A convolutional neural network ends with a fully connected layer of 10 outputs representing 10 animal classes. Which function should be applied to that layer so the model outputs a probability distribution across the 10 classes?

Softmax is the correct choice because it transforms a vector of arbitrary real numbers into a probability distribution, where each value is between 0 and 1, and all values sum to 1. This is ideal for multi-class classification problems like identifying 10 animal classes, as it directly provides the probability of each class. Dropout is a regularization technique to prevent overfitting, not an activation function for output probabilities. Smooth L1 loss is a loss function used during training, not an activation function applied to the output layer. ReLU is an activation function typically used in hidden layers to introduce non-linearity, but it does not produce a probability distribution.