AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

A Concise Yet Effective Model for Non-Aligned Incomplete Multi-View and Missing Multi-Label Learning

Xiang Li, Songcan Chen

TPAMI 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Incomplete multi-view multi-label learning · 116 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 the local low-rank term to the regression loss (ablation)NAIM3L-I · Corel5k (50% incomplete views, 50% missing labels)1-RL (%): 82.73 -> 83.54Improved
add the global high-rank (negative trace norm) term (ablation)NAIM3L-II · Corel5k (50% incomplete views, 50% missing labels)1-RL (%): 83.54 -> 87.84Improved
add the global high-rank (negative trace norm) term (ablation)NAIM3L-II · Corel5k (50% incomplete views, 50% missing labels)AUC (%): 83.80 -> 88.13Improved
add the global high-rank (negative trace norm) term (ablation)NAIM3L-II · Pascal07 (50% incomplete views, 50% missing labels)1-RL (%): 77.35 -> 78.30Improved
add the global high-rank (negative trace norm) term (ablation)NAIM3L-II · ESPGame (50% incomplete views, 50% missing labels)1-RL (%): 79.80 -> 81.81Improved
add the global high-rank (negative trace norm) term (ablation)NAIM3L-II · Mirflickr (50% incomplete views, 50% missing labels)AP (%): 54.98 -> 55.08Improved
add the label-structure regularizer to either variant (ablation)NAIM3L-I / NAIM3L-II · Corel5k (50% incomplete views, 50% missing labels)1-HL (%): 98.70 -> 98.70No clear effect
use NAIM3L (non-aligned) instead of iMVWL (aligned)iMVWL · Corel5k (50% incomplete views, 50% missing labels)AP (%): 28.31 -> 30.88Improved
use NAIM3L instead of iMVWLiMVWL · Mirflickr (50% incomplete views, 50% missing labels)AP (%): 49.48 -> 55.08Improved
use NAIM3L-V instead of iMVWL-V (50% incomplete views, full labels)iMVWL-V · Pascal07 (50% incomplete views, full labels)AP (%): 43.54 -> 49.25Improved
use NAIM3L-L instead of iMVWL-L (50% missing labels, complete views)iMVWL-L · ESPGame (50% missing labels, complete views)AP (%): 25.80 -> 24.90Hurt
give iMVWL full labels instead of 50% missing labelsiMVWL · Mirflickr (50% incomplete views)AP (%): 49.48 -> 48.95Hurt

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