AI Paper Atlas

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

Deep Determinantal Point Process for Large-Scale Multi-Label Classification

Pengtao Xie, Ruslan Salakhutdinov, Luntian Mou, Eric P. Xing

ICCV 2017 · Label-efficient and robust learning › Recognition and long-tail · approach: Partial and multi-label learning · 22 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
learn DPP label correlations jointly with visual features instead of separately (ablation)DDPP · YouTube-8M val (frame-level)MAP (%): 29.1 -> 30.3Improved
learn DPP label correlations jointly with visual features instead of separately (ablation)DDPP · Open Images valMAP: 64.6 -> 65.1Improved
add relational regularization with external must/cannot links (RDDPP) (ablation)DDPP · YouTube-8M val (frame-level)MAP (%): 30.3 -> 31.8Improved
add relational regularization with external must/cannot links (RDDPP) (ablation)DDPP · YouTube-8M val (video-level)MAP (%): 33.8 -> 34.9Improved
add relational regularization with external must/cannot links (RDDPP) (ablation)DDPP · Open Images valMAP: 65.1 -> 67.2Improved
increase number of must/cannot links from 2,000 to 10,000 (ablation)RDDPP · YouTube-8M frame-levelMAPImproved
increase number of must/cannot links beyond 10,000 (ablation)RDDPP · YouTube-8M frame-levelMAPNo clear effect
model high-order label correlations with DPP instead of independent labels (LSTM-IL)LSTM encoder · YouTube-8M val (frame-level)MAP (%): 26.6 -> 30.3Improved
use DDPP instead of SPENstructured MLC · YouTube-8M val (frame-level)MAP (%): 27.4 -> 30.3Improved
use DDPP instead of CGMMLC on video-level features · YouTube-8M val (video-level)MAP (%): 31.5 -> 33.8Improved
use low-rank kernel learning with inducing points instead of full-rank kernelDDPP · YouTube-8Mconvergence time (hours): 19.3 -> 7.4Improved
use low-rank kernel learning with inducing points instead of full-rank kernelDDPP · Open Imagesconvergence time (hours): 38.2 -> 10.4Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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