Flow Matching for Denoised Social Recommendation

Authors: Yinxuan Huang, Ke Liang, Zhuofan Dong, Xiaodong Qu, Wang Tianxiang, Yue Han, Jingao Xu, Bin Zhou, Ye Wang

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

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we present a series of experiments conducted to evaluate the performance of our Rec Flow method, focusing on the following five questions: Q1: How does Rec Flow perform in comparison to other state-of-the-art social recommendation methods? Q2: What are the key contributions of Rec Flow s main modules? Q3: Is Rec Flow robust enough to effectively handle noisy and sparse data in social recommendation (SR)? Q4: How do different settings impact the performance of Rec Flow? Q5: How does the complexity of our method compare to that of alternative approaches?
Researcher Affiliation Academia 1School of Computer Science, National University of Defense Technology, Changsha, Hunan, China 2University of Chicago, Chicago, Illinois, United States 3Harbin Institute of Technology(Shenzhen), Shenzhen, Guangdong, China.
Pseudocode Yes Algorithm 1 Rec Flow Training Input: Users social interaction embedding Es Output: The reconstructed embedding which is denoted as ˆe0 Algorithm 2 Rec Flow Inference Input: Users interaction vectors xu, u = 1, 2, . . . , |U|; optimized parameter θ Output: Predicted user embeddings or interaction outcomes
Open Source Code No Code are available at .
Open Datasets Yes We conducted experiments on three publicly available social recommendation datasets: Ciao, Yelp and Epinions. Detailed statistics for these datasets are provided in Table 2. Our datasets are obtained from Rec Diff (Li et al., 2024a)1 Datasets are available at https://github.com/HKUDS/Rec Diff.
Dataset Splits Yes We applied a 7:1:2 ratio to split each dataset into training, validation, and test sets, adhering to common data partitioning practices in graph-based recommendation systems.
Hardware Specification Yes All experiments are conducted on a machine with an RTX A800 for a fair comparison.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers. It mentions concepts like GELU activation but not a software library and its version.
Experiment Setup Yes The learning rate was tuned within [5e 4, 1e 3, 5e 3] with a 0.96 decay factor per epoch. Batch sizes were selected from [1024, 2048, 4096, 8192], and hidden dimensions from [64, 128, 256, 512]. The parameter γ was set according to the γpctpercentile of node embedding distances for each dataset. The optimal number of GNN layers was chosen from [1, 2, 3, 4]. The Timestep embedding size is selected from 4,8,16,32. And the batch size for Ciao is 2048, while for Yelp and Epinions is 4096. Regularization weights λ1 were selected from [1e 3, 1e 2, 1e 1, 1e0, 1e1].