Permalink
Switch branches/tags
Nothing to show
Find file
Fetching contributors…
Cannot retrieve contributors at this time
20 lines (14 sloc) 1.14 KB
* Overview
This project uses the K-Nearest neighbor classifier and Naive Bayes Classifier for digit recognition. Text files containing broken-down images are formatted with:
- 1 line per image.
- Each number on a line represents the pixels greyscale count, from left to right, pixel 1 through pixel 784 (28x28 greyscale images).
- The example "class" (digit) is the last digit in the line.
The project was basically hacked together in a weekend. The source is messy.
* Approach
For K-Nearest neighbor implementation, a euclidean distance formula is used to determine the 'distance' between two images. The Training data is fed into the system, and using this formula we can determine how close one image is to another. Sorting, Ranking and Voting take place after making the right comparisons with the test data to the training data.
- http://en.wikipedia.org/wiki/K-nearest_neighbor_algorithm
For Bayes, a Naive Bayes is used.
- http://en.wikipedia.org/wiki/Naive_Bayes_classifier
* Upcoming / Wanted
1. Single-hidden-layer neural network implementation.
2. Shape matching using OpenCV (Java interfaces to OpenCV library do not yet exist)