Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Cluster Exploration using Informative Manifold Projections

About

Dimensionality reduction (DR) is one of the key tools for the visual exploration of high-dimensional data and uncovering its cluster structure in two- or three-dimensional spaces. The vast majority of DR methods in the literature do not take into account any prior knowledge a practitioner may have regarding the dataset under consideration. We propose a novel method to generate informative embeddings which not only factor out the structure associated with different kinds of prior knowledge but also aim to reveal any remaining underlying structure. To achieve this, we employ a linear combination of two objectives: firstly, contrastive PCA that discounts the structure associated with the prior information, and secondly, kurtosis projection pursuit which ensures meaningful data separation in the obtained embeddings. We formulate this task as a manifold optimization problem and validate it empirically across a variety of datasets considering three distinct types of prior knowledge. Lastly, we provide an automated framework to perform iterative visual exploration of high-dimensional data.

Stavros Gerolymatos, Xenophon Evangelopoulos, Vladimir Gusev, John Y. Goulermas• 2023

Related benchmarks

TaskDatasetResultRank
Clustering EvaluationMNIST 10,000 random samples
Jaccard Index44
22
ClusteringUCI Image Segmentation
Mean Jaccard0.68
16
Linear ClassificationComplex MNIST+FMNIST Tshirt-Dress (test)
Accuracy88
8
ClassificationCIFAR-100 + FMNIST Tshirt-Shirt (test)
Accuracy78
4
ClassificationCIFAR-100 + FMNIST Tshirt-Coat (test)
Accuracy90
4
Image ClassificationMNIST+FMNIST (test)
Test Accuracy81
4
Image ClassificationCIFAR 100+FMNIST (test)
Mean Test Accuracy95
4
Linear ClassificationComplex MNIST+FMNIST Sandal-Ankle boot (test)
Accuracy89
4
SVM ClassificationMNIST + FMNIST Sandal-Sneaker Complex (test)
Accuracy81
4
SVM ClassificationCIFAR-100 + FMNIST Tshirt-Shirt (test)
Accuracy79
4
Showing 10 of 15 rows

Other info

Follow for update