AI Safety

Trustworthy ai foundations

From Robust Learning to Stable Industrial & Physical AI

Trustworthy AI provides a system-wide foundation for every stage of Industrial & Physical AI, including learning from sensitive data, mitigating algorithmic bias, protecting deployed systems, adapting to changing environments, and maintaining robust and stable representations.
Our research develops learning algorithms and mathematical foundations for robust, stable, privacy-preserving, explainable, and verifiable AI. Key areas include robust perception, predictive process and world models, safe and certifiable decision-making, dynamical-system stability, verified and global optimization, and runtime safety assurance.

System-wide trustworthy AI foundations for Industrial & Physical AI:
privacy, fairness, security, and representation stability

Privacy-preserving & fair ai

Industrial & Physical AI systems often rely on sensitive visual, acoustic, behavioral, spatial, and industrial data. We develop privacy-preserving learning and inference methods based on differential privacy, private inference, and secure learning, together with fair AI methods that reduce group-level disparities.

Selected Publications

- Park et al., Multi-Class SVM with Differential Privacy, NeurIPS, 2025
- Choi et al., Safeguarding Retrieval Data against Membership Inference Attacks, EMNLP, 2025
- Park et al., Public Data Synthesis for Differentially Private Image Classification, CVPR, 2024
- Choi et al., Fair Sampling in Diffusion Models, AAAI, 2024
- Park et al., Differentially Private Sharpness-Aware Training, ICML, 2023
- and more

Privacy-preserving and fairness-aware learning for sensitive multimodal data generated by Industrial & Physical AI systems

Stable & Robust Machine Learning

We develop mathematical and algorithmic methods for stable representations, robust learning, domain adaptation, and reliable generalization from limited, noisy, or partially labeled data. These foundations support dependable perception, prediction, and decision-making under real-world uncertainty.

Selected Publications

- Kim et al., Voronoi Cell-Based Clustering Using a Kernel Support, IEEE TKDE, 2015
- Lee et al., Constructing Sparse Kernel Machines Using Attractors, IEEE TNN, 2009
- Kim and Lee, Clustering Based on Gaussian Processes, Neural Computation, 2007
- Lee and Lee, Equilibrium-Based SVM for Semi-Supervised Classification, IEEE TNN, 2007
- Lee et al., Dynamic Characterization of Cluster Structures for Robust SVC, IEEE TPAMI, 2006
- and more

Mathematical and algorithmic foundations for stable representations, robust clustering, model transferability, and reliable learning

Data-Centric AI Engineering

We develop data-centric methods for heterogeneous data analysis, multimodal integration, ensemble learning, clustering, recommendation, and large-scale data mining. These methods provide the data-engineering foundations for robust Industrial & Physical AI systems.

Selected Publications

- Park et al., Inductive Ensemble Clustering Using Kernel Support Matching, Electronics Letters, 2017
- Kim et al., Improved Churn Prediction by Analyzing a Large Network, Expert Systems with Applications, 2014
- Park et al., Transductive Bayesian Regression via Manifold Learning, Expert Systems with Applications, 2012
- Heo et al., Credit Delinquent Prediction Using Locally Transductive MLP, Neurocomputing, 2009
- Cho et al., A Tandem Clustering Process for Multimodal Datasets, European Journal of Operational Research, 2006
- and more

Foundations for heterogeneous data analysis and data driven decision support

INdustrial & physical AI Safety

What Makes Our Approach Distinctive

Industrial & Physical AI systems operate in dynamic and uncertain real-world environments, often interacting continuously with humans, machines, and physical processes. Such systems must understand human intent, perceive uncertain environments, predict future states, make reliable decisions, execute safe actions when interacting with physical systems, and continuously monitor their behavior and operating conditions.

Many studies in Industrial & Physical AI primarily focus on system architectures, demonstrations, or task-specific performance. We take a complementary perspective, investigating safety and trustworthiness through both learning-based methods and rigorous mathematical systems analysis, with particular emphasis on fairness, privacy, security, stability, robustness, and interpretability.

