The mnist database of handwritten digits

By Yann LeCun, Corinna Cortes. 2005

7,477 citations recorded by the source corpus.

How researchers see this paper

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

In Evaluation & Benchmarks

Citations in the corpus

Intriguing properties of neural networks (2013, methodology): “We perform a number of experiments on a few different networks and three datasets : • For the MNIST dataset, we used the following architectures [11] –”
Scaling description of generalization with number of parameters in deep learning (2019, methodology): “We first consider the task of classifying the parity of digits on the MNIST database [29].”
High-dimensional dynamics of generalization error in neural networks (2017, methodology): “Finally, to investigate the degree to which our findings extend to real world datasets, we trained nonlinear networks on a binary digit recognition task using the MNIST dataset (LeCun, Cortes, and Burges, 1998).”
Overcoming catastrophic forgetting in neural networks (2016, methodology): “Reference ﬁgure 3A 3B 3C learning rate 10 − 3 10 − 5 - 10 We carried out all MNIST experiments with fully-connected networks with rectiﬁed linear units.”
TensorFlow: A system for large-scale machine learning (2016, neutral): “The system must support expressive control-ﬂow and stateful constructs, while also satisfying our other requirements.”
Learning to Generate Samples from Noise through Infusion Training (2017, methodology): “We trained such a model using our infusion training procedure on MNIST (LeCun & Cortes, 1998), Toronto Face Database (Susskind et al., 2010), CIFAR-10 (Krizhevsky & Hinton, 2009), and CelebA (Liu et al., 2015).”
Variational Inference with Normalizing Flows (2015, methodology): “The MNIST digit dataset (LeCun & Cortes, 1998) contains 60,000 training and 10,000 test images of ten handwritten digits (0 to 9) that are 28 × 28 pixels in size.”
Deep Unsupervised Learning using Nonequilibrium Thermodynamics (2015, methodology): “MNIST In order to allow a direct comparison against previous work on a simple dataset, we trained on MNIST digits [26]. t ) passes through several layers of multi-scale convolution (Section 3.2.1).”
NICE: Non-linear Independent Components Estimation (2014, methodology): “We train NICE on MNIST (LeCun and Cortes, 1998), the Toronto Face Dataset 3 (TFD) (Susskind et al., 2010), the Street View House Numbers dataset (SVHN) (Netzer et al., 2011) and CIFAR-10 (Krizhevsky, 2010).”
Score-based Generative Modeling in Latent Space (2021, methodology): “Although this baseline achieves competitive NLL, its sample quality is behind our LSGM and the original SGM. Modeling binarized images: Next, we examine LSGM on dynamically binarized MNIST [93] and OMNIGLOT [74].”
Dual Contradistinctive Generative Autoencoder (2020, neutral): “We train our model on the MNIST dataset [43] and measure the transferability though a classiﬁcation task on the latent embedding vector.”
NVAE: A Deep Hierarchical Variational Autoencoder (2020, methodology): “We examine NVAE on the dynamically binarized MNIST [72], CIFAR-10 [73], ImageNet 32×32 [74], CelebA 64 × 64 [75, 76], CelebA HQ 256×256 [28], and FFHQ 256×256 [77] datasets.”
LVIS: A Dataset for Large Vocabulary Instance Segmentation (2019, methodology): “We owe much of our current success in image recognition to pioneering datasets such as MNIST [16], BSDS [20], Caltech 101 [6], PASCAL VOC [5], ImageNet [23], and COCO [18].”