AI Paper Atlas

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

How can classical multidimensional scaling go wrong?

Rishi Sonthalia, Greg Vanbuskirk, Benjamin Raichel, Anna Gilbert

NeurIPS 2021 · Learning theory and algorithms › Matrix and tensor methods · approach: Randomized sketching for matrices · 1 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
increase embedding dimensioncMDS · perturbed Euclidean metrics (heart)relative SSTRESS vs perturbed inputHurt
increase embedding dimensioncMDS · perturbed Euclidean metrics (heart)relative squared error to original metric (denoising)Hurt
increase embedding dimensioncMDS · Portugal graph and Isomap on heartrelative squared errorHurt
replace cMDS with Lower + cMDScMDS · Celegans, Portugal, heart (perturbed, Isomap)relative SSTRESSImproved
replace cMDS with Lower + cMDScMDS embeddings + 3-layer NN · MNIST, Fashion MNIST, CIFAR10 (perturbed, kNN-graph, missing pixels)classification accuracy as dimension growsImproved
increase embedding dimension for classificationcMDS embeddings + 3-layer NN · MNIST, Fashion MNIST, CIFAR10 non-Euclidean metricsclassification accuracyHurt
use lower bound solution in place of true SSTRESS minimizerLower bound algorithm · relative squared errortracking of true SSTRESS solutionImproved
use Lower + cMDS vs true SSTRESS solution for computeLower + cMDS · Portugal and classification experimentsrunning timeImproved

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