In a multi-class classification neural network, what is the primary role of the softmax activation function when applied to the output layer?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Check this out: when your network finishes calculating everything for a multi-class problem (like identifying if an image is a cat, dog, or bird), it outputs raw scores called logits. Let's say the logits are 2.0, 1.0, and 0.1. Those numbers don't look like probabilities! To fix this, we run them through the softmax function. Softmax squashes each number so it's between 0 and 1, and it makes sure all three outputs add up to exactly 1.0 (or 100%). Now you can tell your boss, 'We are 70% sure this is a dog.' Option C is the correct answer here.
Full explanation below image
Full Explanation
In deep learning, the final layer of a neural network designed for multi-class classification generates a vector of raw, unnormalized real values called logits. The softmax function is applied to these logits to transform them into a probability distribution. Mathematically, for a vector z of length K, the softmax function converts the logits using exponentials normalized by the sum of exponentials of all logits. The exponential function ensures that all resulting outputs are positive. Dividing by the sum of exponentials normalizes the vector such that each value falls within the range (0, 1) and the sum of all values is exactly 1. This allows the network's output to be interpreted as a probability distribution over the K classes, where the class with the highest probability is selected as the prediction. Let's look at why the other options are wrong: Option A refers to feature scaling or input normalization, which is done during preprocessing, not in the output layer. Option B refers to the Sigmoid function, which is used for binary classification tasks (mapping a single output to a value between 0 and 1) rather than multi-class probability vectors. Option C is the correct definition. Option D refers to the linear layers (dense/fully connected layers) themselves, which perform linear transformations before the activation function is applied. For the exam, remember that Softmax is the standard for multi-class classification because it outputs a probability distribution summing to 1.