SpectralOptions<T>
Options & Configuration · AiDotNet.Clustering.Options
Configuration options for Spectral Clustering.
Inheritance: ModelOptions → ClusteringOptions<T> → SpectralOptions<T>
For Beginners
Spectral clustering finds clusters by analyzing connections.
Think of your data as a graph where:
- Each point is a node
- Similar points are connected by edges
- Clusters are groups of densely connected nodes
The algorithm:
- Build a similarity graph (like a social network)
- Find the “natural cuts” in this graph
- These cuts define your clusters
When to use:
- Clusters have unusual shapes (crescents, spirals)
- You can define a good similarity measure
- Data has clear connectivity patterns
How It Works
Spectral clustering uses the eigenvalues of a similarity matrix to perform dimensionality reduction before clustering. It can find non-convex clusters that K-Means cannot.
Constructors
| Constructor | Summary |
|---|---|
SpectralOptions() |
Initializes a new instance of SpectralOptions. |
Properties
| Property | Summary |
|---|---|
AffinityType Affinity { get; set; } |
Gets or sets the affinity/similarity method. |
EigenSolver EigenSolver { get; set; } |
Gets or sets the eigenvalue solver method. |
LaplacianNormalization Normalization { get; set; } |
Gets or sets the type of Laplacian normalization. |
SpectralAssignment AssignLabels { get; set; } |
Gets or sets the assignment strategy after spectral embedding. |
double? Gamma { get; set; } |
Gets or sets the gamma parameter for RBF kernel. |
int NumClusters { get; set; } |
Gets or sets the number of clusters. |
int NumNeighbors { get; set; } |
Gets or sets the number of neighbors for nearest neighbors affinity. |