- Use prepare_data() method to prepare the inputs and labels. The prepare data accepts three parameters
#@files - input file to read
#@indices - tuple of indices to fetch from the file
#@labls - dictionary of labels and their corresponding category value for encoding labels
example usage : data1,label1 = prepare_data('train.data', indices = (0,80), labls = {"class-1": -1, "class-2": 1})
- Create an object of Perceptron class. The constructor recieves two parameters
#@@ __init__ - params : {input_size,gamma}
####@ input_size - input size
####@ gamma - Coeficient of the l2 regularization default is 0 (no regularisation)
- Implement a binary Perceptron
######### Prepare the data ###########
data1,label1 = prepare_data('train.data',(0,80), labls = {"class-1": -1, "class-2": 1})
Perceptron1 = Perceptron(4) ############ Constructor with 4 input size ################
Perceptron1.train(data1,label1,20) ############ Train the Perceptron ##############
print(Perceptron1.data_matrix(data1,label1)) ############ View the data matrix ##############
print(Perceptron1.activation(Perceptron1.predict(inputs = [1,2,3,4]))) ###### Prediction of new data point #####
- Multiple classes
data1,label1 = prepare_data('train.data',(0,120), labls = {"class-1": 1, "class-2": -1, "class-3": -1})
Perceptron1 = Perceptron(4) ############ Constructor with 4 input size ################
Perceptron1.train(data1,label1,20) ############ Train the Perceptron ##############
print(Perceptron1.data_matrix(data1,label1)) ############ View the data matrix ##############
- Perceptron with l2 Regularisation
data1,label1 = prepare_data('train.data',(0,120), labls = {"class-1": 1, "class-2": -1, "class-3": -1})
Perceptron1 = Perceptron(4, gamma = 0.01) ############ Constructor with 4 input size ################
Perceptron1.train(data1,label1,20) ############ Train the Perceptron ##############
print(Perceptron1.data_matrix(data1,label1)) ############ View the data matrix ##############