BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Biomedical Mathematics Group - ECPv6.17.1//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-ORIGINAL-URL:https://www.ibs.re.kr/bimag
X-WR-CALDESC:Events for Biomedical Mathematics Group
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:Asia/Seoul
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:KST
DTSTART:20250101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Asia/Seoul:20260807T100000
DTEND;TZID=Asia/Seoul:20260807T120000
DTSTAMP:20260730T084420Z
CREATED:20260730T084420Z
LAST-MODIFIED:20260730T084420Z
UID:12778-1786096800-1786104000@www.ibs.re.kr
SUMMARY:AutoML-Agent: A Multi-Agent LLM Framework for Full-Pipeline AutoML - Jin Woo Hyun
DESCRIPTION:In this talk\, we discuss the paper “AutoML-Agent: A Multi-Agent LLM Framework for Full-Pipeline AutoML” by Patara Trirat et al\, ICML\, 2025. \nAbstract: \nAutomated machine learning (AutoML) accelerates AI development by automating tasks in the development pipeline\, such as optimal model search and hyperparameter tuning. Existing AutoML systems often require technical expertise to set up complex tools\, which is in general time-consuming and requires a large amount of human effort. Therefore\, recent works have started exploiting large language models (LLM) to lessen such burden and increase the usability of AutoML frameworks via a natural language interface\, allowing non-expert users to build their data-driven solutions. These methods\, however\, are usually designed only for a particular process in the AI development pipeline and do not efficiently use the inherent capacity of the LLMs. This paper proposes AutoML-Agent\, a novel multi-agent framework tailored for full-pipeline AutoML\, i.e.\, from data retrieval to model deployment. AutoML-Agent takes user’s task descriptions\, facilitates collaboration between specialized LLM agents\, and delivers deployment-ready models. Unlike existing work\, instead of devising a single plan\, we introduce a retrieval-augmented planning strategy to enhance exploration to search for more optimal plans. We also decompose each plan into sub-tasks (e.g.\, data preprocessing and neural network design) each of which is solved by a specialized agent we build via prompting executing in parallel\, making the search process more efficient. Moreover\, we propose a multi-stage verification to verify executed results and guide the code generation LLM in implementing successful solutions. Extensive experiments on seven downstream tasks using fourteen datasets show that AutoML-Agent achieves a higher success rate in automating the full AutoML process\, yielding systems with good performance throughout the diverse domains.
URL:https://www.ibs.re.kr/bimag/event/automl-agent-a-multi-agent-llm-framework-for-full-pipeline-automl-jin-woo-hyun/
LOCATION:B232 Seminar Room\, IBS\, 55 Expo-ro Yuseong-gu\, Daejeon\, Daejeon\, 34126\, Korea\, Republic of
CATEGORIES:Journal Club
ORGANIZER;CN="Jae Kyoung Kim":MAILTO:jaekkim@kaist.ac.kr
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Seoul:20260814T100000
DTEND;TZID=Asia/Seoul:20260814T120000
DTSTAMP:20260730T035435Z
CREATED:20260730T035435Z
LAST-MODIFIED:20260730T035435Z
UID:12774-1786701600-1786708800@www.ibs.re.kr
SUMMARY:Evolution of error correction through a need for speed - Kangmin Lee
DESCRIPTION:In this talk\, we discuss the paper “Evolution of error correction through a need for speed” by Riccardo Ravasio et al.\, Science\, 2026. \nAbstract: \nKinetic proofreading is a class of error-correcting mechanisms in biology that expend energy to avoid mistakes during replication\, transcription\, and translation. Proofreading is typically assumed to evolve when selection for fidelity outweighs costs in energy and the speed of replication. We show that when stalling after misincorporations is accounted for\, proofreading can instead speed up replication. Consistent with data on polymerase mutagenesis\, our results suggest that proofreading can evolve under selection for speed alone. We generalize to multicomponent self-assembly and show that analogous error-correcting processes\, such as dynamic instability\, can likewise emerge purely from selection for rapid assembly. Thus\, nonequilibrium error correction can evolve from selection for speed\, even without direct fidelity advantages. We discuss implications for mutation-rate evolution\, molecular assembly processes\, and models of early life.
