GitHub - anuraganands/GSGD: Guided Stochastic Gradient Descent Algorithm (GSGD)

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 31 Commits
+Structure		+Structure
Result		Result
data		data
GSGDv1.m		GSGDv1.m
GSGDv2.m		GSGDv2.m
PrintFinalResults.m		PrintFinalResults.m
ReadMe.txt		ReadMe.txt
SGDvariation.m		SGDvariation.m
collectInconsistentInstances.m		collectInconsistentInstances.m
extractConsistentInstances.m		extractConsistentInstances.m
getAllalgosNames.m		getAllalgosNames.m
getBGD.m		getBGD.m
getError.m		getError.m
getNoiseScoreThreshold.m		getNoiseScoreThreshold.m
getSGD.m		getSGD.m
get_data.m		get_data.m
legappend.m		legappend.m
loadMNISTImages.m		loadMNISTImages.m
loadMNISTLabels.m		loadMNISTLabels.m
makeCrossValidationSets.m		makeCrossValidationSets.m
noiseSorting.m		noiseSorting.m
norm_scale01.m		norm_scale01.m
norm_variance.m		norm_variance.m
readData.m		readData.m
updateLegend.m		updateLegend.m
validate.m		validate.m

Repository files navigation

Guided Stochastic Gradient Descent (GSGD)

Copyright (c) 2018, Anuraganand Sharma - All rights reserved.

This Matlab code is the implementation of GSGD proposed by Anuraganand Sharma in the paper:
A. Sharma, “Guided Stochastic Gradient Descent Algorithm for inconsistent datasets,” 
Applied Soft Computing, vol. 73, pp. 1068–1080, Dec. 2018 

How to run:
1 - The main program is written in GSGD.m. Run this program.
2 - It asks for data files. By default "data" folder is in the same directory as GSGD.m
3 - The results are stored in "Result" folder.
4 - For further analysis run Result\analyseData.m and Result\colorMapTry.m