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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

The last decade of machine learning has seen drastic increases in scale andcapabilities. Deep neural networks (DNNs) are increasingly being deployed inthe real world. However, they are difficult to analyze, raising concerns aboutusing them without a rigorous understanding of how they function. Effectivetools for interpreting them will be important for building more trustworthy AIby helping to identify problems, fix bugs, and improve basic understanding. Inparticular, "inner" interpretability techniques, which focus on explaining theinternal components of DNNs, are well-suited for developing a mechanisticunderstanding, guiding manual modifications, and reverse engineering solutions.  Much recent work has focused on DNN interpretability, and rapid progress hasthus far made a thorough systematization of methods difficult. In this survey,we review over 300 works with a focus on inner interpretability tools. Weintroduce a taxonomy that classifies methods by what part of the network theyhelp to explain (weights, neurons, subnetworks, or latent representations) andwhether they are implemented during (intrinsic) or after (post hoc) training.To our knowledge, we are also the first to survey a number of connectionsbetween interpretability research and work in adversarial robustness, continuallearning, modularity, network compression, and studying the human visualsystem. We discuss key challenges and argue that the status quo ininterpretability research is largely unproductive. Finally, we highlight theimportance of future work that emphasizes diagnostics, debugging, adversaries,and benchmarking in order to make interpretability tools more useful toengineers in practical applications.

Toward Transparent AI: A Survey on Interpreting the Inner Structures of Deep Neural Networks