TriMap

TriMap is a dimensionality reduction method that uses triplets of points to learn a low-dimensional embedding that preserves the global structure of the data.

How It Works

TriMap samples triplets of points (an anchor, a positive point, and a negative point) to capture the relationships between them. It then optimizes a low-dimensional embedding to satisfy these triplet constraints.

Why or When to Use

Offers a balance between preserving local and global structure. It often runs faster than t-SNE and preserves global structure better than UMAP or t-SNE.

Example

How-to (Code)

import * as druid from "@saehrimnir/druidjs";

const data = [
  /* ... multi-dimensional data ... */
];

// 1. Initialize the algorithm
const trimap = new druid.TriMap(data);

// 2. Compute the projection
const projection = trimap.transform();