binary_swarm_intelligence

Library of binary swarm intelligence mainly used for obtaining optimal solution of feature selection

The implemented algorithm

• Binary Genetic Algorithm
• Binary Particle Swarm optimization
• Binary Cuckoo Search
• Binary Firefly algorithm
• Binary Bat Algorithm
• Binary Gravitational Search algorithm
• Binary Dragon Fly Algorithm

Example to use

1. Import library.
2. Set evaluation function in class.
3. Specify the class as the argument of the algorithm.
4. If possible parameter setting.

Example code: Suppose svm is used

import binary_optimization as opt#import library
import numpy as np
from sklearn import svm

class Evaluate:#setting class
    def __init__(self):#set train_data,train_label
        self.tr_x = tr_x
        self.tr_y = tr_y
        self.K = 4
    def evaluate(self,gen):
        """
        Setting of evaluation function.
        Here, the correct answer rate is used.
          anser_label/all_label
        """
        mask = np.array(gen) > 0
        al_data = tr_x[:,np.nonzero(mask)[0]]
        kf = ms.KFold(n_split=self.K, shuffle=True);s=0
        for tr_ix, te_ix in kf.split(self.tr_x):          
          s += svm.LinearSVC().fit(al_data[tr_ix],self.tr_y[tr_ix]).score(al_data[te_ix], self.tr_y[te_ix])
        return s/self.K
        #↑evaluate with fittness function
    def check_dimentions(self,dim):#check number of all feature
        if dim==None:
            return len(self.train_d[0])
        else:
            return dim

print("Algorithm:\n\t{0}  {1} {2}".format("best_pos","best_val","number_of_1s"))

s,g,l=opt.BGA(Eval_Func=Evaluate, n=20, m_i=200)#score, gen_list, gen length of 1
print("BGA:\n\t{0}   {1}  {2}".format("".join(map(str,g)),s,l))

common arguments with algorithms

• Eval_Func: Evaluate function (class)
• n: number of population (int)
• m_i: number of max iteration(int)
• dim: number of all feature(int)
• minf: minimization flag. min or max?(bool)
• prog: Do you use progress bar?(bool)