交叉熵损失函数和均方误差损失函数的代码实现

交叉熵损失函数的代码实现：

import numpy as np

def cross_entropy_loss(y_pred, y_true):
    """
    Computes the cross-entropy loss between predicted and true labels.
    
    Args:
    - y_pred: numpy array of predicted probabilities (shape: batch_size x num_classes)
    - y_true: numpy array of true labels (shape: batch_size x num_classes)
    
    Returns:
    - cross-entropy loss: scalar value
    """
    epsilon = 1e-12 # small value to avoid division by zero
    num_examples = y_pred.shape[0] # number of examples in the batch
    
    # clip predicted values to avoid log(0) errors
    y_pred = np.clip(y_pred, epsilon, 1. - epsilon)
    
    # compute cross-entropy loss
    ce_loss = -np.sum(y_true * np.log(y_pred)) / num_examples
    
    return ce_loss

均方误差损失函数的代码实现：

import numpy as np

def mse_loss(y_pred, y_true):
    """
    Computes the mean squared error loss between predicted and true labels.
    
    Args:
    - y_pred: numpy array of predicted values (shape: batch_size x num_classes)
    - y_true: numpy array of true labels (shape: batch_size x num_classes)
    
    Returns:
    - mean squared error loss: scalar value
    """
    num_examples = y_pred.shape[0] # number of examples in the batch
    
    # compute mean squared error loss
    mse_loss = np.sum(np.square(y_pred - y_true)) / (2 * num_examples)
    
    return mse_loss