Please modify this binary Perceptron implementation to be One versus All Perceptron for multiclass problem.

import numpy as np a = def perceptron(data, step_size=0.01, n=200): weights = np.zeros(len(data[0])) error_list = list for i in range(n): error = 0.0 for line in data: weights[0] for k in range (len(line) - 1): a += weights[k + 1] * line[k] if a >= : d = 1 else: d = 0 line[-1] - d error + error**2 for num in range(len(line) - 1): weights[num + 1] weights[num + 1] + step_size * error * line[num] error_list.append(error) return weights, error_list error = = import numpy as np a = def perceptron(data, step_size=0.01, n=200): weights = np.zeros(len(data[0])) error_list = list for i in range(n): error = 0.0 for line in data: weights[0] for k in range (len(line) - 1): a += weights[k + 1] * line[k] if a >= : d = 1 else: d = 0 line[-1] - d error + error**2 for num in range(len(line) - 1): weights[num + 1] weights[num + 1] + step_size * error * line[num] error_list.append(error) return weights, error_list error = =