Novel Online Dimensionality Reduction Method with Improved Topology Representing and Radial Basis Function Networks

Area of Science:

• Machine Learning
• Data Science
• Network Science

Background:

• Conventional Topology Representing Networks (TRN) have limitations in accurately representing complex data topology.
• High-dimensional data often contains embedded low-dimensional structures that are challenging to extract.

Purpose of the Study:

• To enhance the conventional Topology Representing Network (TRN) for improved topology relationship modeling.
• To propose a novel online dimensionality reduction method integrating an enhanced TRN with Radial Basis Function Networks (RBFN).

Main Methods:

• Developed an improved Topology Representing Network (TRN) to establish more appropriate topology relationships.
• Integrated the enhanced TRN with Radial Basis Function Network (RBFN) for online dimensionality reduction.
• The method is designed to process nonlinear embedded manifolds and map new data in real-time.

Main Results:

• The proposed method effectively identifies meaningful low-dimensional feature structures within high-dimensional data.
• Demonstrated the capability to handle nonlinear embedded manifolds and perform online data mapping.
• The enhanced TRN facilitates efficient processing of large datasets.

Conclusions:

• The novel online dimensionality reduction method offers significant improvements over conventional approaches.
• The integration of enhanced TRN and RBFN provides a robust solution for complex, high-dimensional datasets.
• Experimental results validate the effectiveness and efficiency of the proposed technique.