Muse Spark Safety and Preparedness Report
A comprehensive report for Muse Spark on preparedness, adversarial robustness, and model behaviors.
Research Publications
Publications
AI Safety and LLM Security
Red teaming, jailbreaks, malicious-use evaluation, and safety benchmarks.Agents, Web, and Evaluation
Benchmarks, monitoring, data contamination, and web-agent guardrails.Robustness and Certification
Certified robustness, robust training, and adversarial robustness for vision models.Unlocking Deterministic Robustness Certification on ImageNet
Advances in Neural Information Processing Systems 36, 42993-43011, 2023
Interpretability and Explainability
Attribution, counterfactuals, graph explanations, BERT influence flow, and explainability tooling.On the Feature Alignment of Deep Vision Models Explainability and Robustness Connected at Hip
Carnegie Mellon University, 2023
Exploring Conceptual Soundness with TruLens
NeurIPS 2021 Competitions and Demonstrations Track, 302-307, 2022
Machine learning explainability and robustness: connected at the hip
Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data …, 2021
Interpreting interpretations: Organizing attribution methods by criteria
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern …, 2020
Score-CAM: Score-weighted visual explanations for convolutional neural networks
Proceedings of the IEEE/CVF conference on computer vision and pattern …, 2020
Formal Methods and Systems
Neuro-symbolic reasoning, SMT grounding, and practical ML systems.Grounding neural inference with satisfiability modulo theories
Advances in Neural Information Processing Systems 36, 22794-22806, 2023