skip to main content

Search for: All records

Creators/Authors contains: "Pan, Menghai"

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Free, publicly-accessible full text available May 13, 2025
  2. Free, publicly-accessible full text available December 15, 2024
  3. Free, publicly-accessible full text available August 4, 2024
  4. Free, publicly-accessible full text available August 4, 2024
  5. To make daily decisions, human agents devise their own "strategies" governing their mobility dynamics (e.g., taxi drivers have preferred working regions and times, and urban commuters have preferred routes and transit modes). Recent research such as generative adversarial imitation learning (GAIL) demonstrates successes in learning human decision-making strategies from their behavior data using deep neural networks (DNNs), which can accurately mimic how humans behave in various scenarios, e.g., playing video games, etc. However, such DNN-based models are "black box" models in nature, making it hard to explain what knowledge the models have learned from human, and how the models make such decisions, which was not addressed in the literature of imitation learning. This paper addresses this research gap by proposing xGAIL, the first explainable generative adversarial imitation learning framework. The proposed xGAIL framework consists of two novel components, including Spatial Activation Maximization (SpatialAM) and Spatial Randomized Input Sampling Explanation (SpatialRISE), to extract both global and local knowledge from a well-trained GAIL model that explains how a human agent makes decisions. Especially, we take taxi drivers' passenger-seeking strategy as an example to validate the effectiveness of the proposed xGAIL framework. Our analysis on a large-scale real-world taxi trajectory data shows promising results from two aspects: i) global explainable knowledge of what nearby traffic condition impels a taxi driver to choose a particular direction to find the next passenger, and ii) local explainable knowledge of what key (sometimes hidden) factors a taxi driver considers when making a particular decision. 
    more » « less
  6. Given the historical movement trajectories of a set of individual human agents (e.g., pedestrians, taxi drivers) and a set of new trajectories claimed to be generated by a specific agent, the Human Mobility Signature Identification (HuMID) problem aims at validating if the incoming trajectories were indeed generated by the claimed agent. This problem is important in many real-world applications such as driver verification in ride-sharing services, risk analysis for auto insurance companies, and criminal identification. Prior work on identifying human mobility behaviors requires additional data from other sources besides the trajectories, e.g., sensor readings in the vehicle for driving behavior identification. However, these data might not be universally available and is costly to obtain. To deal with this challenge, in this work, we make the first attempt to match identities of human agents only from the observed location trajectory data by proposing a novel and efficient framework named Spatio-temporal Siamese Networks (ST-SiameseNet). For each human agent, we extract a set of profile and online features from his/her trajectories. We train ST-SiameseNet to predict the mobility signature similarity between each pair of agents, where each agent is represented by his/her trajectories and the extracted features. Experimental results on a real-world taxi trajectory dataset show that our proposed ST-SiamesNet can achieve an F_1 score of 0.8508, which significantly outperforms the state-of-the-art techniques. 
    more » « less
  7. null (Ed.)
    Learning to make optimal decisions is a common yet complicated task. While computer agents can learn to make decisions by running reinforcement learning (RL), it remains unclear how human beings learn. In this paper, we perform the first data-driven case study on taxi drivers to validate whether humans mimic RL to learn. We categorize drivers into three groups based on their performance trends and analyze the correlations between human drivers and agents trained using RL. We discover that drivers that become more efficient at earning over time exhibit similar learning patterns to those of agents, whereas drivers that become less efficient tend to do the opposite. Our study (1) provides evidence that some human drivers do adapt RL when learning, (2) enhances the deep understanding of taxi drivers' learning strategies, (3) offers a guideline for taxi drivers to improve their earnings, and (4) develops a generic analytical framework to study and validate human learning strategies. 
    more » « less
  8. null (Ed.)
    In this paper we advocate a forward-looking, ambitious and disruptive smart cloud commuting system (SCCS) for future smart cities based on shared AVs. Employing giant pools of AVs of varying sizes, SCCS seeks to supplant and integrate various modes of transport in a unified, on-demand fashion, and provides passengers with a fast, convenient, and low cost transport service for their daily commuting needs. To explore feasibility and efficiency gains of the proposed SCCS, we model SCCS as a queueing system with passengers' trip demands (as jobs) being served by the AVs (as servers). Using a 1-year real trip dataset from Shenzhen China, we quantify (i) how design choices, such as the numbers of depots and AVs, affect the passenger waiting time and vehicle utilization; and (ii) how much efficiency gains (i.e., reducing the number of service vehicles, and improving the vehicle utilization) can be obtained by SCCS comparing to the current taxi system. Our results demonstrate that the proposed SCCS system can serve the trip demands with 22% fewer vehicles and 37% more vehicle utilization, which shed lights on the design feasibility of future smart transportation system. 
    more » « less