Mean Shift#
A hierarchical clustering algorithm that uses peak finding to locate the candidate centroids of a training set given a radius constraint. Near-duplicate candidates are merged together in a final post-processing step.
Interfaces: Estimator, Learner, Probabilistic, Verbose, Persistable
Data Type Compatibility: Continuous
Parameters#
| # | Param | Default | Type | Description |
|---|---|---|---|---|
| 1 | radius | float | The bandwidth of the radial basis function. | |
| 2 | ratio | 0.1 | float | The ratio of samples from the training set to seed the algorithm with. |
| 3 | epochs | 100 | int | The maximum number of training rounds to execute. |
| 4 | min change | 1e-4 | float | The minimum change in centroids necessary for the algorithm to continue training. |
| 5 | tree | BallTree | Spatial | The spatial tree used to run range searches. |
| 6 | seeder | Random | Seeder | The seeder used to initialize the cluster centroids. |
Additional Methods#
Estimate the radius of a cluster that encompasses a certain percentage of the total training samples:
public static estimateRadius(Dataset $dataset, float $percentile = 30., ?Distance $kernel = null) : float
Note: Since radius estimation scales quadratically in the number of samples, for large datasets you can speed up the process by running it on a smaller subset of the training data.
Return the centroids computed from the training set:
public centroids() : array
Returns the amount of centroid shift during each epoch of training:
public steps() : array
Example#
use Rubix\ML\Clusterers\MeanShift;
use Rubix\ML\Graph\Trees\BallTree;
use Rubix\ML\Clusterers\Seeders\KMC2;
$radius = MeanShift::estimateRadius($dataset);
$estimator = new MeanShift($radius); // Set radius automatically
$estimator = new MeanShift(2.5, 2000, 1e-6, 0.05, new BallTree(100), new KMC2());
References#
- M. A. Carreira-Perpinan et al. (2015). A Review of Mean-shift Algorithms for Clustering.
- D. Comaniciu et al. (2012). Mean Shift: A Robust Approach Toward Feature Space Analysis.