URL:https://www.ibs.re.kr/bimag/event/evolution-of-error-correction-through-a-need-for-speed-kangmin-lee/
LOCATION:108\, Conference Room\, IBS\, 55 Expo-ro Yuseong-gu\, Daejeon\, Korea\, Republic of
CATEGORIES:Journal Club
ORGANIZER;CN="Jae Kyoung Kim":MAILTO:jaekkim@kaist.ac.kr
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Seoul:20260821T100000
DTEND;TZID=Asia/Seoul:20260821T120000
DTSTAMP:20260730T035554Z
CREATED:20260730T020529Z
LAST-MODIFIED:20260730T035554Z
UID:12772-1787306400-1787313600@www.ibs.re.kr
SUMMARY:Tangent-Space Regularization for Neural-Network Models of Dynamical Systems - Olive Cawiding
DESCRIPTION:In this talk\, we discuss the paper “Tangent-Space Regularization for Neural-Network Models of Dynamical Systems” by Fredrik B Carlson et al\, arXiv (2026). \nAbstract: \nThis work introduces the concept of tangent space regularization for neural-network models of dynamical systems. The tangent space to the dynamics function of many physical systems of interest in control applications exhibits useful properties\, e.g.\, smoothness\, motivating regularization of the model Jacobian along system trajectories using assumptions on the tangent space of the dynamics. Without assumptions\, large amounts of training data are required for a neural network to learn the full non-linear dynamics without overfitting. We compare different network architectures on one-step prediction and simulation performance and investigate the propensity of different architectures to learn models with correct input-output Jacobian. Furthermore\, the influence of L2 weight regularization on the learned Jacobian eigenvalue spectrum\, and hence system stability\, is investigated.
URL:https://www.ibs.re.kr/bimag/event/tangent-space-regularization-for-neural-network-models-of-dynamical-systems-olive-cawiding/
LOCATION:108\, Conference Room\, IBS\, 55 Expo-ro Yuseong-gu\, Daejeon\, Korea\, Republic of
CATEGORIES:Journal Club
ORGANIZER;CN="Jae Kyoung Kim":MAILTO:jaekkim@kaist.ac.kr
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Seoul:20260828T100000
DTEND;TZID=Asia/Seoul:20260828T120000
DTSTAMP:20260730T040743Z
CREATED:20260730T040743Z
LAST-MODIFIED:20260730T040743Z
UID:12776-1787911200-1787918400@www.ibs.re.kr
SUMMARY:Towards a General Intelligence and Interface for Wearable Health Data - Aqsa Awan
DESCRIPTION:In this talk\, we discuss the paper “Towards a General Intelligence and Interface for Wearable Health Data” by Girish Narayanswamy et al.\, arXiv\, 2026. \nAbstract: \nWhile ubiquitous wearable sensors capture a wealth of behavioral and physiological information\, effectively transforming these signals into personalized health insights is challenging. Specifically\, converting low-level sensor data into representations capable of characterizing higher-level states is difficult due to high phenotypic diversity and variation in individual baseline health\, physiology\, and lifestyle factors. Moreover\, collecting wearable data paired with health outcome annotations is laborious and expensive\, and retrospective annotation remains practically unfeasible\, contributing to a scarcity of data with high-quality labels. To overcome these limitations\, we propose a foundation model for wearable health that is pretrained on more than one trillion minutes of unlabeled sensor signals drawn from a large cohort of five million participants. We demonstrate that the joint scaling of model capacity and pretraining data volume leads to systematic improvements in performance\, as evaluated on a diverse set of 35 health prediction tasks\, spanning cardiovascular\, metabolic\, sleep\, and mental health\, as well as lifestyle choices and demographic factors. We find that this population scale representation unlocks label-efficient few-shot learning and generative capabilities for robust daily metric estimation. To further leverage this learned representation\, we deploy a classroom of LLM agents to autonomously search the space of downstream predictive heads built on the model embeddings\, showing broad performance improvements that increase with LLM model capacity. Finally\, we show how integrating these downstream predictors into a Personal Health Agent can support model responses that are more relevant\, contextually aware\, and safe\, and we validate this via 1\,860 ratings from a cohort of clinicians.
URL:https://www.ibs.re.kr/bimag/event/towards-a-general-intelligence-and-interface-for-wearable-health-data-aqsa-awan/
LOCATION:108\, Conference Room\, IBS\, 55 Expo-ro Yuseong-gu\, Daejeon\, Korea\, Republic of
CATEGORIES:Journal Club
ORGANIZER;CN="Jae Kyoung Kim":MAILTO:jaekkim@kaist.ac.kr
END:VEVENT
END:VCALENDAR