pranshushah / binary_classification_python_numpy Goto Github PK

python implementation of binary classification from scratch. i used numpy as only dependency.in this repo we are going to use the Pima Indians onset of diabetes dataset. you can download dataset from data.csv. It describes patient medical record data for Pima Indians and whether they had an onset of diabetes within five years.

in this file I'm using decoupled neural interfaces using synthetic gradients to train the classifier.note that in this classifier i am not using any regularization technique. you can use dropout, batch_norm or(L1, L2) regularization technique, for dropout you can use nn.py as reference

in this file i'm using fully connected layer for binary classification, implemented dropout as regularization technique and sigmoid as activation function. you can also use relu in hidden layer for relu activation use dni.py as refrence

this csv file is data set for our binary classification.last column in csv file is labels of the inputs