• Corpus ID: 702279

Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms

@article{Xiao2017FashionMNISTAN,
  title={Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms},
  author={Han Xiao and Kashif Rasul and Roland Vollgraf},
  journal={ArXiv},
  year={2017},
  volume={abs/1708.07747}
}
We present Fashion-MNIST, a new dataset comprising of 28x28 grayscale images of 70,000 fashion products from 10 categories, with 7,000 images per category. The training set has 60,000 images and the test set has 10,000 images. Fashion-MNIST is intended to serve as a direct drop-in replacement for the original MNIST dataset for benchmarking machine learning algorithms, as it shares the same image size, data format and the structure of training and testing splits. The dataset is freely available… 

Figures and Tables from this paper

Oracle-MNIST: a Realistic Image Dataset for Benchmarking Machine Learning Algorithms

We introduce the Oracle-MNIST dataset, comprising of 28 × 28 grayscale images of 30,222 ancient characters from 10 categories, for benchmarking pattern classification, with particular challenges on

Training on test data: Removing near duplicates in Fashion-MNIST

TLDR
This paper identifies near-duplicate images in Fashion MNIST and proposes a dataset withNear duplicate images between training and testing sets to increase the data quality of FashionMNIST.

Sequential Graph Convolutional Network for Active Learning (Supplementary Material)

TLDR
A extended description of the datasets used to evaluate the algorithms and the compared baselines of the methods evaluated together with the others on four challenging image classification benchmarks: CIFAR10, CIFar-100, FashionMNIST and SVHN.

CNN Model for Image Classification on MNIST and Fashion-MNIST Dataset

TLDR
Review of obtained results and from literature shows that CNN is suitable for image classification for MNIST and Fashion-MNIST dataset.

Image Classification of Fashion-MNIST Dataset Using Long Short-Term Memory Networks

TLDR
The Long Short-Term Memory Networks (LSTMs) are used to build a model which can operate the image classification of Fashion-MNIST dataset, which includes 70000 grayscale images whose size is 28*28.

Pre-convoluted neural networks for fashion classification

TLDR
The pre-convolution neural network with the five layered convoluted deep neural network and the augmented data shows 94% accuracy where it was 93% in VGG16 and 92% in AlexNetnets.

Learning with Dynamic Architectures for Artificial Neural Networks - Adaptive Batch Size Approach

TLDR
The performance of ADANET framework is explored by using custom search space for an image-classification dataset using tensorflow libraries in combination with adaptive batch sizes for learning using simple deep neural network and also with convolutional neural network.

Classification of Fashion Images Using Transfer Learning

TLDR
This study utilizes the feature concatenation property, knowledge consolidation and inductive feature transfer of DenseNet architecture for categorization of fashion MNIST dataset and indicates that the proposed system outperforms the existing architectures.

T RAINING ON TEST DATA : R EMOVING NEAR-DUPLICATES IN F ASHION-MNIST

TLDR
This paper identifies near-duplicate images in Fashion-MNIST and proposes a dataset withNear duplicate images between training and testing sets to artificially increase the testing accuracy of machine learning models.

Overhead-MNIST: Machine Learning Baselines for Image Classification

TLDR
The Overhead-MNIST dataset is a collection of satellite images similar in style to the ubiquitous MNIST hand-written digits found in the machine learning literature, and a convolutional neural network was selected as a baseline for edge deployability and future performance improvement.
...

References

SHOWING 1-6 OF 6 REFERENCES

Learning Multiple Layers of Features from Tiny Images

TLDR
It is shown how to train a multi-layer generative model that learns to extract meaningful features which resemble those found in the human visual cortex, using a novel parallelization algorithm to distribute the work among multiple machines connected on a network.

EMNIST: an extension of MNIST to handwritten letters

TLDR
A variant of the full NIST dataset is introduced, which is called Extended MNIST (EMNIST), which follows the same conversion paradigm used to create the MNIST dataset, and shares the same image structure and parameters as the original MNIST task, allowing for direct compatibility with all existing classifiers and systems.

Multi-column deep neural networks for image classification

TLDR
On the very competitive MNIST handwriting benchmark, this method is the first to achieve near-human performance and improves the state-of-the-art on a plethora of common image classification benchmarks.

ImageNet: A large-scale hierarchical image database

TLDR
A new database called “ImageNet” is introduced, a large-scale ontology of images built upon the backbone of the WordNet structure, much larger in scale and diversity and much more accurate than the current image datasets.

Regularization of Neural Networks using DropConnect

TLDR
This work introduces DropConnect, a generalization of Dropout, for regularizing large fully-connected layers within neural networks, and derives a bound on the generalization performance of both Dropout and DropConnect.

Gradient-based learning applied to document recognition

TLDR
This paper reviews various methods applied to handwritten character recognition and compares them on a standard handwritten digit recognition task, and Convolutional neural networks are shown to outperform all other techniques.