AI Paper Atlas

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

Introspective Classification with Convolutional Nets

Long Jin, Justin Lazarow, Zhuowen Tu

NeurIPS 2017 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 33 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 iteratively synthesized pseudo-negatives rather than using only the CNN training setCNN · MNIST, one-vs-alltest error (%): 0.87 -> 0.78Improved
replace meaningful synthesized pseudo-negatives with random noise (ablation)ICN · MNIST, one-vs-alltest error (%): 0.78 -> 0.89Hurt
add iteratively synthesized pseudo-negativesResNet-32 · MNIST, softmaxtest error (%): 0.50 -> 0.47Improved
add iteratively synthesized pseudo-negativesResNet-32 · CIFAR-10, no data augmentation, one-vs-alltest error (%): 13.44 -> 12.94Improved
add iteratively synthesized pseudo-negativesResNet-32 · CIFAR-10, no data augmentation, softmaxtest error (%): 12.38 -> 11.46Improved
add iteratively synthesized pseudo-negativesResNet-32 · CIFAR-10, data augmentation, softmaxtest error (%): 7.06 -> 6.70Improved
use random noise instead of classifier-synthesized pseudo-negatives (ablation)ICN with ResNet-32 · CIFAR-10, no data augmentation, softmaxtest error (%): 11.46 -> 11.94Hurt
add iteratively synthesized pseudo-negativesResNet-32 · SVHN, softmaxtest error (%): 2.01 -> 1.95Improved
choose between SGD and SGD Langevin sampling, or the listed stopping options (ablation)ICN · MNIST, classification test errortest error (%): 0.78 -> 0.80; 0.81; 0.72; 0.74; 0.75; 0.73No clear effect
use label smoothingResNet-32 · CIFAR-10, no data augmentation, softmaxtest error (%): 12.38 -> 12.65Hurt

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