[Session 1] 

9:00-9:10 Intro

Temporal adaptation 

09:10-09:35 현웅 | Dissecting Channel Dependency in Multivariate Time Series Forecasting: Interaction Scope, Level, and Architecture 

09:35-10:00 윤환 | Aperiodicity-Robust Spatio-Temporal Traffic Forecasting 

10:00-10:20 지연 | Value-aware Cross-Modal Alignment: Learning Numeric-Semantic Representations between Text and Time-Series 

10:20-10:30 Break

Streaming adaptation 

10:30-10:50 주형 | Systematic Literature Review of Online Deep Anomaly Detection

10:50-11:10 명준 | Online Anomaly Detection via Adaptive Boosting with Dynamic Weighting 

11:10-11:35 선호 | Causal Differential Equation Model for Anomaly Detection in Time Series

11:45-13:00 Lunch

[Session 2] Domain-specific adaptation 

13:00-13:25 지효 | AI for Biomedical Challenges: From Literature Insights to Real-world Applications 

13:25-13:50 우준 | Representation Learning: From Foundations to Structured Abstractions in Tabular Data 

13:50-14:10 연우 | Leveraging Intra-Column Relationships for Enhanced Table Understanding 

14:10-14:15 Break

Structured adaptation 

14:15-14:35 승원 | Overcoming Local Sparsity in Continual Knowledge Graph Learning 

14:35-15:00 희정 | Continual Retrieval on Dynamic Streaming Corpora with Adaptive Soft Memory 

15:00-15:20 원준 | Persona-adaptive Memory Module for LLM based Agents 

15:20-15:25 Break

[Session 3] Adaptation to DAIS 😉 

15:25-15:55 현직 | Soft Prompt–driven Multitask Learning with Task Group Adaptation 

15:55-16:25 민열 | From Forgetting to Improvement: Continual GraphRAG for Multi-hop QA 

16:25-16:55 예림 | Tool-Augmented Context Engineering for Hybrid KV-Cache Management in AI Agents 

16:55-17:00 Outro

[Joint Workshop with IDEA Lab on 2025/08/27]

현구 | MGRank: Unsupervised Keyphrase Extraction with Multi-granular Topic Saliency

수빈 | Semantic-Aware Vertical Partitioning for Efficient Table Retrieval

예빈 | Multi-level Diagnosis and Evaluation for Robust Tabular Feature Engineering with Large Language Models 

5:00-5:10 Introduction

Session 1. Into the unknown world (feat. tabular data)

5:10-5:40 수빈: Open-world machine learning and retrieval-enhanced tabular prediction 

• Reference: [P1] [P2]

5:45-6:15 명준: Concept-drifted world and retrieval-enhanced tabular anomaly detection 

• Reference: [P1] [P2]

6:15-6:25 Break

Session 2. Into the structured, continual world (feat. knowledge graph and agents)

6:25-6:55 승원: Knowledge graph for/from large language models

• Reference: [P1]

7:00-7:30 원준: Large language model to perceive, memorize, and act

• Reference: [P1]

7:35-8:00 Wrap-up

Schedule

3:00 - 3:10 Introduction

3:10 - 3:50 Session 1. Key information extraction

• 현구: Keyword extraction with multi-granular topic saliency 

• 지효: Textual and visual understanding of Organ-on-a-Chip research 

3:50 - 3:55 Short break

3:55 - 4:35 Session 2. Structural retrieval

• 희정: Continual retrieval on dynamic streaming corpora

• 민혁: Query-to-columns for human-like table retrieval system 

4:35 - 4:40 Short break

4:40 - 5:20 Session 3. Structural embedding

• 우준: Dynamic exploitation of tabular data by leveraging semi-structured knowledge

• 연우: InterCellFormer: leveraging intra-column relationships for enhanced table understanding 

5:20 - 5:35 Long break (Dinner pack)

5:35 - 6:15 Session 4. Evaluation and alignment

• 예빈: Towards robust feature engineering with large language models 

• 지연: Interpretable prototype: a bridge from text to time-series for LLM 

6:15 - 6:20 Short break

6:20 - 7:00 Session 5. Dependency-aware adaptation

• 현웅: Dissecting dependencies in multi-channel time series forecasting

• 윤환: Spatio-temporal prediction with graph neural network 

7:00 - 7:05 Short break

7:05 - 7:45 Session 6. Time-aware adaptation

• 선호: Look-ahead augmentation for time-series forecasting 

• 주형: Recent trends and frontiers of online deep anomaly detection

7:45 - 7:50 Short break

7:50 - 8:20 Wrap-up

8:20 - 9:00 Invited talk - 이유경 박사 (Boston University)

• Towards goal-oriented language intelligence with robust evaluation