KMNIST
E363690
benchmark dataset
computer vision dataset
handwritten character dataset
image dataset
machine learning dataset
KMNIST is a benchmark image dataset of handwritten Japanese characters (hiragana) designed as a more complex, drop-in replacement for the original MNIST digit dataset.
All labels observed (2)
| Label | Occurrences |
|---|---|
| KMNIST canonical | 1 |
| KMNIST dataset | 1 |
How this entity was disambiguated
This entity first appeared as the object of triple T3507216 — resolving that mention is where its identity was fixed. The disambiguator weighed these candidate entities and picked the highlighted one (or “None”, minting a new entity). This is how homonymy is resolved: the same surface form can point to different entities.
NED1
Entity disambiguation (via context triple)
gpt-5-mini-2025-08-07
Target entity: KMNIST Context triple: [MNIST, inspiredDataset, KMNIST]
-
A.
MNIST
MNIST is a widely used benchmark dataset of handwritten digit images commonly employed for training and evaluating image classification algorithms in machine learning and computer vision.
-
B.
CIFAR
CIFAR (the Canadian Institute for Advanced Research) is a Canadian global research organization that supports long-term, collaborative, interdisciplinary research, including major initiatives in artificial intelligence.
-
C.
LeNet
LeNet is one of the earliest convolutional neural network architectures, pioneering modern deep learning approaches to image recognition and handwritten digit classification.
-
D.
DIGIT
DIGIT is the European Commission’s Directorate‑General responsible for shaping, implementing, and managing the EU institutions’ digital, IT, and cybersecurity strategies and services.
-
E.
KMK
KMK is the central coordinating body of Germany’s state education and cultural ministers, responsible for harmonizing policies across the federal states.
- F. None of above. chosen
- G. Unsure - the case is ambiguous/there is not enough information to decide.
NED2
Entity disambiguation (via description)
gpt-5-mini-2025-08-07
Target entity: KMNIST Target entity description: KMNIST is a benchmark image dataset of handwritten Japanese characters (hiragana) designed as a more complex, drop-in replacement for the original MNIST digit dataset.
-
A.
MNIST
MNIST is a widely used benchmark dataset of handwritten digit images commonly employed for training and evaluating image classification algorithms in machine learning and computer vision.
-
B.
CIFAR
CIFAR (the Canadian Institute for Advanced Research) is a Canadian global research organization that supports long-term, collaborative, interdisciplinary research, including major initiatives in artificial intelligence.
-
C.
LeNet
LeNet is one of the earliest convolutional neural network architectures, pioneering modern deep learning approaches to image recognition and handwritten digit classification.
-
D.
DIGIT
DIGIT is the European Commission’s Directorate‑General responsible for shaping, implementing, and managing the EU institutions’ digital, IT, and cybersecurity strategies and services.
-
E.
KMK
KMK is the central coordinating body of Germany’s state education and cultural ministers, responsible for harmonizing policies across the federal states.
- F. None of above. chosen
Statements (48)
| Predicate | Object |
|---|---|
| instanceOf |
benchmark dataset
ⓘ
computer vision dataset ⓘ handwritten character dataset ⓘ image dataset ⓘ machine learning dataset ⓘ |
| accessMethod | download from the internet ⓘ |
| availability | publicly available ⓘ |
| benchmarkRole | standard benchmark for non-Latin handwritten characters ⓘ |
| category | handwritten character recognition dataset ⓘ |
| characterSet | subset of hiragana ⓘ |
| colorSpace | grayscale ⓘ |
| comparedTo | MNIST ⓘ |
| compatibility | drop-in compatible with MNIST loaders ⓘ |
| complexityRelativeToMNIST | higher ⓘ |
| dataModality | vision ⓘ |
| dataType | grayscale images ⓘ |
| designedAs | more complex drop-in replacement for MNIST ⓘ |
| difficultyLevel | harder than MNIST ⓘ |
| domain | handwritten Japanese characters ⓘ |
| format |
binary files
ⓘ
similar to original MNIST format ⓘ |
| imageChannels | 1 ⓘ |
| imageHeight | 28 ⓘ |
| imageResolution | 28x28 pixels ⓘ |
| imageWidth | 28 ⓘ |
| inputShape | (1, 28, 28) ⓘ |
| inspiredBy | MNIST ⓘ |
| labelType | hiragana character categories ⓘ |
| language | Japanese ⓘ |
| license | permissive research license ⓘ |
| numberOfClasses | 10 ⓘ |
| purpose |
benchmarking image classification algorithms
ⓘ
provide a more complex alternative to MNIST ⓘ serve as a drop-in replacement for MNIST ⓘ |
| researchArea |
computer vision
ⓘ
deep learning ⓘ optical character recognition ⓘ pattern recognition ⓘ |
| script | hiragana ⓘ |
| supports | evaluation of generalization beyond digits ⓘ |
| task | image classification ⓘ |
| typicalUse |
benchmarking representation learning methods
ⓘ
deep learning research ⓘ evaluation of neural network architectures ⓘ supervised learning ⓘ |
| usedIn |
academic research
ⓘ
benchmark studies ⓘ machine learning education ⓘ |
How these facts were elicited
The pipeline generated the facts above by prompting gpt-5.1 with this entity's name + description and the instruction below.
Instruction
You are a knowledge base construction expert. Given a subject entity and a description of it, return factual statements that you know for the subject as a JSON list of dictionaries(triples), where keys must be "subject", "predicate" and "object". The number of facts may be very high, between 25 to 50 or more, for very popular subjects. For less popular subjects, the number of facts can be very low, like 5 or 10. # Requirements - If you don't know the subject at all, return an empty list. - If the subject is not a named entity, return an empty list. - Include at least one triple where predicate is "instanceOf". - Do not get too wordy. - Separate several objects into multiple triples with one object.
Input
Subject: KMNIST Description of subject: KMNIST is a benchmark image dataset of handwritten Japanese characters (hiragana) designed as a more complex, drop-in replacement for the original MNIST digit dataset.
Referenced by (2)
Full triples — surface form annotated when it differs from this entity's canonical label.
linked to: KMNIST