AI Paper Atlas

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

Tensor Gaussian Process with Contraction for Multi-Channel Imaging Analysis

Hu Sun, Ward Manchester, Meng Jin, Yang Liu, Yang Chen

ICML 2023 · Learning theory and algorithms › Matrix and tensor methods · approach: Low-rank tensor completion · 2 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
Add tensor contraction step before a squared-exponential GP (SE+TC) vs. SE on vectorized tensor (ablation)GP with SE kernel · Simulation (N=200, 500)test RMSE: SE 2.504 -> SE+TC 0.627 (N=200); 3.275 -> 0.587 (N=500)Improved
Add tensor contraction (Tensor-GPST) vs. the plain multi-linear Tensor-GPTensor-GP · Simulation (N=200, 500)test RMSE: GP 0.728 -> GPST 0.578 (N=200); 0.664 -> 0.552 (N=500)Improved
Add tensor contraction (Tensor-GPST) vs. the plain multi-linear Tensor-GPTensor-GP · Simulation (N=200, 500)test MSLL: GP 1.162 -> GPST 0.882 (N=200); 1.092 -> 0.835 (N=500)Improved
Add tensor contraction step to SE-kernel GP (SE+TC) vs. SE alone, measured by MSLL (ablation)GP with SE kernel · Simulation (N=200, 500)test MSLL: SE 2.457 -> SE+TC 0.972 (N=200); 2.999 -> 0.919 (N=500)Improved
Add anisotropic TV penalty on contraction factors (GPST with lambda) vs. no TV penalty (SE+TC-like contraction)Tensor-GPST · Simulationtest RMSE: Simulation: GPST 0.578 vs SE+TC 0.627 (N=200); 0.552 vs 0.587 (N=500) — but SE+TC uses a different kernel, so not a cleaMixed
Tensor-GPST vs. Tensor-GP on solar flare intensity predictionTensor-GP · Solar flare forecastingtest RMSE: GP 0.772 -> GPST 0.681Improved
Tensor-GPST vs. CP and Tucker on solar flare forecastingTensor-GPST · Solar flare forecastingtest RMSE: GPST 0.681 vs CP 0.706, Tucker 0.683; GPST only slightly better and not significantly better than TuckerMixed
Tensor-GPST vs. CP baseline in simulationTensor-GPST · Simulation (N=200, 500)test RMSE: GPST 0.578 vs CP 0.550 (N=200); 0.552 vs 0.548 (N=500)No clear effect

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