GB-RVFL: Fusion of Randomized Neural Network and Granular Ball Computing
This work addresses scalability and robustness problems in neural network classification for researchers and practitioners, but it is incremental as it builds on existing RVFL and granular computing methods.
The authors tackled the issues of scalability and noise sensitivity in random vector functional link (RVFL) networks by proposing GB-RVFL, which uses granular balls as inputs to reduce computational complexity and improve robustness, and GE-GB-RVFL, which incorporates graph embedding to preserve geometric structure, achieving superior performance on multiple datasets.
The random vector functional link (RVFL) network is a prominent classification model with strong generalization ability. However, RVFL treats all samples uniformly, ignoring whether they are pure or noisy, and its scalability is limited due to the need for inverting the entire training matrix. To address these issues, we propose granular ball RVFL (GB-RVFL) model, which uses granular balls (GBs) as inputs instead of training samples. This approach enhances scalability by requiring only the inverse of the GB center matrix and improves robustness against noise and outliers through the coarse granularity of GBs. Furthermore, RVFL overlooks the dataset's geometric structure. To address this, we propose graph embedding GB-RVFL (GE-GB-RVFL) model, which fuses granular computing and graph embedding (GE) to preserve the topological structure of GBs. The proposed GB-RVFL and GE-GB-RVFL models are evaluated on KEEL, UCI, NDC and biomedical datasets, demonstrating superior performance compared to baseline models.