Robust perception & State estimation

We develop robust perception and state-estimation methods that integrate physical sensors and industrial data, including cameras, radar, LiDAR, acoustic and tactile sensors, process measurements, logs, and time-series signals, to estimate system, environment, and equipment states.
Our research focuses on maintaining reliable perception under sensor degradation, missing or corrupted observations, process uncertainty, communication failures, distribution shifts, unfamiliar environments, and adversarial attacks, enabling dependable Industrial & Physical AI systems in real-world operation

Selected Publications

- Kim et al., Local Geometry Attention for Time Series Forecasting under Realistic Corruptions, ICLR, 2026
- Kim et al., Towards Undetectable Adversarial Attack on Time Series Classification, Information Sciences, 2025
- Lee et al., Variational Cycle-Consistent Imputation Adversarial Networks for General Missing Patterns, Pattern Recognition, 2022
- Kim and Lee, Nonlinear Dynamic Projection for Noise Reduction of Dispersed Manifolds, IEEE TPAMI, 2014
- and more

Robust and reliable perception algorithms for Industrial & Physical AI in uncertain and adversarial environments

Predictive Process & world models

We develop predictive models of industrial processes and physical environments to forecast future system states, trajectories, interactions, and the consequences of actions.
Our research also focuses on generating realistic synthetic trajectories and rare or safety-critical scenarios, enabling robust training, simulation-based evaluation, and proactive safety assessment for Industrial & Physical AI systems.

Selected Publications

- Park et al., TimeBridge: Better Diffusion Prior Design with Bridge Models for Time Series Generation, KDD, 2026
- Kim et al., Are Self-Attentions Effective for Time Series Forecasting?, NeurIPS, 2024
- Park et al., Fast Sharpness-Aware Training for Periodic Time Series Classification and Forecasting, Applied Soft Computing, 2023
- Kim et al., Self-Correcting Ensemble Using a Latent Consensus Model, Applied Soft Computing, 2016
- and more

Temporal dynamics modeling for state prediction, realistic trajectory generation, and simulation based safety evaluation

Safe planning & robust decision Making

We develop planning and decision-making methods that translate high-level goals and predicted future states into actions that satisfy performance, operational, and safety constraints under uncertainty.
Our research focuses on learning algorithms that remain reliable under distribution shift, adversarial perturbations, model uncertainty, and changing operating conditions, providing foundations for safe planning, risk-aware control, and dependable decision-making in Industrial & Physical AI systems.

Selected Publications

- Kim et al., Fantastic Robustness Measures: The Secrets of Robust Generalization, NeurIPS, 2023
- Lee et al., Implicit Jacobian Regularization Weighted with Impurity of Probability Output, ICML, 2023
- Lee et al., GradDiv: Adversarial Robustness of Randomized Neural Networks via Gradient Diversity Regularization, IEEE TPAMI, 2022
- Kim et al., Understanding Catastrophic Overfitting in Single-Step Adversarial Training, AAAI, 2021
- and more

Learning theoretic foundations for dependable planning and risk aware decisions

Stable control & verified optimization

We develop control and optimization methods that translate decisions into reliable system actions, including process setpoints, motion, force, torque, and other forms of actuation, while maintaining stability, constraint satisfaction, and safe operating conditions.
Our research establishes mathematical foundations for stability analysis, safe and unsafe operating-region characterization, nonlinear systems, trajectory-based optimization, and global optimization, enabling reliable and verifiable operation of Industrial & Physical AI systems.

Selected Publications

- Lee et al., Trajectory-Based Method for Multiple Solutions in Nonlinear Programming, IEEE TAC, 2004
- Lee, Optimization Framework for Computing the Controlling UEP, IEEE TAC, 2004
- Lee et al., Stability Regions of Non-Hyperbolic Dynamical Systems, IEEE TCAS-I, 2002
- Lee et al., Constructive Homotopy for Multiple DC Operating Points, IEEE TCAS-I, 2001
- Lee, Trajectory-Informed Search for Global Optimization, Journal of Global Optimization, 2007
- and more

Mathematical foundations for safe physical behavior via nonlinear optimization and stability analysis

runtime safety assurance & explainability

We develop runtime safety-assurance methods that continuously monitor observations, system states, predictions, decisions, and actions to detect abnormal or unsafe behavior during operation.
Our research integrates anomaly and failure detection, explainable AI, runtime robustness assessment, and security and privacy protection with recovery, replanning, intervention, and human oversight, enabling Industrial & Physical AI systems to maintain safe and trustworthy behavior after deployment.

Selected Publications

- Jeong et al., JoCE: Joint Counterfactual Explanations for Interpretable Time Series Anomaly Detection, Pattern Recognition, 2026
- Choi et al.,Differentially Private Upsampling for Imbalanced Anomaly Detection , Engineering Applications of Artificial Intelligence, 2026
- Byun et al., Privacy-Preserving Inference Resistant to Model Extraction Attacks, Expert Systems with Applications, 2024
- and more

Runtime trustworthiness for Industrial & Physical AI:
explain anomalies, strengthen detection, and protect deployed models and data

Page updated

Google Sites

Report abuse