Pith. sign in

REVIEW 3 cited by

Milking CowMask for Semi-Supervised Image Classification

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2003.12022 v3 pith:UV36QCAL submitted 2020-03-26 cs.CV

Geoff French , Avital Oliver , Tim Salimans This is my paper · ORCID
classification cs.CV
keywords cowmaskdatasemi-supervisedachieveclassificationconsistencyerrorgoogle-research
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Consistency regularization is a technique for semi-supervised learning that underlies a number of strong results for classification with few labeled data. It works by encouraging a learned model to be robust to perturbations on unlabeled data. Here, we present a novel mask-based augmentation method called CowMask. Using it to provide perturbations for semi-supervised consistency regularization, we achieve a state-of-the-art result on ImageNet with 10% labeled data, with a top-5 error of 8.76% and top-1 error of 26.06%. Moreover, we do so with a method that is much simpler than many alternatives. We further investigate the behavior of CowMask for semi-supervised learning by running many smaller scale experiments on the SVHN, CIFAR-10 and CIFAR-100 data sets, where we achieve results competitive with the state of the art, indicating that CowMask is widely applicable. We open source our code at https://github.com/google-research/google-research/tree/master/milking_cowmask

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Data-Driven Registration and Modeling of Brain Deformation for Image-Guided Neurosurgery: A Systematic Review

    eess.IV 2026-02 conditional novelty 4.0 of 10

    AI-based brain-deformation registration methods are fast and increasingly accurate, but still lag classical methods on hard multimodal tasks and lack robust out-of-distribution validation.

  2. Assessing the Advantages and Limitations of Quantum Neural Networks in Regression Tasks

    quant-ph 2025-08 conditional novelty 4.0 of 10

    Quantum neural networks strongly outperform narrow classical networks on smooth function regression, but the advantage depends on comparison design and disappears on discontinuous functions.

  3. Survey of Swarm Intelligence Approaches to Search Documents Based On Semantic Similarity

    cs.AI 2025-07 reject novelty 1.0 of 10

    A survey of swarm intelligence text-mining papers finds a gap: no published work yet combines swarm optimization with sentence embeddings for document search based on semantic similarity.

Pith tools