Digit Recognizer is a Kaggle based competition. In this competition, goal is to correctly identify digits from a dataset of tens of thousands of handwritten images.
With the below methods I have held a rank of [276/1479] among the top 19%
The goal in this competition is to take an image of a handwritten single digit, and determine what that digit is. The data for this competition were taken from the MNIST dataset.
• Implemented different machine learning algorithms (Random Forest, Support Vector Machine, Gradient Boosted Method) to improve the accuracy of model. • Final model submitted is an CNN which resulting in 97.67% training and 95.32% validation accuracy, the performance was boosted using Stochastic gradient descent optimizer and hyper tuning of parameter to 99.77% training and 99.59% validation accuracy. Model achieved 99.04 % accuracy on the test dataset surpassing the previous model with accuracy of 97.51%.
Dataset : 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.
- Accuracy Graph
- Layer wise visualization
- Model architecture