About

I’m a Ph.D. student in the VeriVITAL lab, advised by Taylor T. Johnson. My research is at the intersection of formal methods and machine learning, focused on the safety and reliability of modern AI systems. I’m motivated by what verification can contribute to AI safety in an era of frontier models, where worst-case formal guarantees increasingly give way to questions about the quality of learned behavior. I’m also drawn to neurosymbolic approaches, where symbolic structure can provide interpretability and safety properties that are hard to establish in purely learned systems, and to the broader question of how training procedures shape model internals.

Contact: samuel.sasaki@vanderbilt.edu

Publications

• [NeurIPS'26]Samuel Sasaki, Ben Wooding, Taylor T. Johnson, Learning Reach-Set Geometry for Tighter Probabilistic Neural Network Verification, The Fortieth Annual Conference on Neural Information Processing Systems (NeurIPS'26), 2026.
• [ATVA'26]Anne M. Tumlin*, Samuel Sasaki*, Ben Wooding, Diego Manzanas Lopez, Muhammad Usama Zubair, Navid Hashemi, Hongchao Zhang, Waseem Abbas, Ipek Oguz, Meiyi Ma, Taylor T. Johnson, NNV3: Expanding Neural Network Verification to New Architectures and Domains, The 24th International Symposium on Automated Technology for Verification and Analysis (ATVA'26), December 2026.
• [SAIV'26]Samuel Sasaki, Ben Wooding, Hanchen David Wang, Anne M. Tumlin, Meiyi Ma, Taylor T. Johnson, n2v: Neural Network Verification in Python (Competition Contribution), The 9th International Symposium on AI Verification (SAIV'26), July 2026, Lisbon, Portugal.
• [SAIV'26]Quang Pham, Ben Wooding, Luke Nam, Samuel Sasaki, Taylor T. Johnson, MetaMoE: Formal Verification of Compositional Robustness and Scalability of Mixture-of-Experts Architecture, The 9th International Symposium on AI Verification (SAIV'26), July 2026.
• [NeurIPS'25]Navid Hashemi, Samuel Sasaki, Ipek Oguz, Meiyi Ma, Taylor T. Johnson, Scaling Data-Driven Probabilistic Robustness Analysis for Semantic Segmentation Neural Networks, The Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS'25), December 2025, San Diego, California.
• [ESORICS'25]Preston K. Robinette, Thuy Dung Nguyen, Samuel Sasaki, Taylor T. Johnson, Trigger-Based Fragile Model Watermarking for Image Transformation Networks, 30th European Symposium on Research in Computer Security (ESORICS'25), September 2025, Toulouse, France.
• [SAIV'25]Diego Manzanas Lopez, Samuel Sasaki, Taylor T. Johnson, NNV: A Star Set Reachability Approach (Competition Contribution), The 8th International Symposium on AI Verification (SAIV'25), July 2025, Zagreb, Croatia.
• [NeuS'25]Samuel Sasaki, Diego Manzanas Lopez, Taylor T. Johnson, Neurosymbolic Finite and Pushdown Automata: Improved Multimodal Reasoning versus Vision Language Models (VLMs), 2nd International Conference on Neuro-symbolic Systems (NeuS'25), May 2025, Philadelphia, Pennsylvania.
• [FormaliSE'25]Samuel Sasaki, Diego Manzanas Lopez, Preston K. Robinette, Taylor T. Johnson, Robustness Verification of Video Classification Neural Networks, 13th International Conference on Formal Methods in Software Engineering (FormaliSE'25), April 2025, Ottawa, Canada. Acceptance rate: 31% (15/48).
• [ARCH-COMP'26]Samuel Sasaki, Ben Wooding, Taylor T. Johnson, Matthias Althoff, Luis Benet, Samuel Coogan, Marcelo Forets, Akash Harapanahalli, Lukas Koller, Tobias Ladner, Christian Schilling, Huan Zhang, Xiangru Zhong, ARCH-COMP26 Category Report: Artificial Intelligence and Neural Network Control Systems (AINNCS) for Continuous and Hybrid System Plants, 13th International Workshop on Applied Verification of Continuous and Hybrid Systems (ARCH-COMP'26), 2026.
• [Preprint]Navid Hashemi, Samuel Sasaki, Diego Manzanas Lopez, Ipek Oguz, Meiyi Ma, Taylor T. Johnson, Probabilistic Robustness Analysis in High Dimensional Space: Application to Semantic Segmentation Network, arXiv, November 2025.

Service

Conference Organization

[ARCH-COMP’26] Co-lead, AINNCS track.

Artifact Evaluation

[QEST+FORMATS’25] Member of the artifact evaluation committee.

Outreach

[YSJ’26] Reviewer, Young Scientist Journal. A peer-reviewed journal of high school research.

Teaching

Teaching assistant for the Introduction to Engineering - Computer Science course at Vanderbilt University in Fall 2023. Responsibilities included holding office hours, grading assignments, and assisting in-class activities.

Teaching assistant for the Embedded Systems course at Vanderbilt University in Spring 2024. Responsibilities included leading weekly lab sections in which students would work through assignments, managing distribution of equipment, and grading assignments.