Community-Aware Multi-Task Transportation Demand Prediction
Authors: Hao Liu, Qiyu Wu, Fuzhen Zhuang, Xinjiang Lu, Dejing Dou, Hui Xiong320-327
AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, extensive experiments on two real-world datasets demonstrate the effectiveness of our approach compared with seven baselines. |
| Researcher Affiliation | Collaboration | 1Baidu Research, Beijing, China 2Peking University, China 3Key Lab of IIP of Chinese Academy of Sciences (CAS), ICT, CAS, Beijing 100190, China 4University of Chinese Academy of Sciences, Beijing 100049, China 5Rutgers University, USA |
| Pseudocode | No | No pseudocode or algorithm blocks were found in the paper. |
| Open Source Code | No | The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available. |
| Open Datasets | No | We use two real-world datasets, Beijing and Shanghai. ... The paper uses internal datasets ('Beijing and Shanghai') without providing a public link, DOI, repository, or formal citation to access them. They are presented as data collected and used by the authors, not explicitly made available for public access. |
| Dataset Splits | Yes | We chronologically order each dataset, set one hour as the unit time step, use the first 60% as the training set, the next 20% as the validation set, and the last 20% for testing. |
| Hardware Specification | Yes | All models run on a Linux server with Intel Xeon 5117 CPU, 128 GB Memory, and NVIDIA Tesla P40 GPU. |
| Software Dependencies | No | The paper mentions using 'Adam' for optimization and refers to 'XGBoost (Chen and Guestrin 2016)', but it does not provide specific version numbers for these or any other software libraries, programming languages, or environments. |
| Experiment Setup | Yes | Specifically, we set the learning rate to 0.0001, hidden dimension d = 512 and α = 0.5. We set input length L = 18 and τ = 3 for prediction. The number of task groups is set to 4. The activation function in GNN is Leaky Re LU with slope ratio 0.1. ... we apply Z-score normalization. |