RiskOracle: A Minute-Level Citywide Traffic Accident Forecasting Framework

Authors: Zhengyang Zhou, Yang Wang, Xike Xie, Lianliang Chen, Hengchang Liu1258-1265

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on two real-world datasets demonstrate the effectiveness and scalability of our Risk Oracle framework.
Researcher Affiliation Academia 1School of Computer Science and Technology, University of Science and Technology of China 2School of Software Engineering, University of Science and Technology of China 3University of Electronic Science and Technology of China {zzy0929, cll006}@mail.ustc.edu.cn, {angyan, xkxie, }@ustc.edu.cn, liu.heng.chang@gmail.com
Pseudocode No The paper does not include any pseudocode or clearly labeled algorithm blocks.
Open Source Code Yes More details are available on the website4. https://github.com/zzyy0929/AAAI2020-Risk Oracle/.
Open Datasets Yes We conduct experiments on two real-world datasets: NYC Opendata and Suzhou Industrial Park (SIP) dataset. ... More details are available on the website4. https://github.com/zzyy0929/AAAI2020-Risk Oracle/.
Dataset Splits Yes For experiments, we select 60%, 30% and 10% of dataset for training, evaluation and validation, respectively.
Hardware Specification No The paper does not provide specific hardware details such as CPU/GPU models or memory specifications used for experiments. It only lists funding sources.
Software Dependencies No The multi-task DTGN is trained with back propagation and Adam method (Kingma 2014). No specific version numbers for software libraries or dependencies are provided.
Experiment Setup Yes We stack 9 GCN layers with 384 filters in each layer. The weights of the loss function are set as λ1 = 0.8, λ2 = 1, λ3 = 1e 4. The multi-task DTGN is trained with back propagation and Adam method (Kingma 2014).