About

Radu Grosu is a Professor and Head of the Dependable-Systems Group at the Faculty of Informatics of the Vienna University of Technology, and a Research Professor at the Computer Science Department of the State University of New York at Stony Brook. He earned his Dr.rer.nat. in Computer Science from the Technical University of München and was a Research Associate in the Computer Science Department of the University of Pennsylvania. Grosu's primary research focus is on developing formal methods and tools to support the modeling and automated analysis of complex computational systems, including software, embedded, and biological systems, with an emphasis on approaches that scale well for realistic applications. His notable contributions include establishing a noncommutative Cayley-Hamilton theorem for finite automata, relating minimal nondeterministic finite automata via linear transformations, automatically detecting emergent properties in networks of cardiac myocytes, learning efficient models for excitable cells, and defining advanced model checking techniques that balance time, space, and confidence. Additionally, he has contributed to the development of compositional models for automata, semantics and refinement rules for UML diagrams, and denotational semantics for reconfigurable systems. Grosu has received several awards, including the National Science Foundation Career Award, the State University of New York Research Foundation Promising Inventor Award, and the ACM Service Award.

Research topics

• Computer Science
• Artificial Intelligence
• Machine Learning
• Mathematics
• Applied mathematics
• Physics
• Mathematical analysis
• Statistics

Selected publications

• Looped SSMs: Depth-Recurrence and Input Reshaping for Time Series Classification

  ArXiv.org · 2026-05-15

  articleOpen accessSenior author

  State Space Models (SSMs) are inherently recurrent along the sequence dimension, yet depth-recurrence - reusing the same block repeatedly across layers, as recently applied in looped transformers - has not been explored in this model family. We show that a looped SSM with $k$ parameters iterated $L$ times consistently closely matches or outperforms a standard SSM with $k \cdot L$ independent parameters across four architectures (LRU, S5, LinOSS, LrcSSM) and six time series classification benchmarks, despite operating within a strictly smaller hypothesis space, as we formally establish. Since the larger model contains the looped model as a special case, this dominance cannot be explained by expressivity and instead points to parameter sharing across depth as a beneficial inductive bias that simplifies optimization. These results demonstrate that depth-recurrence is orthogonal to sequence-recurrence and independently beneficial. We further show that input reshaping is an equally neglected design axis: concatenating timesteps for low-dimensional inputs, or flattening and rechunking the joint feature-time dimension for high-dimensional ones, yields accuracy gains of 1-6% across all models, confirmed over 5 random seeds. Both techniques provide standalone improvements that compound when combined, suggesting that depth and input reshaping are two independent and underexplored design axes for SSMs on time series.

  PublisherOA PDF

• Synaptic Activation and Dual Liquid Dynamics for Interpretable Bio-Inspired Models

  arXiv (Cornell University) · 2026-02-13

  articleOpen accessSenior author

  In this paper, we present a unified framework for various bio-inspired models to better understand their structural and functional differences. We show that liquid-capacitance-extended models lead to interpretable behavior even in dense, all-to-all recurrent neural network (RNN) policies. We further demonstrate that incorporating chemical synapses improves interpretability and that combining chemical synapses with synaptic activation yields the most accurate and interpretable RNN models. To assess the accuracy and interpretability of these RNN policies, we consider the challenging lane-keeping control task and evaluate performance across multiple metrics, including turn-weighted validation loss, neural activity during driving, absolute correlation between neural activity and road trajectory, saliency maps of the networks' attention, and the robustness of their saliency maps measured by the structural similarity index.

  PublisherOA PDF

• Supplementary dataset for paper: "Towards Safe Learning-Based Non-Linear Model Predictive Control through Recurrent Neural Network Modeling"

  arXiv (Cornell University) · 2026-03-26

  datasetOpen access

  Supplementary datasets and pretrained models for the paper "Towards Safe Learning-Based Non-Linear Model Predictive Control through Recurrent Neural Network Modeling" (Clement et al., 2026). This record contains the data and models used in the Sequential-AMPC framework available at the GitHub repository seq-ampc(1) vehicle kinematic-obstacle and vehicle dynamic-obstacle datasets generated for this work; and(2) pretrained neural network models used in the experiments. The original SOEAMPC supplementary dataset is available at DOI: 10.5281/zenodo.7846094 from which the quadcopter dataset was used. The code used to generate, train, and evaluate the models in this record is available at the associated GitHub repository for Sequential-AMPC. File organization: vehicle_obs_N_55000.tar.lz: kinematic bicycle model with static obstacle avoidance vehicle_8state_obs_N_116000.tar.lz: dynamic single-track vehicle model with static obstacle avoidance models.zip: pretrained models for the experiments in this work For each dataset, files contain initial conditions, MPC input trajectories, predicted state sequences, and associated parameters as documented in the repository README.

