t-SNE Explorer

A) Synthetic Data Generator

Generate n points in d dimensions with k distinct distance types. Optimal: k=1 (n≤d+1 simplex), k=2 (n=5 pentagon), k=3 (n=7 heptagon). Larger n uses approximate lattice constructions.

n (number of points):

d (dimensions):

k (distinct distance types):

seed:

B) t-SNE Runner

Data Source: ✓ v6: tqdm-style progress bars for loading & t-SNE!

📊 MNIST Configuration

Subset:

Number of Samples:

Select Numeric Columns:

Handle Missing Values:

Embedding Method:

Initialization:

Custom Initial Coordinates (JSON format):

Provide n×2 array of initial 2D coordinates

Perplexity:

Learning Rate:

Iterations:

Early Exaggeration:

Momentum:

Seed:

C) Outputs & Internals

2D t-SNE Result

Cluster Distribution

Run t-SNE to see cluster results

Cost (KL Divergence) Over Iterations

P Matrix (High-D Affinities)

Q Matrix (Low-D Affinities)

Distances Between y_i (Embedding)

2D Coordinates

Clustering

Method:

k (clusters):

t-SNE Explorer

A) Synthetic Data Generator

Generated Points

Pairwise Distances

B) t-SNE Runner

C) Outputs & Internals

2D t-SNE Result

Cluster Distribution

Cost (KL Divergence) Over Iterations

P Matrix (High-D Affinities)

Q Matrix (Low-D Affinities)

Distances Between y_i (Embedding)

2D Coordinates

Clustering

Upload Data

CSV Files

Images

Uploaded Datasets

A) Synthetic Data Generator

Generated Points

Pairwise Distances

B) t-SNE Runner

C) Outputs & Internals

2D t-SNE Result

Cluster Distribution

Cost (KL Divergence) Over Iterations

P Matrix (High-D Affinities)

Q Matrix (Low-D Affinities)

Distances Between yi (Embedding)

2D Coordinates

Clustering

Upload Data

CSV Files

Images

Uploaded Datasets

Distances Between y_i (Embedding)