AI Paper Atlas

Research map / Learning theory and algorithms / Optimal transport

Aggregated Wasserstein Distance and State Registration for Hidden Markov Models

Yukun Chen, Jianbo Ye, Jia Li

TPAMI 2019 · Learning theory and algorithms › Optimal transport · approach: Wasserstein barycenters and GW · 35 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 MAW or IAW instead of KL divergence as HMM distanceMAW · Mocap motion classification (6 joint-group)accuracy (%): 90.53% -> 92.90%Improved
use IAW instead of KL divergence as HMM distanceIAW · Mocap motion classification (27 joint-group)accuracy (%): 95.86% -> 98.22%Improved
use MAW instead of KL on 27 joint-group schemeMAW · Mocap motion classification (27 joint-group)accuracy (%): 95.86% -> 94.67%Hurt
use IAW instead of KL in 1-NN phoneme classificationIAW · TIMIT phoneme k-NN (k=1)accuracy: 0.492 -> 0.532Improved
use IAW instead of MAW in phoneme k-NN classificationIAW · TIMIT phoneme k-NNaccuracy: 0.553 -> 0.584 (k=5)Improved
use MAW instead of KL in phoneme k-NN classificationMAW · TIMIT phoneme k-NNaccuracy: 0.492 -> 0.525 (k=1); 0.560 -> 0.547 (k=10)Mixed
use closed-form MAW instead of sampling-based KL for computationMAW · Synthetic datatime per distance: 5.8ms -> 3.8msImproved
use IAW instead of KL for computationIAW · Synthetic datatime per distance: 5.8ms -> 57.9msHurt

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