Self-Training With Noisy Student Improves ImageNet Classification
Citations
12,690 citations
Cites background from "Self-Training With Noisy Student Im..."
...The use of additional data sources allows to achieve state-ofthe-art results on standard benchmarks (Mahajan et al., 2018; Touvron et al., 2019; Xie et al., 2020)....
[...]
...Currently, Noisy Student is the state of the art on ImageNet and BiT-L on the other datasets reported here....
[...]
...Therefore, in large-scale image recognition, classic ResNetlike architectures are still state of the art (Mahajan et al., 2018; Xie et al., 2020; Kolesnikov et al., 2020)....
[...]
...The second is Noisy Student (Xie et al., 2020), which is a large EfficientNet trained using semi-supervised learning on ImageNet and JFT300M with the labels removed....
[...]
10,601 citations
1,027 citations
585 citations
576 citations
Cites methods from "Self-Training With Noisy Student Im..."
...The current state-of-the-art supervised model [154] on ImageNet follows the self-training paradigm, where we first train an EfficientNet model on labeled ImageNet images and use it as a teacher to generate pseudo labels on 300M unlabeled images....
[...]
References
123,388 citations
73,978 citations
49,914 citations
40,257 citations
33,597 citations
"Self-Training With Noisy Student Im..." refers methods in this paper
...We use stochastic depth [29], dropout [63] and RandAugment [14] to noise the student....
[...]
...To noise the student, we use dropout [63], data augmentation [14] and stochastic depth [29] during its training....
[...]
...In our experiments, we use dropout [63], stochastic depth [29], data augmentation [14] to noise the student....
[...]