AI Paper Atlas

Research map / Learning theory and algorithms / Matrix and tensor methods

Learning from Binary Multiway Data: Probabilistic Tensor Decomposition and its Statistical Optimality

Miaoyan Wang, Lexin Li

JMLR 2020 · Learning theory and algorithms › Matrix and tensor methods · approach: Randomized tensor decomposition algorithms · 27 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use binary (logistic) tensor decomposition instead of continuous-valued decompositionBTF_Alternating · Enron tensor completionAUC: 79.6% -> 94.3%Improved
use binary instead of continuous-valued tensor decompositionBTF_Alternating · Kinship, Nations, HCP completionAUC: 0.9436 -> 0.9708Improved
use binary instead of continuous-valued tensor decompositionBTF_Alternating · Enron tensor completionRMSE: 6.3 x 10^-5 -> 6.4 x 10^-3Hurt
increase tensor dimension dConstrained MLE · Simulated CP tensorestimation errorImproved
increase true rank RConstrained MLE · Simulated CP tensorestimation errorHurt
increase noise level sigma in high SNR regimeConstrained MLE · Simulated CP tensor, d=50estimation errorImproved
increase dimension d from 20 to 40 for BIC rank selectionBIC · Rank selectionselected rank: 8.7 -> 10Improved
use probabilistic Bernoulli model rather than BooleanTFBTF methods · Boolean tensor model dataRMSEImproved
use BTF_Alternating or BTF_Gradient instead of BTF_BayesianBTF_Alternating · Boolean tensor model datamisclassification errorImproved
fit logistic link to probit-generated data (misspecification)BTF_Alternating · Stochastic multiway block modelrelative loss: 0.22 to 0.48 relative loss across sub-modelsNo clear effect
use rank-R CP exact low-rank instead of unfolding into matrixMLE upper bound · Theoryconvergence rate: d^-(K-1)/4 -> d^-(K-1)/2Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy