MNIST ("Modified National Institute of Standards and Technology") is the de facto “hello world” dataset of computer vision. Since its release in 1999, this classic dataset of handwritten images has served as the basis for benchmarking classification algorithms. As new machine learning techniques emerge, MNIST remains a reliable resource for researchers and learners alike.
Results
You can find more details about the results here.
Do not forget to upvote if you like it. ;) Thanks.
Author:
Links:
- Digit Recognizer Competition
- Kaggle Kernel | Step-By-Step: Improving CNN (0.98414 -> 0.99714)
- Yann LeCun's MINST database
Licence :
MIT License. See license.txt