Safe, Robust & Resilient AI Publications

Overview

Publications that are relevant to Safe, Robust & Resilient AI. The selection emphasizes work that contributes to robustness under perturbations, risk/uncertainty-aware decision making, privacy-preserving learning, trustworthy/reliable AI behavior, security and AI forensics, and protection/attribution of model IP.

Adversarial Robustness

• 2026
  J Dong, Y Wen, Z Li, W Gao, Z Zhang, S Wang, X Qu, YS Ong
  SwiftDistill: Efficient robust knowledge transfer via dual-branch adversarial distillation and hardness-balanced augmentation (Pattern Recognition, 2026)

  Brief: Transfers adversarial robustness efficiently through dual-branch distillation and hardness-balanced augmentation.

• 2026
  D Liu, Y Tao, J Dong, K Tang, P Zhou, W Hu, YS Ong
  Hard-Label Black-Box Attacks on 3D Point Clouds (IEEE Transactions on Dependable and Secure Computing, 2026)

  Brief: Evaluates 3D point-cloud vulnerability when an attacker can query only predicted labels.

• 2026
  J Dong, C Zhang, X Qu, QR Sua, T Nguyen, W Pan, X Li, T Liu, P Koniusz, YS Ong
  Tug-of-War No More: Harmonizing Accuracy and Robustness in Vision-Language Models via Stability-Aware Task Vector Merging (The Fourteenth International Conference on Learning Representations, 2026)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2026
  J Dong, RZ Moayedi, YS Ong, SM Moosavi-Dezfooli
  Allies Teach Better than Enemies: Inverse Adversaries for Robust Knowledge Distillation (IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2026
  J Dong, Y Zhang, H Zhu, YS Ong, P Koniusz
  Hierarchically Robust Zero-shot Vision-language Models (Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2025
  J Dong, C Zhang, X Qu, Z Ma, P Koniusz, YS Ong
  Robust SuperAlignment: Weak-to-Strong Robustness Generalization for Vision-Language Models (The Thirty-Ninth Annual Conference on Neural Information Processing Systems, 2025) (Spotlight)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2025
  J Dong, J Liu, X Qu, YS Ong
  Confound from All Sides, Distill with Resilience: Multi-Objective Adversarial Paths to Zero-Shot Robustness (Proceedings of the IEEE/CVF International Conference on Computer Vision, 2025) (Highlight)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2025
  J Dong, P Koniusz, Y Zhang, H Zhu, W Liu, X Qu, YS Ong
  Improving Zero-Shot Adversarial Robustness in Vision-Language Models by Closed-form Alignment of Adversarial Path Simplices (International Conference on Machine Learning, 2025) (Spotlight; PREMIA Best Student Paper Award 2025 — Merit Award)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2025
  J Dong, P Koniusz, X Qu, YS Ong
  Stabilizing modality gap & lowering gradient norms improve zero-shot adversarial robustness of vlms (Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2025)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2024
  J Dong, P Koniusz, J Chen, YS Ong
  Adversarially robust distillation by reducing the student-teacher variance gap (European Conference on Computer Vision, 2024)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2024
  J Dong, P Koniusz, J Chen, X Xie, YS Ong
  Adversarially robust few-shot learning via parameter co-distillation of similarity and class concept learners (Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2024
  J Dong, Y Wang, X Xie, J Lai, YS Ong
  Generalizable and discriminative representations for adversarially robust few-shot learning (IEEE Transactions on Neural Networks and Learning Systems, 2024)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2024
  J Dong, P Koniusz, J Chen, ZJ Wang, YS Ong
  Robust distillation via untargeted and targeted intermediate adversarial samples (Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2022
  A Chan, YS Ong, C Tan
  How does frequency bias affect the robustness of neural image classifiers against common corruption and adversarial perturbations? (International Joint Conferences on Artificial Intelligence, 2022)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2022
  X Qu, P Wei, M Gao, Z Sun, YS Ong, Z Ma
  Synthesising audio adversarial examples for automatic speech recognition (Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discover and Data Mining, 2022)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2021
  A Chan, L Ma, F Juefei-Xu, YS Ong, X Xie, M Xue, Y Liu
  Breaking neural reasoning architectures with metamorphic relation-based adversarial examples (IEEE Transactions on Neural Networks and Learning Systems, 2021)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2020
  A Chan, Y Tay, YS Ong
  Jacobian adversarially regularized networks for robustness (International Conference on Learning Representations, 2020)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2020
  A Chan, Y Tay, YS Ong, A Zhang
  Poison attacks against text datasets with conditional adversarially regularized autoencoder (Conference on Empirical Methods in Natural Language Processing, 2020)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2020
  A Chan, Y Tay, YS Ong
  What It Thinks Is Important Is Important: Robustness Transfers Through Input Gradients (Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020) (Oral)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2020
  X Han, T He, YS Ong, Y Zhong
  Precise object detection using adversarially augmented local/global feature fusion (Engineering Applications of Artificial Intelligence, 2020)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

Safety, Risk & Uncertainty

• 2026
  Y Ni, J Houssineau, YS Ong, P Koniusz
  Possibilistic Predictive Uncertainty for Deep Learning (International Conference on Machine Learning (ICML), 2026)

  Brief: Models epistemic uncertainty using possibilistic posterior predictions with an efficient training objective.

