Neural Network Implementation of the Renormalization Group for Fault Diagnosis with Class Imbalance

The application of machine learning models in practical tasks faces challenges such as class imbalance and multidimensional noise. This study proposes RGNet, a neural network architecture based on the concept of the renormalization group (RG), for hierarchical coarse-graining of the feature space. The model sequentially compresses the input dimensionality and concatenates all scales before classification, allowing it to capture both local details and global patterns. The notion of RG-flows is introduced --- interpretable low-dimensional representations whose visualization via t-SNE reveals a discrete curvilinear structure confirming the effectiveness of coarse-graining. On the AI4I2020 dataset, RGNet achieves a recall of 0.9118 at threshold 0.5 and a balanced variant with F1=0.630 at threshold 0.92, reflecting a deliberate trade-off: the architecture prioritizes fault detection over precision, consistent with the asymmetric cost of missed failures in predictive maintenance. The obtained results demonstrate that RGNet is an interpretable and competitive solution for fault prediction in applications with imbalanced classes.

@article{Nikulchev2026,
  title     = {Neural Network Implementation of the Renormalization Group for Fault Diagnosis with Class Imbalance},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {17},
  number    = {7},
  year      = {2026},
  publisher = {The Science and Information Organization},
  author    = {Evgeny Nikulchev and Dmitry Ilin},
  doi       = {10.14569/IJACSA.2026.0170705},
  url       = {https://doi.org/10.14569/IJACSA.2026.0170705}
}