Corpus ID: 218862858

From ImageNet to Image Classification: Contextualizing Progress on Benchmarks

@article{Tsipras2020FromIT,
  title={From ImageNet to Image Classification: Contextualizing Progress on Benchmarks},
  author={Dimitris Tsipras and Shibani Santurkar and Logan Engstrom and Andrew Ilyas and Aleksander Madry},
  journal={ArXiv},
  year={2020},
  volume={abs/2005.11295}
}
  • Dimitris Tsipras, Shibani Santurkar, +2 authors Aleksander Madry
  • Published 2020
  • Computer Science, Mathematics
  • ArXiv
  • Building rich machine learning datasets in a scalable manner often necessitates a crowd-sourced data collection pipeline. In this work, we use human studies to investigate the consequences of employing such a pipeline, focusing on the popular ImageNet dataset. We study how specific design choices in the ImageNet creation process impact the fidelity of the resulting dataset---including the introduction of biases that state-of-the-art models exploit. Our analysis pinpoints how a noisy data… CONTINUE READING

    Citations

    Publications citing this paper.

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 42 REFERENCES

    Bag of Tricks for Image Classification with Convolutional Neural Networks

    • Tong He, Zhi Zhang, +3 authors Mu Li
    • Computer Science
    • 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
    • 2019

    Learning Transferable Architectures for Scalable Image Recognition

    Microsoft COCO: Common Objects in Context

    VIEW 2 EXCERPTS
    HIGHLY INFLUENTIAL

    Rethinking the Inception Architecture for Computer Vision