• 2026
  G Ai, C Niu, H Yan, JT Zhou, YS Ong, G Pang
  PromptDyG: Test-Time Prompt Adaptation on Dynamic Graphs (International Conference on Machine Learning (ICML), 2026)

  Brief: Adapts dynamic graph predictions at test time to address evolving patterns and distribution shift.

• 2026
  J Yang, J Tu, H Liu, X Wang, C Zheng, Z Zhang, S Cui, C Chen, T He, H Wang, YS Ong, M Huang
  BARREL: Boundary-Aware Reasoning for Factual and Reliable LRMs (International Conference on Learning Representations (ICLR), 2026)

  Brief: Improves factual reliability by encouraging reasoning models to recognize their knowledge boundaries.

• 2026
  T Peng, J Dong, YS Ong
  Learning Latent Concepts for Detecting Out-of-Distribution Objects (Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026) (Oral) (Best Paper Candidate)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2025
  N Lu, S Liu, J Wu, W Chen, Z Zhang, YS Ong, Q Wang, K Tang
  Safe Delta: Consistently Preserving Safety when Fine-Tuning LLMs on Diverse Datasets (International Conference on Machine Learning, 2025)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2025
  L Zhang, G Yu, J Yao, YS Ong, IW Tsang, JT Kwok
  Sandbox: safeguarded multi-label learning through safe optimal transport (Machine Learning, 2025)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2018
  J Rubio-Hervas, A Gupta, YS Ong
  Data-driven risk assessment and multicriteria optimization of UAV operations (Aerospace Science and Technology, 2018)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2016
  S Faghih-Roohi, YS Ong, S Asian, AN Zhang
  Dynamic conditional value-at-risk model for routing and scheduling of hazardous material transportation networks (Annals of Operations Research, 2016)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2014
  JJ Lim, AN Zhang, YS Ong, PS Tan
  A Risk-Averse Inventory Cost Model Using CVaR (Proceedings of the 18th Asia Pacific Symposium on Intelligent and Evolutionary Systems, 2014)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2013
  Z Li, PS Tan, G. Yee, QC Nyguen, YS Ong, XS Chen
  A review of complex systems technologies for supply chain risk management (IEEE International Conference on Systems, Man, and Cybernetics, 2013)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2013
  X Chen, YS Ong, PS Tan, NS Zhang, Z Li
  Agent-based modeling and simulation for supply chain risk management-a survey of the state-of-the-art (IEEE International Conference on Systems, Man, and Cybernetics, 2013)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2007
  D Lim, YS Ong, MH Lim, Y Jin
  Single/multi-objective inverse robust evolutionary design methodology in the presence of uncertainty (Evolutionary Computation in Dynamic and Uncertain Environments, 2007)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2006
  YS Ong, Z Zhou, D Lim
  Curse and blessing of uncertainty in evolutionary algorithm using approximation (IEEE international conference on evolutionary computation, 2006)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2005
  D Lim, YS Ong, BS Lee
  Inverse multi-objective robust evolutionary design optimization in the presence of uncertainty (Proceedings of the 7th annual workshop on genetic and evolutionary computation, 2005)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

Privacy & Data Governance

• 2025
  J Dong, H Zhu, Y Zhang, X Qu, YS Ong, P Koniusz
  Machine Unlearning via Task Simplex Arithmetic (The Thirty-Ninth Annual Conference on Neural Information Processing Systems, 2025)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2025
  FX Fan, C Tan, R Wattenhofer, YS Ong
  Position Paper: Rethinking Privacy in RL for Sequential Decision-making in the Age of LLMs (International Joint Conference on Neural Networks, 2025)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2025
  FX Fan, C Tan, YS Ong, R Wattenhofer, WT Ooi
  Fedrlhf: A convergence-guaranteed federated framework for privacy-preserving and personalized rlhf (Proceedings of the 24th International Conference on Autonomous Agents and Multiagent Systems, 2025)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2024
  K Pan, YS Ong, M Gong, H Li, AK Qin, Y Gao
  Differential privacy in deep learning: A literature survey (Neurocomputing, 2024)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

• 2024
  W Liu, X Zheng, C Chen, J Xu, X Liao, F Wang, Y Tan, YS Ong
  Reducing item discrepancy via differentially private robust embedding alignment for privacy-preserving cross domain recommendation (International Conference on Machine Learning, 2024)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.

Security & Forensics

• 2026
  W Hong, Z Rong, L Wang, F Chang, J Zhu, K Tang, Z Zhu, YS Ong
  Evolving Jailbreaks: Automated Multi-Objective Long-Tail Attacks on Large Language Models (IEEE Transactions on Evolutionary Computation, 2026)

  Brief: Uses multi-objective evolutionary search to systematically evaluate long-tail jailbreak vulnerabilities in language models.

• 2025
  A Żychowski, A Gupta, YS Ong, J Mańdziuk
  Augmented Decision Spaces for Stackelberg Security Games: Sparse evolution begets scalability (Proceedings of the Genetic and Evolutionary Computation Conference, 2025)

  Brief: This work contributes to Safe, Robust & Resilient AI by addressing robustness under uncertainty, reliability of learning systems, or safety-critical optimization and decision-making.