Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Hypergraph-Enhanced Contrastive Learning for Multi-View Clustering with Hyper-Laplacian Regularization
Authors: Zhibin Gu, weili wang
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on diverse datasets demonstrate the effectiveness of our approach. This section presents a comprehensive empirical evaluation of the proposed HOPER model, encompassing experimental settings, performance comparisons, parameter sensitivity analysis, feature visualizations, and ablation studies. |
| Researcher Affiliation | Academia | Zhibin Gu1,2,3 Weili Wang4 1College of Computer and Cyber Security, Hebei Normal University, China ... 4Independent Researcher, China EMAIL, EMAIL |
| Pseudocode | Yes | Algorithm 1 Hypergraph-enhanced Multi-view Representation Learning |
| Open Source Code | Yes | The code is provided in the appendix, and all datasets used are publicly available, ensuring reproducibility of the results. |
| Open Datasets | Yes | To comprehensively evaluate the effectiveness of the proposed HOPER framework, we conduct experiments on six publicly available multi-view datasets. Their statistical details are shown in Table 1. |
| Dataset Splits | No | The paper mentions several benchmark datasets (BBCSport, Synthetic3d, Web KB, COIL-20, Handwritten, Hdigit) and their statistics but does not specify explicit training, validation, or test splits. The clustering results are obtained by applying k-means to the unified representations without explicitly defined data splits for model training vs. evaluation. |
| Hardware Specification | Yes | All experiments are conducted using the PyTorch framework on an NVIDIA GeForce RTX 3090 GPU. |
| Software Dependencies | No | The paper mentions 'PyTorch framework' but does not specify a version number for PyTorch or any other software libraries used, which is required for a reproducible description of ancillary software. |
| Experiment Setup | Yes | In the initialization module, we utilize a four-layer autoencoder to obtain the latent embeddings. The number of nearest neighbors of hyperedge construction is tuned over {5, 10, 15, 20, 25, 30} on different datasets. In the fine-tuning module, we optimize the hyperparameters λ1 and λ2. Based on empirical observations, we perform a grid search over the values {0.0001, 0.001, 0.01, 0.1, 1, 5, 10} to select the optimal values for both hyperparameters on multiple datasets. The best performance is obtained under the combination of λ1 = 5 and λ2 = 0.0001. These specific values were then used for all experiments reported in this paper. The training process consists of 2000 epochs for the autoencoder initialization phase and 200 epochs for the fine-tuning phase. We employ a cosine learning rate decay to adjust the learning rate dynamically. |