Deep Generative Neural Embeddings for High Dimensional Data Visualization

TL;DR

This paper introduces a neural network-based visualization method that uses generative embeddings for high-dimensional data, offering flexible manipulation and scalability, outperforming traditional methods like t-SNE and VAE.

Contribution

It presents a novel neural embedding and generative network approach that enables independent manipulation and scalable visualization of high-dimensional data.

Findings

Effective data visualization demonstrated on ImageNet.

Outperforms t-SNE and VAE in visualization tasks.

Allows independent editing of embeddings for human-in-the-loop applications.

Abstract

We propose a visualization technique that utilizes neural network embeddings and a generative network to reconstruct original data. This method allows for independent manipulation of individual image embeddings through its non-parametric structure, providing more flexibility than traditional autoencoder approaches. We have evaluated the effectiveness of this technique in data visualization and compared it to t-SNE and VAE methods. Furthermore, we have demonstrated the scalability of our method through visualizations on the ImageNet dataset. Our technique has potential applications in human-in-the-loop training, as it allows for independent editing of embedding locations without affecting the optimization process.