AI Paper Atlas

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

Learning with Holographic Reduced Representations

Ashwinkumar Ganesan, Hang Gao, Sunil Gandhi, Edward Raff, Tim Oates, James Holt, Mark McLean

NeurIPS 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 0 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
replace the fully-connected output layer with the HRR dense label outputFC · BibtexP@1: 46.4 -> 60.3Improved
replace the fully-connected output layer with the HRR dense label outputFC · Delicious-200KP@1: 21.8 -> 44.9Improved
replace the fully-connected output layer with the HRR dense label outputFC · Delicious-200KPSP@1: 10.5 -> 6.84Hurt
replace the fully-connected output layer with the HRR dense label outputCNN · Wiki10-31KP@1: 60.0 -> 74.3Improved
replace the fully-connected output layer with the HRR dense label outputLSTM · EURLex-4KP@1: 63.0 -> 70.4Improved
replace the fully-connected output layer with the HRR dense label outputCNN · Amazon-12KP@1: 89.1 -> 84.5Hurt
replace the fully-connected output layer with the HRR dense label outputFC · Amazon-670KP@1: 34.6 -> 19.9Hurt
replace the fully-connected output layer with the HRR dense label outputFC · Amazon-670KPSP@1: 5.22 -> 8.45Improved
replace the fully-connected output layer with the HRR dense label outputFC · Amazon-670Ktraining time speedup: 6.28Improved
replace the fully-connected output layer with the HRR dense label outputFC · Bibtextraining time speedup: 0.35Hurt
make baseline FC deeper (2 to 3 layers, h=512) (ablation)FC · Wiki10-31KP@5: 46.64 -> 41.92Hurt
increase HRR output dimension d' and network size (h=512,d'=400 to 3 layers h=2048,d'=3000) (ablation)HRR-FC · Wiki10-31KP@5: 55.06 -> 58.51Improved

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