Four teams that include faculty and students from the Whiting School of Engineering’s Department of Civil and Systems Engineering will present their research at this year’s annual NeurIPS conference, December 2 – 7 in San Diego.
Now in its 39th year, NeurIPS is recognized as the flagship event in machine learning and artificial intelligence, bringing together the latest work in fields like AI, statistics, optimization, and neuroscience.
NeurIPS received 21,575 valid submissions in its main program track and accepted just 5,290 papers—an acceptance rate of 24.52%—with a breakdown of 4,525 as posters, 688 as spotlights and 77 as oral.
Accepted papers from civil and systems engineering faculty include:
- Personalized Decision Modeling: Utility Optimization or Textualized-Symbolic Reasoning (selected as a Conference Spotlight)
Yibo Zhao, Yang Zhao, Hongru Du, Hao Frank Yang
Human decisions consist of numbers, narratives, and nuance. Population-level models optimize using utility for the collective yet always miss the individual’s causal calculus shaped by identity, timing, and constraints. Inspired by Dr. Judea Pearl’s personalized decision-making lens (Mueller & Pearl, 2022), the team designed ATHENA—short for Adaptive Textual-symbolic Human-centric Reasoning—based on LLM technology, which learns a shared symbolic utility backbone and then semantically adapts to each person’s beliefs and context in text.
- Towards Physics-informed Spatial Intelligence with Human Priors: An Autonomous Driving Pilot Study (selected as a Conference Spotlight)
Guanlin Wu, Boyan Su, Yang Zhao, Pu Wang, Yichen Lin, Hao Frank Yang
Integrating and verifying spatial intelligence in foundation models remains a challenge. In response, this study introduces Spatial Intelligence Grid (SIG): a structured, grid-based schema that explicitly encodes object layouts, inter-object relations, and physically grounded priors. As a complementary channel to text, SIG provides a faithful, compositional representation of scene structure for foundation-model reasoning. Building on SIG, we derive SIG-informed evaluation metrics that quantify a model’s intrinsic VSI, which separates spatial capability from language priors. In few-shot in-context learning with state-of-the-art multimodal LLMs, SIG yields consistently larger, more stable, and more comprehensive gains across all VSI metrics compared to VQA-only representations, indicating its promise as a data-labeling and training schema for learning VSI. We also release SIGBench, a benchmark of 1.4K driving frames annotated with ground-truth SIG labels and human gaze traces, supporting both grid-based machine VSI tasks and attention-driven, human-like VSI tasks in autonomous-driving scenarios.
- How to Auto-optimize Prompts for Domain Tasks? Adaptive Prompting and Reasoning through Evolutionary Domain Knowledge Adaptation
Yang Zhao, Pu Wang, Hao Frank Yang
The team sought to understand how to write and optimize prompts and the reasoning process for domain-specific tasks, such as transportation, public health, and human behavior, so that an LLM’s reasoning ability is fully leveraged. The researchers propose Evolutionary Graph Optimization for Prompting (EGO-Prompt), an automated framework that designs better prompts, uses efficient reasoning processes and provides enhanced causal-informed process. Testing the framework on three real-world public health, transportation and human behavior tasks, EGO-Prompt achieves 7.32%-12.61% higher F1 than leading methods and allows small models to reach the performance of larger models at under 20% of the original cost. It also outputs a refined, domain-specific SCG that improves interpretability.
- Learning to Optimize for Mixed-Integer Non-linear Programming with Feasibility Guarantees (Workshop paper for ScaleOPT: GPU-Accelerated and Scalable Optimization)
Bo Tang, Elias B. Khalil, Ján Drgoňa
Mixed-integer nonlinear programs (MINLPs) arise in domains as diverse as energy systems and transportation, but are notoriously difficult to solve, particularly at scale. While learning-to-optimize (L2O) methods have been successful at continuous optimization, extending them to MINLPs is challenging due to integer constraints. To overcome this, we propose a novel L2O approach with two integer correction layers to ensure the integrality of the solution and a projection step to ensure the feasibility of the solution. We prove that the projection step converges, providing a theoretical guarantee for our method. Our experiments show that our methods efficiently solve MINLPs with up to tens of thousands of variables, providing high-quality solutions within milliseconds, even for problems where traditional solvers and heuristics fail. This is the first general L2O method for parametric MINLPs, finding solutions to some of the largest instances reported to date. - From Image Generation to Infrastructure Design: a Multi-agent Pipeline for Street Design Generation (Workshop paper for UrbanAI: Harnessing Artificial Intelligence for Smart Cities)
Chenguang Wang, Xiang Yan, Yilong Dai, Ziyi Wang, Susu Xu
Realistic visual renderings of street-design scenarios are essential for public engagement in active transportation planning. Traditional approaches are labor-intensive, hindering collective deliberation and collaborative decision-making. While AI-assisted generative design shows transformative potential by enabling rapid creation of design scenarios, existing generative approaches typically require large amounts of domain-specific training data and struggle to enable precise spatial variations of design/configuration in complex street-view scenes. We introduce a multi-agent system that edits and redesigns bicycle facilities directly on real-world street-view imagery. The framework integrates lane localization, prompt optimization, design generation, and automated evaluation to synthesize realistic, contextually appropriate designs. Experiments across diverse urban scenarios demonstrate that the system can adapt to varying road geometries and environmental conditions, consistently yielding visually coherent and instruction-compliant results. This work establishes a foundation for applying multi-agent pipelines to transportation infrastructure planning and facility design.