Author name cluster

Can Li

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

4 papers

2 author rows

NeurIPS Conference 2025 Conference Paper

Enforcing Hard Linear Constraints in Deep Learning Models with Decision Rules

Gonzalo E. Constante
Hao Chen
Can Li

Deep learning models are increasingly deployed in safety-critical tasks where predictions must satisfy hard constraints, such as physical laws, fairness requirements, or safety limits. However, standard architectures lack built-in mechanisms to enforce such constraints, and existing approaches based on regularization or projection are often limited to simple constraints, computationally expensive, or lack feasibility guarantees. This paper proposes a model-agnostic framework for enforcing input-dependent linear equality and inequality constraints on neural network outputs. The architecture combines a task network trained for prediction accuracy with a safe network trained using decision rules from the stochastic and robust optimization literature to ensure feasibility across the entire input space. The final prediction is a convex combination of the two subnetworks, guaranteeing constraint satisfaction during both training and inference without iterative procedures or runtime optimization. We prove that the architecture is a universal approximator of constrained functions and derive computationally tractable formulations based on linear decision rules. Empirical results on benchmark regression tasks show that our method consistently satisfies constraints while maintaining competitive accuracy and low inference latency.

PDF Details

IROS Conference 2025 Conference Paper

Learning-based Keypoints Detection with Topological Order on Deformable Linear Objects from Incomplete Point Clouds

Can Li
Jingyang Liu
Lei Sun

Detection of deformable linear objects (DLOs) in three-dimensional space is essential for robotic manipulation of DLOs. However, their complex deformations and high degrees of freedom make perception highly susceptible to occlusions, noise, and data missing. To address these challenges, we propose a deep learning-based method that leverages the topological properties of DLOs to robustly detect keypoints from incomplete point clouds while preserving the topological order of keypoints. Our approach initializes a sequence of keypoints that adheres to the topological structure of DLOs. Then, these ordered keypoints are refined through bidirectional sequence learning. Simulation results demonstrate that our method generates accurate, uniform, and smooth keypoint sequences under varying levels of occlusion. Compared to existing baselines, our approach achieves superior performance. Real-world experiments further validate the generalization capability of our method in unseen and challenging scenarios involving occlusion and self-occlusion while maintaining real-time performance.

Details

EAAI Journal 2024 Journal Article

Integration of the wind and solar power for the dynamic economic emission dispatch with the charging and discharging of plug-in electric vehicles

Dexuan Zou
Lejie Ma
Can Li
Haibin Ouyang

Wind and solar power are incorporated into the dynamic economic emission dispatch with the plug-in electric vehicles (DEED-PEV) in this paper, seeking to reduce the generation cost and pollutant emission significantly. A Laplacian non-dominated sorting genetic algorithm-II (LNSGA-II) and a constraint handling method (CHM) are proposed for the DEED-PEV. The LNSGA-II applies the Laplace distribution to both the crossover and mutation. The Laplace-based crossover makes each individual carry out unidirectional search towards its partner and enhances the information exchange between each pair of individuals. The Laplace-based mutation makes each individual carry out bi-directional search surrounding itself and adjust the searching range adaptively. Therefore, the LNSGA-II is able to explore and exploit the decision space of the DEED-PEV sufficiently, contributing to the reductions of the generation cost and pollutant emission in the objective space. For the variables of every individual, the CHM limits them within the feasible operating zones. The CHM also limits them within the dynamic boundaries. It further adjusts them according to the power balances. The power generation constraints and up/down ramp rate constraints can be always satisfied while the elimination of the total violation of the other constraints can be expedited. Therefore, the CHM can transform infeasible individuals to feasible ones rapidly. Four strategies are investigated for six PEV charging/discharging scenarios. The first three strategies consider no more than one kind of renewable energy generation while the fourth strategy involves both the wind power and solar power. Experimental results suggest that the fourth strategy is more economical and environment-friendly than the other three strategies. Furthermore, the LNSGA-II outperforms the other three competitor algorithms for six DEED-PEV problems in the light of hypervolume, coverage rate and spacing.

Details DOI

NeurIPS Conference 2020 Conference Paper

Adaptive Graph Convolutional Recurrent Network for Traffic Forecasting

Lei Bai
Lina Yao
Can Li
Xianzhi Wang
Can Wang

Modeling complex spatial and temporal correlations in the correlated time series data is indispensable for understanding the traffic dynamics and predicting the future status of an evolving traffic system. Recent works focus on designing complicated graph neural network architectures to capture shared patterns with the help of pre-defined graphs. In this paper, we argue that learning node-specific patterns is essential for traffic forecasting while pre-defined graph is avoidable. To this end, we propose two adaptive modules for enhancing Graph Convolutional Network (GCN) with new capabilities: 1) a Node Adaptive Parameter Learning (NAPL) module to capture node-specific patterns; 2) a Data Adaptive Graph Generation (DAGG) module to infer the inter-dependencies among different traffic series automatically. We further propose an Adaptive Graph Convolutional Recurrent Network (AGCRN) to capture fine-grained spatial and temporal correlations in traffic series automatically based on the two modules and recurrent networks. Our experiments on two real-world traffic datasets show AGCRN outperforms state-of-the-art by a significant margin without pre-defined graphs about spatial connections.

PDF Details