Graduate students

Soyeon Kim

PhD student (20.09 ~ )

PhD student, 2020 Fall – present Master of Landscape Architecture from Seoul National University, 2019 *outstanding master’s thesis award, summa cum laude Bachelor of Landscape Architecture from Kangwon National University, 2016 *summa cum laude)

PhD student, 2020 Fall – present
Master of Landscape Architecture from Seoul National University, 2019 *outstanding master’s thesis award, summa cum laude
Bachelor of Landscape Architecture from Kangwon National University, 2016 *summa cum laude)
Research interest: Feature Attribution, Representation Learning, Spatiotemporal Analysis, Geospatial Analysis, Domain Adaptation(Meteorology)
soyeon.k (at) kaist.ac.kr

3 notes

Research

All research

2025
Diverse Rare Image Generation
Latent optimization and feature-space density estimation generate rare images with controllable rarity, diversity, and reference similarity.
2025
Diverse Rare Sample Generation with Pretrained GANs
This study proposes a novel approach for generating diverse rare samples from high-resolution image datasets with pretrained GANs. Our method employs gradient-based optimization of latent vectors within a multi-objective framework and utilizes normalizing flows for density estimation on the feature space. This enables the generation of diverse rare images, with controllable parameters for rarity, diversity, and similarity to a reference image. We demonstrate the effectiveness of our approach both qualitatively and quantitatively across various datasets and GANs without retraining or fine-tuning the pretrained GANs.
2023
Explainable AI-Based Interface System for Weather Forecasting Model
As machine learning (ML) becomes more prevalent in meteorological decision-making, user-centered explainable artificial intelligence (XAI) studies have yet to extend into this domain. This study identifies three key requirements for explaining black-box models in meteorology: model bias in different rainfall scenarios, model reasoning, and output confidence. Appropriate XAI methods are mapped to these requirements and evaluated both quantitatively and qualitatively. The results show that intuitive explanations improve decision-making utility and user trust, highlighting the need for user-centered XAI to enhance the usability of AI systems in practice.

11 papers

Publications

Full record

2026

[114]
Diffusion Integrated Gradients: Controllable Path Generation for Flexible Feature Attribution
Soyeon Kim, Kyowoon Lee, and Jaesik Choi
European Conference on Computer Vision (ECCV) 2026.Explainable AI Generative models
arXiv
[112]
Spectral Integrated Gradients for Coarse-to-Fine Feature Attribution
Soyeon Kim, Seongwoo Lim, Kyowoon Lee, and Jaesik Choi
Knowledge Discovery and Data Mining (KDD) 2026.Explainable AI Data mining
arXiv
[109]
Manifold-Aligned Guided Integrated Gradients for Reliable Feature Attribution
Soyeon Kim, Seongwoo Lim, Kyowoon Lee, and Jaesik Choi
International Conference on Machine Learning (ICML), 2026.Explainable AI
Paper
[106]
K-MetBench: A Multi-Dimensional Benchmark for Fine-Grained Evaluation of Expert Reasoning, Locality, and Multimodality in Meteorology
Soyeon Kim, Cheongwoong Kang, Myeongjin Lee, Eun-Chul Chang, Jaedeok Lee, and Jaesik Choi
Findings of the Annual Meeting of the Association for Computational Linguistics (ACL), 2026.Language models Weather and climate
arXiv

2025

[102]
Unsupervised Concept Discovery for Deep Weather Forecast Models with High-Resolution Radar Data
Soyeon Kim, Junho Choi, Subeen Lee and Jaesik Choi
PLOS Climate, 2025.Explainable AI Weather and climate
arXiv
[97]
Example-Based Concept Analysis Framework for Deep Weather Forecast Models
Soyeon Kim, Junho Choi, Subeen Lee and Jaesik Choi
Artificial Intelligence for the Earth Systems (AIES), 2025.Explainable AI Weather and climate
Paper
[87]
Diverse Rare Sample Generation with Pretrained GANs
Subeen Lee, Jiyeon Han, Soyeon Kim and Jaesik Choi
AAAI Conference on Artificial Intelligence (AAAI), 2025.Generative models
Paper

2024

—
Capsule Neural Networks as Noise Stabilizer for Time Series Data
김소연, 성지현, 한현경, 최재식
정보과학회 논문지(Journal of KIISE), 2024.Time series

2023

[75]
Explainable AI-Based Interface System for Weather Forecasting Model
Soyeon Kim, Junho Choi, Yeji Choi, Subeen Lee, Artyom Stitsyuk, Minkyoung Park, Seongyeop Jeong, Youhyun Baek and Jaesik Choi
International Conference on Human-Computer Interaction International (HCII), 2023.Explainable AI Weather and climate
Paper
—
Capsule Neural Networks as Noise Stabilizer for Time Series Data
김소연, 성지현, 한현경, 최재식
한국컴퓨터종합학술대회(KCC), 2023. (우수논문상)Time series

2021

[56]
Conditional Temporal Neural Processes with Covariance Loss
Boseon Yu, Jiwoo Lee, Janghoon Ju, Seijun Chung, Soyeon Kim and Jaesik Choi
International Conference on Machine Learning (ICML), 2021.Time series
Paper

Soyeon Kim

Research

Diverse Rare Image Generation

Diverse Rare Sample Generation with Pretrained GANs

Explainable AI-Based Interface System for Weather Forecasting Model