Abstract:Clustering, as an unsupervised technique, plays a pivotal role in various data analysis applications. Among clustering algorithms, Spectral Clustering on Euclidean Spaces has been extensively studied. However, with the rapid evolution of data complexity, Euclidean Space is proving to be inefficient for representing and learning algorithms. Although Deep Neural Networks on hyperbolic spaces have gained recent traction, clustering algorithms or non-deep machine learning models on non-Euclidean Spaces remain underexplored. In this paper, we propose a spectral clustering algorithm on Hyperbolic Spaces to address this gap. Hyperbolic Spaces offer advantages in representing complex data structures like hierarchical and tree-like structures, which cannot be embedded efficiently in Euclidean Spaces. Our proposed algorithm replaces the Euclidean Similarity Matrix with an appropriate Hyperbolic Similarity Matrix, demonstrating improved efficiency compared to clustering in Euclidean Spaces. Our contributions include the development of the spectral clustering algorithm on Hyperbolic Spaces and the proof of its weak consistency. We show that our algorithm converges at least as fast as Spectral Clustering on Euclidean Spaces. To illustrate the efficacy of our approach, we present experimental results on the Wisconsin Breast Cancer Dataset, highlighting the superior performance of Hyperbolic Spectral Clustering over its Euclidean counterpart. This work opens up avenues for utilizing non-Euclidean Spaces in clustering algorithms, offering new perspectives for handling complex data structures and improving clustering efficiency.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to solve the problem that spectral clustering algorithms in traditional Euclidean spaces perform poorly when dealing with complex data structures (such as hierarchical and tree - like structures). Specifically, the paper proposes a hyperbolic - space - based spectral clustering algorithm (Hyperbolic Spectral Clustering Algorithm, HSCA) to make up for the deficiencies of existing non - deep - learning machine - learning models in non - Euclidean spaces. #### Main problems: 1. **Limitations of Euclidean space**: - Euclidean space is inefficient in representing and learning complex data structures (such as hierarchical and tree - like structures). - For data sets with complex topological structures, Euclidean space cannot effectively capture their intrinsic patterns, thus affecting the performance of learning algorithms. 2. **Gaps in existing research**: - Although there has been some progress in the application of deep neural networks in hyperbolic spaces, the exploration of non - deep - learning machine - learning models or clustering algorithms in non - Euclidean spaces is still relatively scarce. #### Proposed solutions: - **Hyperbolic similarity matrix**: Replace the traditional Euclidean similarity matrix with an appropriate hyperbolic similarity matrix to better represent complex data structures. - **Theoretical analysis and proof**: Provide a weak - consistency theoretical analysis of the proposed algorithm and prove that its convergence speed is at least the same as that of the spectral clustering algorithm in Euclidean space. - **Experimental verification**: Through the experimental results on the Wisconsin Breast Cancer Dataset, show the superior performance of the hyperbolic spectral clustering algorithm compared to its Euclidean version. ### Specific contributions: 1. **Propose hyperbolic spectral clustering algorithm**: Develop a spectral clustering algorithm based on hyperbolic space and replace the Euclidean similarity matrix with a hyperbolic similarity matrix. 2. **Theoretical guarantee**: Provide a proof of weak - consistency of the algorithm to ensure that its convergence speed is not lower than that of the Euclidean spectral clustering algorithm. 3. **Extended application**: Propose hyperbolic versions of some well - known Euclidean spectral clustering algorithm variants in hyperbolic space, such as FastESC (Fast Spectral Clustering) and Approximate Spectral Clustering with k - means - based Landmark Selection. ### Conclusion: This paper opens up new avenues for the application of non - deep - learning machine - learning models in non - Euclidean spaces, especially showing significant advantages in dealing with complex data structures and improving clustering efficiency.

Consistent Spectral Clustering in Hyperbolic Spaces

Parallel spectral clustering algorithm

Spectral Clustering on Large Datasets: When Does it Work? Theory from Continuous Clustering and Density Cheeger-Buser

The Hidden Convexity of Spectral Clustering

A Unified Framework for Representation-Based Subspace Clustering of Out-of-Sample and Large-Scale Data.

Fast Spectral Clustering Learning with Hierarchical Bipartite Graph for Large-Scale Data

Spectral clustering algorithm for the allometric extension model

Spectral Clustering Via Half-Quadratic Optimization

A High Performance Implementation of Spectral Clustering on CPU-GPU Platforms

HyperSpec: Ultrafast Mass Spectra Clustering in Hyperdimensional Space

Spectral Clustering in Convex and Constrained Settings

Subspace Clustering Via Variance Regularized Ridge Regression

Context-Aware Hypergraph Construction for Robust Spectral Clustering

The geometry of kernelized spectral clustering

Federated Classification in Hyperbolic Spaces via Secure Aggregation of Convex Hulls

Euclidean and Poincaré space ensemble Xgboost

Hypergraphs with Edge-Dependent Vertex Weights: Spectral Clustering based on the 1-Laplacian

Robust Clustering using Hyperdimensional Computing

Spectral clustering and semi-supervised learning using evolving similarity graphs

Hypergraphs with Edge-Dependent Vertex Weights: p-Laplacians and Spectral Clustering

Research on Spectral Clustering Algorithms and Prospects