AI Paper Atlas

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

ProtoAttend: Attention-Based Prototypical Learning

Sercan O. Arik, Tomas Pfister

JMLR 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Prototype and concept dictionary models · 42 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 sparsemax instead of softmax attention to reduce prototypes (ablation)ProtoAttend · MNISTprototypes for 50% of decision: 365 -> 2Improved
use sparsemax instead of softmax attention (ablation)ProtoAttend · DBPediaaccuracy (%): 98.20 -> 97.74Hurt
add sparsity regularization on top of sparsemax (ablation)ProtoAttend · Fashion-MNISTprototypes for 50% of decision: 4 -> 1Improved
add sparsity regularization on top of sparsemax (ablation)ProtoAttend · CIFAR-10accuracy (%): 91.44 -> 91.26Hurt
use ProtoAttend with sparsemax + sparsity reg. instead of baseline encoderbaseline encoder · CIFAR-10 label noise 0.8test accuracy (%): 57.02 -> 60.50Improved
use ProtoAttend instead of baseline encoderbaseline encoder · CIFAR-10 label noise 0.4test accuracy (%): 77.47 -> 80.04Improved
use ProtoAttend instead of dropoutdropout · CIFAR-10 label noise 0.8test accuracy (%): 56.76 -> 60.50Improved
train only with prototype output loss L(y,yhat(1)) instead of adding yhat(0) loss (ablation)ProtoAttend training loss · Fashion-MNISTaccuracy (%) of yhat(0): 94.37 -> 10.92Hurt
train only with L(y,yhat(0)) instead of adding yhat(1) loss (ablation)ProtoAttend training loss · Fashion-MNISTaccuracy (%) of yhat(1): 94.38 -> 13.13Hurt
use yhat(0.5) loss only instead of L0+L1 (ablation)ProtoAttend training loss · Fashion-MNISTconfidence for correct cases: 0.931 -> 0.927Hurt
add L(y,yhat(0.5)) to L0+L1 (ablation)ProtoAttend training loss · Fashion-MNISTaccuracy (%) of yhat(1): 94.38 -> 94.45No clear effect
use confidence score from prototype label agreementProtoAttend · confidence-controlled predictionaccuracy vs ratio of samplesImproved

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