FairTP: A Prolonged Fairness Framework for Traffic Prediction
Authors: Jiangnan Xia, Yu Yang, Jiaxing Shen, Senzhang Wang, Jiannong Cao
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments in two real-world datasets show that Fair TP significantly improves prediction fairness without causing significant accuracy degradation. |
| Researcher Affiliation | Academia | 1School of Computer Science and Engineering, Central South University 2Centre for Learning, Teaching, and Technology, The Education University of Hong Kong 3School of Data Science, Lingnan University 4The Department of Computing, The Hong Kong Polytechnic University |
| Pseudocode | No | The paper describes methods and processes using prose and mathematical equations but does not include a clearly labeled pseudocode or algorithm block. |
| Open Source Code | Yes | Code https://github.com/jiangnanx129/Fair TP Extended version https://github.com/jiangnanx129/Fair TP |
| Open Datasets | Yes | Dataset. We use two real-world datasets for regional traffic prediction: the HK and the SD datasets. The HK dataset contains six months of taxi trajectory data with 938 road sensors. The SD dataset includes data from 716 road sensors, sourced from the Pe MS platform in 2019. Details are provided in the extended version. |
| Dataset Splits | No | The paper mentions using two datasets (HK and SD) but does not provide specific details on how these datasets were split into training, validation, and test sets (e.g., percentages, sample counts, or references to predefined splits). |
| Hardware Specification | Yes | All models are implemented on the Ge Force RTX 3090. |
| Software Dependencies | No | The paper mentions that models are implemented but does not provide specific version numbers for software dependencies or libraries (e.g., Python, PyTorch, TensorFlow, CUDA). |
| Experiment Setup | Yes | We set the sampled number Nsam to 200 for both the SD and HK datasets. The dynamic time length Td is fixed at 3, representing 3 batches. We set the hyperparameters λ1 and λ2 to 0.01 and 0.1, respectively. |