POI Recommendation via Multi-Objective Adversarial Imitation Learning

Authors: Zhenglin Wan, Anjun Gao, Xingrui Yu, Pingfu Chao, Jun Song, Maohao Ran

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Comprehensive experiments reveal the superior performance for MOAIR compared with baselines, especially with sparse training data.
Researcher Affiliation Collaboration 1School of Data Science, The Chinese University of Hong Kong-Shenzhen, Shenzhen, China 2School of Computer Science and Technology, Soochow University, Suzhou, China 3Centre for Frontier AI Research, Agency for Science, Technology and Research (A*STAR), Singapore 4Department of Geography, Hong Kong Baptist University, Hong Kong 5Metasequoia Intelligence, Shenzhen, China
Pseudocode No The paper describes the methodology using textual explanations and mathematical formulations, but does not include any clearly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statements about releasing source code, nor does it provide a link to a code repository or mention code in supplementary materials.
Open Datasets Yes We utilize check-in data from both Foursquare and Gowalla, as these datasets are commonly employed in prior studies.
Dataset Splits Yes Each dataset is organized by user, sorted chronologically, and split with the first 80% used for training and the remaining 20% for testing.
Hardware Specification Yes In our experiments, we uses four A40 48G GPUs, an AMD EPYC 7543P 32-core CPU, and a Linux operating system.
Software Dependencies No The paper mentions general software concepts but does not provide specific version numbers for any libraries or frameworks used in the implementation.
Experiment Setup Yes The embedding dimension for the state is set at 128. The dimension of latent variable z is set to 32, the learning rate and PPO-clip hyper-parameter of each PPO unit are set to be 3 10 4 and 0.1, and the decay rate γ of IL agent is 0.99. The masking probability is set to 0.15 at first and finally achieves 0.25 along with the epoch increasing.