English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Most examinations of neural networks' learned latent spaces typically employdimensionality reduction techniques such as t-SNE or UMAP. While these methodseffectively capture the overall sample distribution in the entire learnedlatent space, they tend to distort the structure of sample distributions withinspecific classes in the subset of the latent space. This distortion complicatesthe task of easily distinguishing classes identifiable by neural networks. Inresponse to this challenge, we introduce the k* Distribution methodology. Thisapproach focuses on capturing the characteristics and structure of sampledistributions for individual classes within the subset of the learned latentspace using local neighborhood analysis. The key concept is to facilitate easycomparison of different k* distributions, enabling analysis of how variousclasses are processed by the same neural network. This provides a more profoundunderstanding of existing contemporary visualizations. Our study reveals threedistinct distributions of samples within the learned latent space subset: a)Fractured, b) Overlapped, and c) Clustered. We note and demonstrate that thedistribution of samples within the network's learned latent space significantlyvaries depending on the class. Furthermore, we illustrate that our analysis canbe applied to explore the latent space of diverse neural network architectures,various layers within neural networks, transformations applied to inputsamples, and the distribution of training and testing data for neural networks.We anticipate that our approach will facilitate more targeted investigationsinto neural networks by collectively examining the distribution of differentsamples within the learned latent space.

k* Distribution: Evaluating the Latent Space of Deep Neural Networks using Local Neighborhood Analysis