  PublisherOA PDFDOI

• Multi-structure segmentation in CBCT volumes: The ToothFairy2 challenge

  Medical Image Analysis · 2026-04-23

  articleOpen access

  Cone-beam computed tomography (CBCT) is widely used for dento-maxillofacial diagnostics and treatment planning, and comprehensive multi-structure segmentation remains time-consuming, limiting large-scale, reproducible research. In this article, we present ToothFairy2, a MICCAI 2024 challenge on multi-structure segmentation in maxillofacial CBCT. The accompanying dataset comprises 530 CBCT volumes (480 public training, 50 hidden test) with expert 3D annotations of 42 classes, including maxilla, mandible, crowns, bridges, implants, inferior alveolar canals, maxillary sinuses, pharynx, and teeth labeled according to the International Tooth Numbering System (FDI). 26 international teams participated in ToothFairy2, and their methods were run and evaluated for voxel-wise multi-class segmentation using a standardized protocol. This report extends the evaluation of teeth to also investigate the current capabilities of tooth detection and FDI numbering. Furthermore, ranking stability was analyzed to assess the robustness of the final challenge outcome. Overall, challenge participants achieved consistently high performance for large, high-contrast structures such as jawbones, pharynx, and most teeth, while maxillary sinuses, dental restorations, and fine structures remain challenging due to class imbalance and metal artifacts. Analysis of tooth-related metrics further revealed that assigning correct FDI numbers was more challenging than delineating individual teeth. By releasing CBCT data, 3D annotations, baseline models, and evaluation code, ToothFairy2 establishes a long-term benchmark to drive the development of automated methods for robust, clinically meaningful multi-structure segmentation in maxillofacial CBCT.

  PublisherDOI

• Supplementary dataset for paper: "Towards Safe Learning-Based Non-Linear Model Predictive Control through Recurrent Neural Network Modeling"

  Zenodo (CERN European Organization for Nuclear Research) · 2026-03-26

  datasetOpen access

  Supplementary datasets and pretrained models for the paper "Towards Safe Learning-Based Non-Linear Model Predictive Control through Recurrent Neural Network Modeling" (Clement et al., 2026). This record contains the data and models used in the Sequential-AMPC framework available at the GitHub repository seq-ampc(1) vehicle kinematic-obstacle and vehicle dynamic-obstacle datasets generated for this work; and(2) pretrained neural network models used in the experiments. The original SOEAMPC supplementary dataset is available at DOI: 10.5281/zenodo.7846094 from which the quadcopter dataset was used. The code used to generate, train, and evaluate the models in this record is available at the associated GitHub repository for Sequential-AMPC. File organization: vehicle_obs_N_55000.tar.lz: kinematic bicycle model with static obstacle avoidance vehicle_8state_obs_N_116000.tar.lz: dynamic single-track vehicle model with static obstacle avoidance models.zip: pretrained models for the experiments in this work For each dataset, files contain initial conditions, MPC input trajectories, predicted state sequences, and associated parameters as documented in the repository README.

  PublisherDOI

• Adaptive Control in Autonomous Driving via Real-Time Recurrent RL

  Open MIND · 2026-02-02

  preprintSenior author

  We study online fine-tuning of pretrained control policies for autonomous driving using Real-Time Recurrent Reinforcement Learning (RTRRL), a memory-efficient algorithm that updates policy parameters at every time step without backpropagation through time. We extend RTRRL to support LrcSSM, a recently proposed nonlinear diagonal state-space model, and combine offline behavioral cloning with online RTRRL fine-tuning to adapt policies to distribution shifts at deployment. We validate the approach in the CarRacing simulation and on a 1:10-scale RoboRacer platform equipped with an event camera, where a pretrained policy is fine-tuned online during real-world line-following. To our knowledge, this is the first demonstration of online RL fine-tuning with event-camera observations on standard (non-spiking) hardware in closed-loop control. LrcSSM-based policies improve fastest and most consistently across both settings.

