEEE Fellow in 2024. Her leadership positions include serving as LMU-Director of the Konrad Zuse School of Excellence in Reliable AI and as spokesperson of the AI-HUB
LMU. Her research work covers, in particular, applied harmonic analysis, artificial intelligence, compressed sensing, imaging sciences, and applications to life sciences, robotics, and telecommunication.

Language

English

On Mon April 28, 2025

Speaker

최진석

Title

TBA

Abstract

TBA

Bio

TBA

Language

Korean

On Mon May 12, 2025

Speaker

이훈상

Title

TBA

Abstract

TBA

Bio

TBA

Language

Korean

On Mon May 19, 2025

Speaker

정경미

Title

Digital Phynotyping과 Digital Therapeutics: 심리학적 접근

Abstract

최근 big data와 생성형AI의 급격한 발전은 심리서비스 영역에서 전통적인 방식의 심리평가/진단/치료에 대한 대안적 서비스전달체계로써 DP와 DTx의 가능성에 힘을 실어주고있다. 기술의 성공적인 적용은 기존 심리서비스 이론과 전달방식에 대한 이해 뿐 아니라 현행 기술적 접근방식의 문제점에 대한 이해에서 출발했을 때 가능할 것이다. 본 강의에서는 우울장애를 예로 평가와 진단에 대한 적용기술인 DP와 생성형AI를 이용한 인지행동치료 챗봇 개발의 문제점과 제안점을 제시할 것이다.

Bio

정경미 교수는 임상심리학 박사로 국내와 미국의 심리학자 자격증과 면허증을 가지고 있으며, 우울, 자폐성 장애, 그리고 신체장애로 인한 2차 정신장애 환자들에게 평가, 진단 및 행동치료, 행동분석 및 인지행동치료를 제공해 왔다. 정경미 교수는 Digital Mental Health 랩을 운영하면서, 2012년부터 다수의 국가 및 민간과제를 통해 심리서비스앱을 개발하고 그 효과성을 검증해 왔으며, 핸드폰을 통해 수집된 자료로 머신러닝과 복잡계이론에 근거한 통계방식을 이용해 우울을 예측하는 Digital phynotyping연구를 수행하고 있다.

Language

Korean

On Mon May 26, 2025

Speaker

Youngsok Kim

Title

Realizing the Benefits of Processing-in-DIMM for In-Memory Databases

Abstract

Modern Dual In-line Memory Modules (DIMMs) can now support Processing-In-Memory (PIM) by placing In-DIMM Processors (IDPs) near their memory banks. PIM can greatly accelerate in-memory joins, whose performance is frequently bounded by main-memory accesses, by offloading the operations of the join from host CPUs to the IDPs. However, as real PIM hardware has not been available until very recently, the prior PIM-assisted join algorithms have relied on PIM hardware simulators which assume many PIM hardware characteristics significantly different from those of real PIM hardware. To realize the benefits of PIM on real systems, I will first present PID-Join, a fast in-memory join algorithm exploiting UPMEM DIMMs, currently the only publicly-available PIM-enabled DIMMs. PID-Join optimizes all three join types (i.e., hash, sort-merge, nested-loop) for the IDPs, enables fast inter-IDP communication using host CPU cache streaming and vector instructions, and facilitates fast data transfer between the IDPs and the host memory. I will then present SPID-Join, a skew-resistant in-memory join algorithm leveraging the PIM-enabled DIMMs. SPID-Join overcomes PID-Join's performance and scalability limitations with skewed input records by replicating popular join keys across multiple IDPs. Doing so allows SPID-Join to process popular join keys with much higher aggregate memory bandwidth and computational throughput offered by multiple IDPs. As there exists a trade-off between the join throughput and the degree of join key replication, SPID-Join employs a cost model to identify the optimal join key replication ratio for given join and system configurations.

Bio

Youngsok Kim is currently an associate professor of the Department of Computer Science and Engineering at Yonsei University. His research interests span computer architecture and system software with an emphasis on architecture-conscious database management systems and processor performance modeling. He received his BSc and PhD in Computer Science and Engineering from POSTECH. He was a post-doc researcher at Seoul National University before joining Yonsei University. During his PhD studies, he was an intern at Consumer Hardware, Google Inc. and S.LSI Business, Samsung Electronics.

Language

English

On Mon June 02, 2025

Speaker

Gim Hee Lee

Title

Learning to Reconstruct and Comprehend Our 3D World

Abstract

The ability to perceive, reconstruct, and understand the 3D world is essential for a wide range of applications, including robotics, augmented reality, and autonomous driving. Recent advancements in deep learning and neural representations have revolutionized how we capture and interpret 3D environments, enabling high-fidelity reconstruction and semantic scene understanding even from sparse, incomplete, or ambiguous inputs. In this talk, I will present our recent work on neural 3D reconstruction and learning-based 3D scene understanding. Specifically, I will discuss our efforts in multi-view surface reconstruction, large-scale reconstruction for novel view synthesis, and reconstruction under occlusions and sparse-view settings. Additionally, I will highlight our research on open-world and vocabulary-free 3D scene understanding, pushing the boundaries of semantic comprehension in complex environments.

Bio

Dr. Gim Hee Lee is currently an Associate Professor in the Department of Computer Science at the National University of Singapore (NUS), where he leads the Computer Vision and Robotic Perception Laboratory. Prior to joining NUS, he was a researcher at Mitsubishi Electric Research Laboratories (MERL), USA. He obtained his PhD in Computer Science from ETH Zurich. He has served or will serve as an Area Chair for major computer vision and machine learning conferences such as CVPR, ICCV, ECCV, ICLR, NeurIPS, etc. He was part of the organizing committee as the Program Chair for 3DV 2022 and Demo Chair for CVPR 2023, and he is organizing 3DV 2025 in Singapore as the General Chair. He is a recipient the Singapore NRF Investigatorship, Class of 2024. His research interests include 3D computer vision, machine learning and robotics.

Language

English

On Mon June 09, 2025

Speaker

Hyunwoo J Kim

Title

Efficient Deep Video Understanding Towards AGI

Abstract

Video has become one of the most popular modalities that modern individuals consume and produce. However, developing AI systems that deeply understand videos is still a challenging goal due to the difficulty of annotations, the sheer volume of data, and the substantial computational burden required for training and inference of video models. To address these problems, I introduce new strategies for pre-training and fine-tuning video foundation models, including parameter-efficient fine-tuning (PEFT). Additionally, to deploy video models to users, I present training-free cost-efficient inference techniques for video transformers. To demonstrate the generalizability of video foundation models, I highlight our recent work in 'Video Question Answering' which implicitly requires tackling various subtasks and achieving a deeper understanding of videos. Lastly, I discuss how Video QA and Multimodal QA systems can serve as stepping stones towards artificial general intelligence, and outline future research directions.

Bio

Hyunwoo J. Kim is an associate professor at Korea University, where he leads Machine Learning and Vision Lab (MLV). His lab focuses on developing techniques for general-purpose AI systems, including multimodal foundation models, multi-modal question answering, efficient inference, and new neural network architectures. Prior to this position, he worked at Amazon Lab126 in Sunnyvale, California. He obtained a Ph.D. in Computer Sciences at the University of Wisconsin-Madison (Ph.D minor in statistics). He has served (or is serving) as an Area Chair for ICLR 2025, ICCV 2025, CVPR 2025, 2024 and co-organized the 1st and 2nd MICCAI workshops on Foundation Models for General Medical AI in 2023 and 2024.

Language

English