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

Self-supervised visual representation learning has recently attractedsignificant research interest. While a common way to evaluate self-supervisedrepresentations is through transfer to various downstream tasks, we insteadinvestigate the problem of measuring their interpretability, i.e. understandingthe semantics encoded in raw representations. We formulate the latter asestimating the mutual information between the representation and a space ofmanually labelled concepts. To quantify this we introduce a decodingbottleneck: information must be captured by simple predictors, mapping conceptsto clusters in representation space. This approach, which we call reverselinear probing, provides a single number sensitive to the semanticity of therepresentation. This measure is also able to detect when the representationcontains combinations of concepts (e.g., "red apple") instead of justindividual attributes ("red" and "apple" independently). Finally, we propose touse supervised classifiers to automatically label large datasets in order toenrich the space of concepts used for probing. We use our method to evaluate alarge number of self-supervised representations, ranking them byinterpretability, highlight the differences that emerge compared to thestandard evaluation with linear probes and discuss several qualitativeinsights. Code at: {\scriptsize{\url{https://github.com/iro-cp/ssl-qrp}}}.

Measuring the Interpretability of Unsupervised Representations via Quantized Reverse Probing