  DOI

• Liquid Resistance Liquid Capacitance Networks

  Lecture notes in computer science · 2026-01-01

  book-chapterSenior author
  PublisherDOI

• Synaptic Activation and Dual Liquid Dynamics for Interpretable Bio-Inspired Models

  Open MIND · 2026-02-13

  preprintSenior author

  In this paper, we present a unified framework for various bio-inspired models to better understand their structural and functional differences. We show that liquid-capacitance-extended models lead to interpretable behavior even in dense, all-to-all recurrent neural network (RNN) policies. We further demonstrate that incorporating chemical synapses improves interpretability and that combining chemical synapses with synaptic activation yields the most accurate and interpretable RNN models. To assess the accuracy and interpretability of these RNN policies, we consider the challenging lane-keeping control task and evaluate performance across multiple metrics, including turn-weighted validation loss, neural activity during driving, absolute correlation between neural activity and road trajectory, saliency maps of the networks' attention, and the robustness of their saliency maps measured by the structural similarity index.

  DOI

• Towards Safe Learning-Based Non-Linear Model Predictive Control through Recurrent Neural Network Modeling

  arXiv (Cornell University) · 2026-03-25

  preprintOpen access

  The practical deployment of nonlinear model predictive control (NMPC) is often limited by online computation: solving a nonlinear program at high control rates can be expensive on embedded hardware, especially when models are complex or horizons are long. Learning-based NMPC approximations shift this computation offline but typically demand large expert datasets and costly training. We propose Sequential-AMPC, a sequential neural policy that generates MPC candidate control sequences by sharing parameters across the prediction horizon. For deployment, we wrap the policy in a safety-augmented online evaluation and fallback mechanism, yielding Safe Sequential-AMPC. Compared to a naive feedforward policy baseline across several benchmarks, Sequential-AMPC requires substantially fewer expert MPC rollouts and yields candidate sequences with higher feasibility rates and improved closed-loop safety. On high-dimensional systems, it also exhibits better learning dynamics and performance in fewer epochs while maintaining stable validation improvement where the feedforward baseline can stagnate.

  PublisherDOI

• Looped SSMs: Depth-Recurrence and Input Reshaping for Time Series Classification

  arXiv (Cornell University) · 2026-05-15

  preprintOpen accessSenior author

  State Space Models (SSMs) are inherently recurrent along the sequence dimension, yet depth-recurrence - reusing the same block repeatedly across layers, as recently applied in looped transformers - has not been explored in this model family. We show that a looped SSM with $k$ parameters iterated $L$ times consistently closely matches or outperforms a standard SSM with $k \cdot L$ independent parameters across four architectures (LRU, S5, LinOSS, LrcSSM) and six time series classification benchmarks, despite operating within a strictly smaller hypothesis space, as we formally establish. Since the larger model contains the looped model as a special case, this dominance cannot be explained by expressivity and instead points to parameter sharing across depth as a beneficial inductive bias that simplifies optimization. These results demonstrate that depth-recurrence is orthogonal to sequence-recurrence and independently beneficial. We further show that input reshaping is an equally neglected design axis: concatenating timesteps for low-dimensional inputs, or flattening and rechunking the joint feature-time dimension for high-dimensional ones, yields accuracy gains of 1-6% across all models, confirmed over 5 random seeds. Both techniques provide standalone improvements that compound when combined, suggesting that depth and input reshaping are two independent and underexplored design axes for SSMs on time series.

  PublisherDOI

Recent grants

• CAREER: Model-Based Design and Verification of Embedded Systems

  NSF · $300k · 2002–2010

Frequent coauthors

• Scott A. Smolka

  133 shared

• Guodong Wang

  72 shared

• Ezio Bartocci

  68 shared

• Ching‐Nung Yang

  National Dong Hwa University

  64 shared

• Bo Chen

  Technical University of Denmark

  64 shared

• Weizhi Meng

  Lancaster University

  64 shared

• Xiaodong Liu

  64 shared

• Naixue Xiong

  Sul Ross State University

  64 shared

Awards & honors

• National Science Foundation Career Award
• State University of New York Research Foundation Promising I…
• ACM Service Award