SUBCLU<T>
Models & Types · AiDotNet.Clustering.Subspace
SUBCLU (SUBspace CLUstering) density-connected subspace clustering algorithm.
Inheritance: ClusteringBase<T> → SUBCLU<T>
Implements: IGeneratedParameterRegistrar<T>
For Beginners
SUBCLU efficiently finds density-based clusters:
The key insight is “downward closure”:
- If points are NOT a cluster in 2D, they can’t be a cluster in 3D or higher
- So we can skip many subspace combinations!
Example:
- Check features 1 and 2: no cluster found
- Skip checking features 1, 2, and 3 (waste of time)
- Only check subspaces that might have clusters
This makes SUBCLU much faster than brute-force subspace search.
How It Works
SUBCLU is a subspace clustering algorithm based on the DBSCAN density concept. It exploits the monotonicity property: if a cluster exists in a k-dimensional subspace, it must exist in all (k-1)-dimensional projections of that subspace.
Algorithm steps:
- Apply DBSCAN to each 1-D subspace
- Generate candidate 2-D subspaces from 1-D clusters
- Apply DBSCAN to candidate subspaces
- Use monotonicity to prune: no cluster in 2-D means skip higher dimensions
- Continue until no more candidates or max dimension reached
Example
using AiDotNet;
using AiDotNet.Data.Loaders;
using AiDotNet.Clustering.Subspace;
using AiDotNet.Tensors.LinearAlgebra;
var data = new Matrix<double>(6, 2);
double[][] rows = { new[] { 1.0, 1.0 }, new[] { 1.2, 0.9 }, new[] { 1.1, 1.1 },
new[] { 8.0, 8.0 }, new[] { 8.2, 7.9 }, new[] { 7.9, 8.1 } };
for (int i = 0; i < 6; i++) { data[i, 0] = rows[i][0]; data[i, 1] = rows[i][1]; }
var result = await new AiModelBuilder<double, Matrix<double>, Vector<double>>()
.ConfigureModel(new SUBCLU<double>())
.ConfigureDataLoader(DataLoaders.FromMatrix(data))
.BuildAsync();
var labels = result.Predict(data);
Console.WriteLine($"SUBCLU: clustered {labels.Length} points.");
Constructors
| Constructor | Summary |
|---|---|
SUBCLU(SUBCLUOptions<T> options) |
Initializes a new SUBCLU instance. |
Properties
| Property | Summary |
|---|---|
IReadOnlyList<SubspaceCluster> SubspaceClusters { get; } |
Gets the discovered subspace clusters. |
Methods
| Method | Summary |
|---|---|
IFullModel<T, Matrix<T>, Vector<T>> WithParameters(Vector<T> parameters) |
|
ModelOptions GetOptions() |
|
Vector<T> Predict(Matrix<T> x) |
|
void Train(Matrix<T> x, Vector<T> y) |