Dongmin Kim

AAAI Conference 2026 System Paper

AEGIS: Toward Expert-in-the-loop Industrial Anomaly Detection

• Dongmin Kim
• Ye Seul Sim
• Suhee Yoon
• Sanghyu Yoon
• Seungdong Yoa
• Soonyoung Lee
• Woohyung Lim

Anomaly detection platforms in real-world environments require continuous interaction between automated systems and domain experts, as anomalies evolve dynamically and their definitions vary across contexts. Therefore, an effective platform must collaborate with experts and incorporate their feedback to update the system. This paper introduces AEGIS, an anomaly detection platform that aims to support interaction between domain experts and data-driven agents through three core capabilities: (1) data-driven insights through real-time monitoring, explanations, and distribution shift detection, which invoke customized tools to generate appropriate responses, (2) an expert feedback interface for labeling and direct updates via chat-based interaction, and (3) autonomous model construction that leverages expert-labeled data with LLM-driven hyperparameter optimization. Through this design, AEGIS fosters continuous interaction in which the platform provides insights while experts guide model improvement, ensuring user intent is reflected and robustness is maintained under evolving data distributions.

AAAI Conference 2025 Conference Paper

Representation Space Augmentation for Effective Self-Supervised Learning on Tabular Data

• Moonjung Eo
• Kyungeun Lee
• Hye-Seung Cho
• Dongmin Kim
• Ye Seul Sim
• Woohyung Lim

Tabular data, widely used across industries, remains underexplored in deep learning. Self-supervised learning (SSL) shows promise for pre-training deep neural networks (DNNs) on tabular data, but its potential is hindered by challenges in designing suitable augmentations. Unlike image and text data, where SSL leverages inherent spatial or semantic structures, tabular data lacks such explicit structure. This makes traditional input-level augmentations, like modifying or removing features, less effective due to difficulties in balancing critical information preservation with variability. To address these challenges, we propose RaTab, a novel method that shifts augmentation from input-level to representation-level using matrix factorization, specifically truncated SVD. This approach preserves essential data structures while generating diverse representations by applying dropout at various stages of the representation, thereby significantly enhancing SSL performance for tabular data.

AAAI Conference 2024 Conference Paper

When Model Meets New Normals: Test-Time Adaptation for Unsupervised Time-Series Anomaly Detection

• Dongmin Kim
• Sunghyun Park
• Jaegul Choo

Time-series anomaly detection deals with the problem of detecting anomalous timesteps by learning normality from the sequence of observations. However, the concept of normality evolves over time, leading to a "new normal problem", where the distribution of normality can be changed due to the distribution shifts between training and test data. This paper highlights the prevalence of the new normal problem in unsupervised time-series anomaly detection studies. To tackle this issue, we propose a simple yet effective test-time adaptation strategy based on trend estimation and a self-supervised approach to learning new normalities during inference. Extensive experiments on real-world benchmarks demonstrate that incorporating the proposed strategy into the anomaly detector consistently improves the model's performances compared to the existing baselines, leading to robustness to the distribution shifts.

NeurIPS Conference 2022 Conference Paper

WaveBound: Dynamic Error Bounds for Stable Time Series Forecasting

• Youngin Cho
• Daejin Kim
• Dongmin Kim
• MOHAMMAD AZAM KHAN
• Jaegul Choo

Time series forecasting has become a critical task due to its high practicality in real-world applications such as traffic, energy consumption, economics and finance, and disease analysis. Recent deep-learning-based approaches have shown remarkable success in time series forecasting. Nonetheless, due to the dynamics of time series data, deep networks still suffer from unstable training and overfitting. Inconsistent patterns appearing in real-world data lead the model to be biased to a particular pattern, thus limiting the generalization. In this work, we introduce the dynamic error bounds on training loss to address the overfitting issue in time series forecasting. Consequently, we propose a regularization method called WaveBound which estimates the adequate error bounds of training loss for each time step and feature at each iteration. By allowing the model to focus less on unpredictable data, WaveBound stabilizes the training process, thus significantly improving generalization. With the extensive experiments, we show that WaveBound consistently improves upon the existing models in large margins, including the state-of-the-art model.

ICML Conference 2010 Conference Paper

A scalable trust-region algorithm with application to mixed-norm regression

• Dongmin Kim
• Suvrit Sra
• Inderjit S. Dhillon

IROS Conference 1993 Conference Paper

Determination of displacement and velocity of an object handled by a three-fingered robot hand

• Dongmin Kim
• Hariharan Krishnan

The authors consider the problem of determining the displacement (position and orientation) and the velocity (linear and angular) of a rigid body handled by a three-fingered robot hand with frictional point contact. Direct measurement of configuration and velocity of the grasped object is difficult in practice. It is well known that the displacement of the rigid body is completely specified by the initial and final positions of three noncolinear points fixed on the object. Similarly, the instantaneous velocity of the rigid body can be determined from the linear velocity of three noncolinear points fixed on the body. Previously developed algorithms are computationally involved by the need to compute several trigonometric functions or matrix inverses. The authors present alternate computationally efficient algorithms for determining the displacement and the velocity of an object handled by a three-fingered robot hand.

ICRA Conference 1989 Conference Paper

Adaptive coordinated motion control of two manipulator arms

• Michael W. Walker
• Dongmin Kim
• Joseph Dionise

An adaptive controller is presented for the coordinated motion control of two manipulators handling an object of unknown mass. Global convergence in tracking of both position and internal force trajectories of the object is proved, assuming perfect models for both manipulators. The computational algorithm is similar to the Newton-Euler inverse dynamics algorithm with complexity linear in the number of links in the manipulators. A significant feature of the control method is that both manipulators use the identical computational algorithm. Thus, the concept of master/slave relationship between the two manipulators is avoided. Two simulations are presented. The first assumes an ideal model for each manipulator. As expected, the controller is stable and provides excellent tracking ability of the manipulator. The second simulation investigates the effects of modeling errors in the mass properties of each arm. The position trajectories track very close to their desired values; however, they never completely converge. >