This repository contains two innovative methods for solving manifold alignment problems. The first method, Shortest Paths on the Union of Domains (SPUD), focuses on finding the optimal path between points across multiple domains by using shortest path algorithms. The second method, Manifold Alignment via Stochastic Hopping (MASH), introduces random stochastic hopping to improve alignment quality in noisy datasets. You can learn more about these methods by reading our linked paper.
Created Sept, 2024.
To install the mashspud package, run the following command:
pip install git+https://github.com/rustadadam/mashspud.gitBelow is a quick code example of how to use SPUD from the mashspud package. The code implementation to MASH is similar. You can view full demonstrations of both MASH and SPUD in their respective deomnstration notebook files.
Example:
from mashspud import SPUD
from demonstration_helper import *
# Initialize the data
iris_features, iris_labels = prepare_dataset("csv_files/iris.csv")
iris_domainA, iris_domainB = split_features(iris_features, split = "distort") #Create domains
iris_anchors = create_anchors(int(len(iris_features)))[:10] #Create 10 random anchors
# Initialize SPUD and fit it
iris_spud = SPUD(knn = 8, verbose = 3)
iris_spud.fit(dataA = iris_domainA, dataB = iris_domainB, known_anchors=iris_anchors)
#Plot Embedding
iris_spud.plot_emb(labels = iris_labels)
- Rhodes, Jake
- Rustad, Adam