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..
Structured Spectral Reasoning for Frequency-Adaptive Multimodal Recommendation
Authors: Wei Yang, Rui Zhong, Yiqun Chen, Chi Lu, Peng Jiang
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on three real-world benchmarks show consistent gains over strong baselines, particularly under sparse and cold-start settings. Additional analyses indicate that structured spectral modeling improves robustness and provides clearer diagnostics of how different bands contribute to performance. |
| Researcher Affiliation | Collaboration | 1Kuaishou Technology 2Renmin University of China |
| Pseudocode | No | The paper describes methods in text and mathematical formulas, but does not contain a dedicated section or figure explicitly labeled 'Pseudocode' or 'Algorithm'. |
| Open Source Code | Yes | The code is available at https://github.com/llm-ml/SSR.git. |
| Open Datasets | Yes | We conduct experiments on three representative subsets of the Amazon Product Review corpus3, widely adopted in multimodal recommendation literature [15, 51]. ... 3http://jmcauley.ucsd.edu/data/amazon/ |
| Dataset Splits | Yes | We build splits in chronological order with an 80/10/10 ratio for train, validation, and test. |
| Hardware Specification | No | The paper states in its NeurIPS checklist that it provides details on compute resources in Section 4 and Appendix B. However, upon reviewing these sections, no specific hardware details such as GPU models, CPU models, or detailed cloud/cluster specifications are provided. Only software frameworks and optimization strategies are mentioned. |
| Software Dependencies | No | The paper mentions implementing the method using PyTorch and optimizing with the Adam optimizer, and applying Xavier initialization. However, it does not provide specific version numbers for these software components or any other libraries used, which is required for reproducible software dependency information. |
| Experiment Setup | Yes | We implement our method using Py Torch within the MMRec benchmarking framework [91]. We use 64-dimensional embedding vectors for all item and user representations and apply Xavier initialization [92]. The model is optimized using the Adam optimizer [93] with a batch size of 2048. Learning rates are tuned from {0.0001, 0.0005, 0.001, 0.005} using the validation set. Early stopping is triggered if Recall@20 does not improve within 20 validation steps. We conduct extensive ablations on the number of frequency bands M {2, 3, 4, 5} and evaluate regularization weights for the spectral consistency loss, contrastive term and regularization term from {0.0001, 0.001, 0.01, 0.1, 1.0}. All hyperparameter choices are selected based on validation performance. |