I found the code of this multiclass perceptron online. I can't remember the source. I have not read enough machine learning to be able to inspect it myself but I am wondering if the code is correct or not. I am posting the entire example below. (If I'm not mistaken, the code may have been generated from Microsoft Copilot.) import numpy as np class MultiClassPerceptron: def __init__(self, input_dim, num_classes, learning_rate=0.01, epochs=100): self.weights = np.zeros((num_classes, input_dim)) self.bias = np.zeros(num_classes) self.learning_rate = learning_rate self.epochs = epochs def predict(self, X): scores = np.dot(X, self.weights.T) + self.bias return np.argmax(scores, axis=1) def fit(self, X, y): for epoch in range(self.epochs): for i in range(X.shape[0]): x_i = X y_true = y y_pred = self.predict(x_i.reshape(1, -1))[0] if y_pred != y_true: self.weights[y_true] += self.learning_rate * x_i self.weights[y_pred] -= self.learning_rate * x_i self.bias[y_true] += self.learning_rate self.bias[y_pred] -= self.learning_rate # Example usage if __name__ == "__main__": # Sample dataset (X: features, y: labels) X = np.array([[2, 1], [1, 3], [2, 3], [3, 2], [4, 1], [3, 4]]) y = np.array([0, 1, 1, 0, 0, 1]) # Labels for two classes (0 and 1) # Initialize and train the perceptron perceptron = MultiClassPerceptron(input_dim=2, num_classes=2, learning_rate=0.1, epochs=10) perceptron.fit(X, y) # Predict on new data test_data = np.array([[3, 3], [1, 2]]) predictions = perceptron.predict(test_data) print("Predictions:", predictions) Continue reading...
