FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4324134093> ?p ?o ?g. }

• W4324134093 endingPage "025006" @default.
• W4324134093 startingPage "025006" @default.
• W4324134093 abstract "We develop the use of mutual information (MI), a well-established metric in information theory, to interpret the inner workings of deep learning models. To accurately estimate MI from a finite number of samples, we present GMM-MI (pronounced $``$Jimmie$$), an algorithm based on Gaussian mixture models that can be applied to both discrete and continuous settings. GMM-MI is computationally efficient, robust to the choice of hyperparameters and provides the uncertainty on the MI estimate due to the finite sample size. We extensively validate GMM-MI on toy data for which the ground truth MI is known, comparing its performance against established mutual information estimators. We then demonstrate the use of our MI estimator in the context of representation learning, working with synthetic data and physical datasets describing highly non-linear processes. We train deep learning models to encode high-dimensional data within a meaningful compressed (latent) representation, and use GMM-MI to quantify both the level of disentanglement between the latent variables, and their association with relevant physical quantities, thus unlocking the interpretability of the latent representation. We make GMM-MI publicly available." @default.
• W4324134093 created "2023-03-15" @default.
• W4324134093 creator A5003176673 @default.
• W4324134093 creator A5027649787 @default.
• W4324134093 creator A5036614505 @default.
• W4324134093 creator A5042055184 @default.
• W4324134093 creator A5062434214 @default.
• W4324134093 creator A5082949304 @default.
• W4324134093 date "2023-04-11" @default.
• W4324134093 modified "2023-09-26" @default.
• W4324134093 title "A robust estimator of mutual information for deep learning interpretability" @default.
• W4324134093 cites W1827461051 @default.
• W4324134093 cites W1965555277 @default.
• W4324134093 cites W1980751286 @default.
• W4324134093 cites W1982197353 @default.
• W4324134093 cites W1988115241 @default.
• W4324134093 cites W1994618660 @default.
• W4324134093 cites W1995920349 @default.
• W4324134093 cites W2012119171 @default.
• W4324134093 cites W2025234854 @default.
• W4324134093 cites W2027657888 @default.
• W4324134093 cites W2040704490 @default.
• W4324134093 cites W2049633694 @default.
• W4324134093 cites W2056749682 @default.
• W4324134093 cites W2060213761 @default.
• W4324134093 cites W2080722576 @default.
• W4324134093 cites W2082290707 @default.
• W4324134093 cites W2092939357 @default.
• W4324134093 cites W2103496339 @default.
• W4324134093 cites W2113051562 @default.
• W4324134093 cites W2114771311 @default.
• W4324134093 cites W2122378596 @default.
• W4324134093 cites W2122477496 @default.
• W4324134093 cites W2127234432 @default.
• W4324134093 cites W2133128370 @default.
• W4324134093 cites W2134478553 @default.
• W4324134093 cites W2137983211 @default.
• W4324134093 cites W2142635246 @default.
• W4324134093 cites W2150593711 @default.
• W4324134093 cites W2158196600 @default.
• W4324134093 cites W2163922914 @default.
• W4324134093 cites W2168175751 @default.
• W4324134093 cites W2170794187 @default.
• W4324134093 cites W2219726305 @default.
• W4324134093 cites W2521255850 @default.
• W4324134093 cites W2548006506 @default.
• W4324134093 cites W2548984383 @default.
• W4324134093 cites W2791833235 @default.
• W4324134093 cites W2898674840 @default.
• W4324134093 cites W2907100055 @default.
• W4324134093 cites W2936324353 @default.
• W4324134093 cites W2950418866 @default.
• W4324134093 cites W2964325296 @default.
• W4324134093 cites W2967281065 @default.
• W4324134093 cites W2989653627 @default.
• W4324134093 cites W2996320484 @default.
• W4324134093 cites W3010093582 @default.
• W4324134093 cites W3014517510 @default.
• W4324134093 cites W3015544277 @default.
• W4324134093 cites W3033188742 @default.
• W4324134093 cites W3087196548 @default.
• W4324134093 cites W3088227899 @default.
• W4324134093 cites W3089235226 @default.
• W4324134093 cites W3098516077 @default.
• W4324134093 cites W3100110286 @default.
• W4324134093 cites W3100196865 @default.
• W4324134093 cites W3103877238 @default.
• W4324134093 cites W3112135819 @default.
• W4324134093 cites W3116286104 @default.
• W4324134093 cites W3216431012 @default.
• W4324134093 cites W4226357961 @default.
• W4324134093 cites W4232613155 @default.
• W4324134093 cites W4281631709 @default.
• W4324134093 cites W572470661 @default.
• W4324134093 doi "https://doi.org/10.1088/2632-2153/acc444" @default.
• W4324134093 hasPublicationYear "2023" @default.
• W4324134093 type Work @default.
• W4324134093 citedByCount "0" @default.
• W4324134093 crossrefType "journal-article" @default.
• W4324134093 hasAuthorship W4324134093A5003176673 @default.
• W4324134093 hasAuthorship W4324134093A5027649787 @default.
• W4324134093 hasAuthorship W4324134093A5036614505 @default.
• W4324134093 hasAuthorship W4324134093A5042055184 @default.
• W4324134093 hasAuthorship W4324134093A5062434214 @default.
• W4324134093 hasAuthorship W4324134093A5082949304 @default.
• W4324134093 hasBestOaLocation W43241340931 @default.
• W4324134093 hasConcept C105795698 @default.
• W4324134093 hasConcept C119857082 @default.
• W4324134093 hasConcept C124101348 @default.
• W4324134093 hasConcept C146849305 @default.
• W4324134093 hasConcept C151730666 @default.
• W4324134093 hasConcept C152139883 @default.
• W4324134093 hasConcept C153180895 @default.
• W4324134093 hasConcept C154945302 @default.
• W4324134093 hasConcept C162324750 @default.
• W4324134093 hasConcept C176217482 @default.
• W4324134093 hasConcept C17744445 @default.
• W4324134093 hasConcept C185429906 @default.

• next