Reading Digits in Natural Images with Unsupervised Feature Learning

By Yuval Netzer, Tao Wang, Adam Coates, Alessandro Bissacco, Bo Wu, Andrew Y. Ng. 2011

8,121 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

FitNets: Hints for Thin Deep Nets (2014, methodology): “The SVHN dataset (Netzer et al., 2011) Table 3 shows that our FitNet achieves comparable accuracy than the teacher despite using only 32% of teacher capacity.”
Network In Network (2013, methodology): “We evaluate NIN on four benchmark datasets: CIFAR-10 [12], CIFAR-100 [12], SVHN [13] and MNIST [1].”
Maxout Networks (2013, methodology): “The SVHN (Netzer et al., 2011) For SVHN, we did not train on the validation set at all.”
Pylearn2: a machine learning research library (2013, methodology): “These include the datasets used for DARPA’s unsupervised and transfer learning challenge [25] , the dataset used for the NIPS 2011 workshops challenge in representation learning [32], the C IFAR-10 and CIFAR-100 datasets [29], the MNIST dataset [34], some of the MNIST variations dataset s [31], the ”
Learning Transferable Visual Models From Natural Language Supervision (2021, background): “Three of these datasets MNIST (LeCun), SVHN (Netzer et al., 2011), and IIIT5K (Mishra et al., 2012) directly check the ability of a model to perform low-level character and word recognition, while Hateful Memes (Kiela et al., 2020) and SST-2 (Socher et al., 2013) check the ability of a model to use…”
Unsupervised Data Augmentation for Consistency Training (2019, methodology): “Firstly, as shown in Table 6, if we apply the augmentation policy found on SVHN by AutoAugment to CIFAR-10 (denoted by Switched Augment), the error rate increases from 5 .”
An Empirical Model of Large-Batch Training (2018, methodology): “• VAE and Autoencoder We train a VAE [KW13] and a simple Autoencoder on the SVHN dataset [NWC + 11]; we were motivated to compare these models because VAEs introduce additional stochasticity.”
On Robustness and Transferability of Convolutional Neural Networks (2020, background): “13 [46] Yuval Netzer, Tao Wang, Adam Coates, Alessandro Bissacco, Bo Wu, and Andrew Y.”
A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark (2019, methodology): “The group includes: Caltech101, CIFAR-100, DTD, Flowers102, Pets, Sun397, and SVHN.”
S4L: Self-Supervised Semi-Supervised Learning (2019, methodology): “Semi-supervised learning algorithms are typically [28, 22, 2, 39, 1, 21] evaluated on small-scale datasets such as CIFAR-10 [17] and SVHN [23].”
Spatial Transformer Networks (2015, methodology): “4.2 we test spatial transformer networks on a challenging real-world dataset, Street View House Numbers [25], for number recognition, showing stateof-the-art results using multiple spatial transformers embedded in the convolutional stack of a CNN. Finally, in Sect.”
Challenges in representation learning: A report on three machine learning contests (2013, methodology): “The dataset is an obfuscated subset of the second (MNIST-like) format of the Street View House Numbers dataset[3].”
Randaugment: Practical automated data augmentation with a reduced search space (2019, methodology): “We report the computational cost, the search space size, and the test accuracy achieved for AutoAugment (AA) [4], Fast AutoAugment [19], Population Based Augmentation (PBA) [15] and the proposed RandAugment (RA) on CIFAR10 [16], SVHN [28], and ImageNet [5] classification tasks.”
VQ-DRAW: A Sequential Discrete VAE (2020, methodology): “Stage conditioning resulted in large improvements on CIFAR-10, where the refinement sequence is an order of magnitude longer than the sequences for MNIST and SVHN. 2.”
Predictive Sampling with Forecasting Autoregressive Models (2020, methodology): “191 for SVHN (To obtain the bits per latent dimension, multiply these by the dimensionality reduction factor 12 ).”
Flow Contrastive Estimation of Energy-Based Models (2019, methodology): “We conduct experiments on the Street View House Numbers (SVHN) [53], CIFAR-10 [35] and CelebA [45] datasets.”
BIVA: A Very Deep Hierarchy of Latent Variables for Generative Modeling (2019, background): “(3) for various values of k both on the CIFAR-10 test set and on the SVHN dataset (Netzer et al., 2011).”
Variational Walkback: Learning a Transition Operator as a Stochastic Recurrent Net (2017, methodology): “We then describe experimental results illustrating the soundness of the proposed approach on the MNIST, CIFAR-10, SVHN and CelebA datasets.”
Learning Generative ConvNets via Multi-grid Modeling and Sampling (2017, methodology): “We learn multi-grid models from ﬁve datasets: CelebA [27], Large-scale Scene Understanding (LSUN) [41], CIFAR-10 [20], Street View Housing Numbers (SVHN) [34] and MIT places205 [47].”