KyungTae Lim

Semantic Hardness Is Not Visual Hardness: Sign-Aware Hard Negative Mining for Sign Language Retrieval

Association for Computational Linguistics 2026, S Cho, ChangSu Choi, Fitsum Gaim et autres

Sign Language Retrieval (SLRet) enables efficient access to sign language content but remains fragile in fine-grained scenarios where visually similar signs must be distinguished. We show that this limitation does not stem from model capacity, but from ineffective hard negative supervision. Specifically, …

kr, ca (code pays fourni par la source)

Unveiling the Performance Gap Between Deep Text Mining and Human Coders in Environmental Information Extraction

JungJin Kim, KyungTae Lim, Jan Adamowski, Hanseok Jeong

Automated text mining is increasingly used in government monitoring and policy-oriented workflows, but its value depends on how closely automated labels align with human interpretation. This study compares a BiLSTM classifier with two trained human coders in classifying Korean environmental news across …

Unveiling the Performance Gap Between Deep Text Mining and Human Coders in Environmental Information Extraction

JungJin Kim, KyungTae Lim, Jan Adamowski, Hanseok Jeong

Automated text mining is increasingly used in government monitoring and policy-oriented workflows, but its value depends on how closely automated labels align with human interpretation. This study compares a BiLSTM classifier with two trained human coders in classifying Korean environmental news across …

Unveiling the Performance Gap Between Deep Text Mining and Human Coders in Environmental Information Extraction

Jinwoo Kim, KyungTae Lim, Jan Adamowski, Hanseok Jeong

Automated text mining is increasingly used in government monitoring and policy-oriented workflows, but its value depends on how closely automated labels align with human interpretation. This study compares a BiLSTM classifier with two trained human coders in classifying Korean environmental news across …

kr, ca (code pays fourni par la source)

Analyzing the effect of reasoning-based supervision on face anti-spoofing

Jimin Min, KyungTae Lim, Minjun Kim, Dongsu Kim et autres

Face anti-spoofing (FAS) has become a crucial component in securing face recognition systems against presentation attacks, such as printed photos, replay videos, and 3D masks. While recent advances have improved generalization to unseen spoofing attempts, many existing methods remain black-box models that …

kr (code pays fourni par la source)

TELLME: Test-Enhanced Learning for Language Model Enrichment

Association for Computational Linguistics 2026, Wooyoung Go, Minjun Kim, MinKyu Kim et autres

Continual pre-training (CPT) has been widely adopted as a method for domain expansion in large language models. However, CPT has consistently been accompanied by challenges, such as the difficulty of acquiring large-scale domain-specific datasets and high computational costs. In this study, we …

kr, ca (code pays fourni par la source)

TReX: Tokenizer Regression for Optimal Data Mixture

Association for Computational Linguistics 2026, Minkyung Cho, KyungTae Lim, Jungyeul Park et autres

Building effective tokenizers for multilingual Large Language Models (LLMs) requires careful control over language-specific data mixtures. While a tokenizer’s compression performance critically affects the efficiency of LLM training and inference, existing approaches rely on heuristics or costly large-scale searches to determine optimal …

kr, ca (code pays fourni par la source)

Beyond Accuracy: Alignment and Error Detection across Languages in the Bi-GSM8K Math-Teaching Benchmark

Association for Computational Linguistics 2026, KyungTae Lim, JOON-HO LIM, Jieun Park

Recent advancements in LLMs have significantly improved mathematical problem-solving, with models like GPT-4 achieving human-level performance. However, proficiently solving mathematical problems differs fundamentally from effectively teaching mathematics. To bridge this gap, we introduce the Bi-GSM8K benchmark, a bilingual English-Korean dataset enriched with …

kr, ca (code pays fourni par la source)

ELO: Efficient Layer-Specific Optimization for Continual Pretraining of Multilingual LLMs

Association for Computational Linguistics 2026, SH Alexalex225225, ChangSu Choi, Minjun Kim et autres

We propose an efficient layer-specific optimization (ELO) method designed to enhance continual pretraining (CP) for specific languages in multilingual large language models (MLLMs). This approach addresses the common challenges of high computational cost and degradation of source language performance associated with traditional …

kr, ca (code pays fourni par la source)

Enriching the Korean learner corpus for grammatical error correction and writing assessment

Jayoung Song, KyungTae Lim, Jungyeul Park

Abstract Despite growing global interest in Korean language education, learner corpora tailored to Korean L2 writing remain scarce. This paper introduces KoLLA v2.0, the first Korean learner corpus to incorporate both multi-reference GEC annotations and rubric-based essay scoring. We extend the original …

us, kr (code pays fourni par la source)

Beyond Accuracy: Alignment and Error Detection across Languages in the Bi-GSM8K Math-Teaching Benchmark

Jieun Park, KyungTae Lim, Joon-ho Lim

Enhanced Conditional Generation of Double Perovskite by Knowledge-Guided Language Model Feedback

Junhyeong Lee, Jong‐Won Park, KyungTae Lim, Seunghwa Ryu

Double perovskites (DPs) are promising candidates for sustainable energy technologies due to their compositional tunability and compatibility with low-energy fabrication, yet their vast design space poses a major challenge for conditional materials discovery. This work introduces a multi-agent, text gradient-driven framework that …