Xiyao Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Xiyao Wang has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Andrew I. Cohen
Department of Philosophy and Jean Beer Blumenfeld Center for Ethics, Georgia State University
1 shared publication - Furong Huang
Department of Optoelectronic Engineering Jinan University Guangzhou Guangdong China
1 shared publication - Huilin Jiang
Nanjing Xiaozhuang University, Nanjing, China
1 shared publication - Jintao Wan
School of Transportation, Jilin University, Changchun, China
1 shared publication - Jiong Fu
Shanghai Jiao Tong University
1 shared publication - Penglong Li
Jilin University
1 shared publication
Latest Publications
- 18 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (1) , 1164-1174
An Innovative Intention-Driven Forecast Predictor and Discriminative Normalization for Multiagent Trajectory Prediction
In complex traffic scenarios, accurate and efficient trajectory prediction is crucial for downstream planning and decision-making. However, existing trajectory prediction methods are computationally expensive, and the extraction of interaction features often leads to prediction distortion due to over-smoothing.To address this, we propose a lightweight Relative Spatial Instance Framework, which replaces the traditional agent-centric approach with a local coordinate system, thereby reducing computational burden and improving prediction efficiency.To further enhance model robustness, we designed a Discriminative Normalization Layer (DNL) that optimizes the feature matrix to improve feature discrimination, effectively alleviating over-smoothing.In addition, our Intention-Driven Forecast Predictor (IDFP) captures the initial state of each agent and accurately predicts its future trajectory.Experimental results show that our method achieves state-of-the-art prediction performance on the public dataset Argoverse 1 with fewer parameter costs.Overall, this method balances efficiency and accuracy, providing an efficient and reliable solution for the field of trajectory prediction.
- 18 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (1) , 1164-1174
An Innovative Intention-Driven Forecast Predictor and Discriminative Normalization for Multiagent Trajectory Prediction
In complex traffic scenarios, accurate and efficient trajectory prediction is crucial for downstream planning and decision-making. However, existing trajectory prediction methods are computationally expensive, and the extraction of interaction features often leads to prediction distortion due to over-smoothing.To address this, we propose a lightweight Relative Spatial Instance Framework, which replaces the traditional agent-centric approach with a local coordinate system, thereby reducing computational burden and improving prediction efficiency.To further enhance model robustness, we designed a Discriminative Normalization Layer (DNL) that optimizes the feature matrix to improve feature discrimination, effectively alleviating over-smoothing.In addition, our Intention-Driven Forecast Predictor (IDFP) captures the initial state of each agent and accurately predicts its future trajectory.Experimental results show that our method achieves state-of-the-art prediction performance on the public dataset Argoverse 1 with fewer parameter costs.Overall, this method balances efficiency and accuracy, providing an efficient and reliable solution for the field of trajectory prediction.
- Open Access
- 14 May 2023
- Published by Darcy & Roy Press Co. Ltd. in Frontiers in Business, Economics and Management
- Vol. 9 (1) , 238-243
Take the Cities in The Yellow River Basin as An Example
Under the two-carbon target, the road of green development in China has reached a new height. This paper is based on the SBM-GML model of 2001-2019 the Yellow River basin nine provinces of agricultural green total factor productivity, combined with the number of Beijing university Digital Inclusive Financial index, from the perspective of time series, space to explore the Digital Inclusive Financial and the relationship between agricultural green low carbon development, for the Yellow River city Green High Quality Development.
- 14 May 2023
- Published by Darcy & Roy Press Co. Ltd. in Frontiers in Business, Economics and Management
- Vol. 9 (1) , 238-243
Open AccessTake the Cities in The Yellow River Basin as An Example
Under the two-carbon target, the road of green development in China has reached a new height. This paper is based on the SBM-GML model of 2001-2019 the Yellow River basin nine provinces of agricultural green total factor productivity, combined with the number of Beijing university Digital Inclusive Financial index, from the perspective of time series, space to explore the Digital Inclusive Financial and the relationship between agricultural green low carbon development, for the Yellow River city Green High Quality Development.
- Open Access
Transfer RL across Observation Feature Spaces via Model-Based Regularization
In many reinforcement learning (RL) applications, the observation space is specified by human developers and restricted by physical realizations, and may thus be subject to dramatic changes over time (e.g. increased number of observable features). However, when the observation space changes, the previous policy will likely fail due to the mismatch of input features, and another policy must be trained from scratch, which is inefficient in terms of computation and sample complexity. Following theoretical insights, we propose a novel algorithm which extracts the latent-space dynamics in the source task, and transfers the dynamics model to the target task to use as a model-based regularizer. Our algorithm works for drastic changes of observation space (e.g. from vector-based observation to image-based observation), without any inter-task mapping or any prior knowledge of the target task. Empirical results show that our algorithm significantly improves the efficiency and stability of learning in the target task.
Open AccessTransfer RL across Observation Feature Spaces via Model-Based Regularization
In many reinforcement learning (RL) applications, the observation space is specified by human developers and restricted by physical realizations, and may thus be subject to dramatic changes over time (e.g. increased number of observable features). However, when the observation space changes, the previous policy will likely fail due to the mismatch of input features, and another policy must be trained from scratch, which is inefficient in terms of computation and sample complexity. Following theoretical insights, we propose a novel algorithm which extracts the latent-space dynamics in the source task, and transfers the dynamics model to the target task to use as a model-based regularizer. Our algorithm works for drastic changes of observation space (e.g. from vector-based observation to image-based observation), without any inter-task mapping or any prior knowledge of the target task. Empirical results show that our algorithm significantly improves the efficiency and stability of learning in the target task.
- 20 September 2019
- Published by Springer Nature
- p. 78-84
Steering Wheel Interaction Design Based on Level 3 Autonomous Driving Scenario
- 20 September 2019
- Published by Springer Nature
- p. 78-84
Steering Wheel Interaction Design Based on Level 3 Autonomous Driving Scenario