AI Paper Atlas

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

The Decoupled Extended Kalman Filter for Dynamic Exponential-Family Factorization Models

Carlos A. Gomez-Uribe, Brian Karrer

JMLR 2021 · Learning theory and algorithms › Matrix and tensor methods · approach: Structured low-rank learning

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 block-diagonal per-entity covariance (DEKF) instead of full EKFEKF · simulated regression, MF, TF (static)predictionNo clear effect
use per-entity blocks instead of full diagonal covarianceDEKF · simulated datapredictionImproved
use full EKF instead of DEKFDEKF · simulated dynamic tensor factorizationpredictionImproved
model parameter dynamics instead of assuming static parameters (ablation)DEKF · simulated data with drifting parameterspredictionImproved
use Thompson sampling instead of posterior-mean recommendationDEKF · simulated explore-exploit, static parametersnormalized cumulative regretImproved
increase parameter drift by reducing half life to 10000 then 1000Thompson sampling with MF · simulated explore-exploitcumulative regretHurt
add parameter dynamics (ablation)DEKF · MovieLens-20M Gaussian-MFRMSEImproved
use diagonal DEKF instead of user/item entitiesDEKF · MovieLens-20M Bernoulli-MFnormalized cross-entropyImproved
use diagonal DEKF instead of user/item entitiesDEKF · MovieLens-20M Gaussian-MFRMSEHurt

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