Robotics & Embodied AI
Fuzzy topological system shows higher accuracy on eleven regression datasets
In a 2025 paper in IEEE Transactions on Fuzzy Systems, researchers at Liaoning Shihua University report that a method called a hierarchical fuzzy topological system, or HFTS, showed higher accuracy on eleven KEEL regression datasets. High-dimensional regression is a prediction task with many input features that can interact in nonlinear ways. The method combines fuzzy logic, which permits partial membership rather than strict all-or-nothing categories, with a graph neural network that groups similar features into a topological structure. The paper reports that this approach kept a balance between interpretability and predictive accuracy.
What problem does the method address?
High-dimensional data regression is hard because features can relate to each other in nonlinear ways, intermediate variables can become too numerous, and rules can expand too quickly, a problem the paper calls rule explosion. These conditions make it difficult for a model to capture complex features and keep regression accuracy high.
How does HFTS handle high-dimensional data?
HFTS has a modular design. Each layer is an independent fuzzy logic system, so the layers can work flexibly with the distribution of features and the required result. A graph neural network groups high-dimensional data by mapping features into nodes and setting edges based on similarity. This forms a topological structure that supports high-density feature representation through neighborhood aggregation. The system also uses a cross-layer rule-sharing mechanism and an interpolation expansion algorithm to smooth fuzzy rules and lower interaction complexity. An adaptive weight adjustment strategy changes feature importance to improve robustness and predictive accuracy.
What did the tests show?
When the researchers applied HFTS to eleven KEEL regression datasets, the method was more accurate on those datasets. The paper reports that it addressed high-dimensional interactions while keeping a balance between interpretability and performance.
Limits and what's next
The result is reported on eleven KEEL regression datasets, so the evidence is tied to those datasets. The abstract describes one method and one set of tests, and it does not report an independent replication by another team. It also does not say whether the accuracy result would hold on data outside those eleven datasets or in a real-world setting. The abstract reports no further limits beyond that dataset scope and single-study test.
Source: Mengxue Yao et al., “Hierarchical Fuzzy Topological System for High-Dimensional Data Regression Problems,” IEEE Transactions on Fuzzy Systems, 2025.