arXiv subjects

Junwoo Kim

Publications and source records attributed to Junwoo Kim.

4 recordsLinked to original sources

arXiv2026

HELEA: Hard-Negative Benchmark and LLM-based Reranking for Robust Entity Alignment

Entity Alignment (EA) is essential for knowledge graph (KG) fusion, but existing benchmarks often allow models to exploit name overlap rather than relational structure. This makes it difficult to evaluate whether models can reject same-name entities that refer to different real-world objects. Our primary contribution is a same-name hard-negative augmentation strategy that simultaneously yields quality-controlled evaluation benchmarks (DW-HN29K, DY-HN27K) and augmented training corpora (DW-Train, DY-Train), by mining same-name but distinct entity pairs from KG name-collision groups. We further introduce HELEA, a two-stage framework integrating (i) entity encoder retrieval trained on hard-negative-augmented training corpora with 1-hop KG context, and (ii) LLM-based reranking without additional training. Experiments show that name-dependent baselines collapse to near-random performance on our hard-negative benchmarks, while HELEA achieves F1 0.967 on DW-HN29K while maintaining Hit@1 0.993 on standard DW-15K.

cs.CL↗

arXiv2026

Mi:dm 2.0 Korea-centric Bilingual Language Models

We introduce Mi:dm 2.0, a bilingual large language model (LLM) specifically engineered to advance Korea-centric AI. This model goes beyond Korean text processing by integrating the values, reasoning patterns, and commonsense knowledge inherent to Korean society, enabling nuanced understanding of cultural contexts, emotional subtleties, and real-world scenarios to generate reliable and culturally appropriate responses. To address limitations of existing LLMs, often caused by insufficient or low-quality Korean data and lack of cultural alignment, Mi:dm 2.0 emphasizes robust data quality through a comprehensive pipeline that includes proprietary data cleansing, high-quality synthetic data generation, strategic data mixing with curriculum learning, and a custom Korean-optimized tokenizer to improve efficiency and coverage. To realize this vision, we offer two complementary configurations: Mi:dm 2.0 Base (11.5B parameters), built with a depth-up scaling strategy for general-purpose use, and Mi:dm 2.0 Mini (2.3B parameters), optimized for resource-constrained environments and specialized tasks. Mi:dm 2.0 achieves state-of-the-art performance on Korean-specific benchmarks, with top-tier zero-shot results on KMMLU and strong internal evaluation results across language, humanities, and social science tasks. The Mi:dm 2.0 lineup is released under the MIT license to support extensive research and commercial use. By offering accessible and high-performance Korea-centric LLMs, KT aims to accelerate AI adoption across Korean industries, public services, and education, strengthen the Korean AI developer community, and lay the groundwork for the broader vision of K-intelligence. Our models are available at https://huggingface.co/K-intelligence. For technical inquiries, please contact midm-llm@kt.com.

cs.CL↗

arXiv2022

Precise and extensive characterization of an optical resonator for cavity-based quantum networks

Cavity-based quantum node is a competitive platform for distributed quantum networks. Here, we characterize a high-finesse Fabry-Perot optical resonator for coupling single or few atomic quantum registers. Our cavity consists of two mirrors with different reflectivities: One has minimal optical loss, and the other high transmission loss where more than 90% of the intracavity photons would be emitted. Cavity finesse, birefringent effects, and mechanical resonances are measured using the lasers at 780, 782, and 795 nm. In order to obtain cavity geometric parameters, we drive the adjacent longitudinal or transverse modes with two lasers simultaneously, and measure those frequencies using a precision wavelength meter (WLM). A major novelty of this method is that the parameters' uncertainties are solely determined by the resolution of the WLM, eliminating all of the temporal environment fluctuations. Our scheme makes it possible to quantify the atom-cavity coupling constant up to four significant figures, the most precise and accurate estimation so far, which would become a key ingredient for benchmarking a cavity-based quantum node. Furthermore, the distortion of polarized photonic qubits would be minimized owing to the small birefringent splitting, below 4.9% of the cavity linewidth. Our system should operate in the intermediate atom-cavity coupling regime that would allow us to implement various quantum network protocols.

quant-ph↗

arXiv2021

Locking Multi-laser Frequencies to a Precision Wavelength Meter: Application to Cold Atoms

We herein report a simultaneous frequency stabilization of two 780-nm external cavity diode lasers using a precision wavelength meter (WLM). The laser lock performance is characterized by the Allan deviation measurement in which we find $σ_{y}=10^{-12}$ at an averaging time of 1000 s. We also obtain spectral profiles through a heterodyne spectroscopy, identifying the contribution of white and flicker noises to the laser linewidth. The frequency drift of the WLM is measured to be about 2.0(4) MHz over 36 hours. Utilizing the two lasers as a cooling and repumping field, we demonstrate a magneto-optical trap of $^{87}$Rb atoms near a high-finesse optical cavity. Our laser stabilization technique operates at broad wavelength range without a radio frequency element.

quant-ph↗