EMNIST: an extension of MNIST to handwritten letters

Abstract

The MNIST dataset has become a standard benchmark for learning, classification and computer vision systems. Contributing to its widespread adoption are the understandable and intuitive nature of the task, its relatively small size and storage requirements and the accessibility and ease-of-use of the database itself. The MNIST database was derived from a larger dataset known as the NIST Special Database 19 which contains digits, uppercase and lowercase handwritten letters. This paper introduces a variant of the full NIST dataset, which we have called Extended MNIST (EMNIST), which follows the same conversion paradigm used to create the MNIST dataset. The result is a set of datasets that constitute a more challenging classification tasks involving letters and digits, and that shares the same image structure and parameters as the original MNIST task, allowing for direct compatibility with all existing classifiers and systems. Benchmark results are presented along with a validation of the conversion process through the comparison of the classification results on converted NIST digits and the MNIST digits.

The dataset is now available for download from https://www.westernsydney.edu.au/bens/home/reproducible_research/emnist. This link is also included in the revised article

Cited by in corpus (93)

• Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
• Decentralized Federated Learning: Fundamentals, State of the Art, Frameworks, Trends, and Challenges
• FedML: A Research Library and Benchmark for Federated Machine Learning
• Spectrally-Encoded Single-Pixel Machine Vision Using Diffractive Networks
• Ditto: Fair and Robust Federated Learning Through Personalization
• Expanding the Reach of Federated Learning by Reducing Client Resource Requirements
• Federated Learning for Computationally-Constrained Heterogeneous Devices: A Survey
• FedMix: Approximation of Mixup under Mean Augmented Federated Learning
• Diffusion Schrödinger Bridge with Applications to Score-Based Generative Modeling
• Unseen Class Discovery in Open-world Classification
• Disentanglement by Nonlinear ICA with General Incompressible-flow Networks (GIN)
• From the digital data revolution to digital health and digital economy toward a digital society: Pervasiveness of Artificial Intelligence
• Out-of-distribution Detection in Classifiers via Generation
• Why is the Mahalanobis Distance Effective for Anomaly Detection?
• FedScale: Benchmarking Model and System Performance of Federated Learning at Scale
• Probabilistic Modeling of Deep Features for Out-of-Distribution and Adversarial Detection
• SEALion: a Framework for Neural Network Inference on Encrypted Data
• FLrce: Resource-Efficient Federated Learning with Early-Stopping Strategy
• Learning Generative Models across Incomparable Spaces
• Enhancing the Privacy of Federated Learning with Sketching
• Dual Manifold Adversarial Robustness: Defense against Lp and non-Lp Adversarial Attacks
• Continuous learning of spiking networks trained with local rules
• Flexible Dataset Distillation: Learn Labels Instead of Images
• Federated Learning with Superquantile Aggregation for Heterogeneous Data
• Robust Watermarking of Neural Network with Exponential Weighting
• Communication-Efficient Federated Distillation
• Deep Amortized Clustering
• What Do We Mean by Generalization in Federated Learning?
• Towards Self-Adaptive Metric Learning On the Fly
• Efficient and Private Federated Learning with Partially Trainable Networks
• Learning Optimal Conformal Classifiers
• Fast Real-time Counterfactual Explanations
• Heterogeneous Data-Aware Federated Learning
• A New Loss Function for CNN Classifier Based on Pre-defined Evenly-Distributed Class Centroids
• Adaptive Federated Dropout: Improving Communication Efficiency and Generalization for Federated Learning
• Bootstrapping Neural Processes
• Federated Learning on Non-IID Data: A Survey
• Fluctuation-dissipation Type Theorem in Stochastic Linear Learning
• An optical diffractive deep neural network with multiple frequency-channels
• Meta-learning algorithms for Few-Shot Computer Vision
• Evaluating State-of-the-Art Classification Models Against Bayes Optimality
• Privacy-preserving Decentralized Aggregation for Federated Learning
• AutoAssist: A Framework to Accelerate Training of Deep Neural Networks
• Information Theoretic Meta Learning with Gaussian Processes
• A needle-based deep-neural-network camera
• BaCOUn: Bayesian Classifers with Out-of-Distribution Uncertainty
• Learning with Algorithmic Supervision via Continuous Relaxations
• Assessing Intelligence in Artificial Neural Networks
• Automating Crystal-Structure Phase Mapping: Combining Deep Learning with Constraint Reasoning
• SimEx: Express Prediction of Inter-dataset Similarity by a Fleet of Autoencoders
• Efficient and Less Centralized Federated Learning
• DeltaGAN: Towards Diverse Few-shot Image Generation with Sample-Specific Delta
• Artificial Liver Classifier: A New Alternative to Conventional Machine Learning Models
• Augmentation-Interpolative AutoEncoders for Unsupervised Few-Shot Image Generation
• Effectiveness of Optimization Algorithms in Deep Image Classification
• You Only Query Once: Effective Black Box Adversarial Attacks with Minimal Repeated Queries
• One for One, or All for All: Equilibria and Optimality of Collaboration in Federated Learning
• Application of Quantum Pre-Processing Filter for Binary Image Classification with Small Samples
• F2GAN: Fusing-and-Filling GAN for Few-shot Image Generation
• Source-Free Adaptation to Measurement Shift via Bottom-Up Feature Restoration
• A Deep Learning Framework for Lifelong Machine Learning
• Learning Translation Invariance in CNNs
• Large-Scale Semi-Supervised Learning via Graph Structure Learning over High-Dense Points
• Representation of Federated Learning via Worst-Case Robust Optimization Theory
• Bespoke vs. Prêt-à-Porter Lottery Tickets: Exploiting Mask Similarity for Trainable Sub-Network Finding
• Schrödinger's Camera: First Steps Towards a Quantum-Based Privacy Preserving Camera
• Information Condensing Active Learning
• Conditional Coupled Generative Adversarial Networks for Zero-Shot Domain Adaptation
• Verification of Neural Networks: Specifying Global Robustness using Generative Models
• Meta-Learning Bidirectional Update Rules
• Inherent Weight Normalization in Stochastic Neural Networks
• Estimating informativeness of samples with Smooth Unique Information
• Reward-Based 1-bit Compressed Federated Distillation on Blockchain
• Domain2Vec: Domain Embedding for Unsupervised Domain Adaptation
• Improved Robustness to Open Set Inputs via Tempered Mixup
• Adversarial Learning for Zero-shot Domain Adaptation
• ADDS: Adaptive Differentiable Sampling for Robust Multi-Party Learning
• The Compact Support Neural Network
• Learning with Collaborative Neural Network Group by Reflection
• Learning from Similarity-Confidence Data
• The Uncanny Similarity of Recurrence and Depth
• Robust Federated Learning by Mixture of Experts
• Task Fingerprinting for Meta Learning in Biomedical Image Analysis
• Enabling Highly Efficient Capsule Networks Processing Through A PIM-Based Architecture Design
• Canoe : A System for Collaborative Learning for Neural Nets
• Widely Linear Kernels for Complex-Valued Kernel Activation Functions
• Max and Coincidence Neurons in Neural Networks
• Few-shot learning using pre-training and shots, enriched by pre-trained samples
• Smart Rewritings of the Basic Equations for Quantitative Non-Linear Inverse Scattering
• CatFedAvg: Optimising Communication-efficiency and Classification Accuracy in Federated Learning
• Consistency of Extreme Learning Machines and Regression under Non-Stationarity and Dependence for ML-Enhanced Moving Objects
• ImitAL: Learning Active Learning Strategies from Synthetic Data
• A Biologically Plausible Audio-Visual Integration Model for Continual Learning