En el contexto de aprendizaje automático and inteligencia artificial, logits refer to the capa de salida of a red neuronal before any función de activación, such as softmax or sigmoid, is applied. These raw scores represent the model’s confidence for each class in classification tasks. The term is commonly used in the field of deep learning, particularly in clasificación multiclase problemas.
Cuando una red neuronal procesa datos de entrada, produce un conjunto de logits que pueden considerarse como puntuaciones de valores reales sin límites. Por ejemplo, en una tarea de clasificación con tres clases, la red podría generar logits como [2.0, -1.5, 0.5]. Estos valores no representan probabilidades; pueden ser positivos o negativos, y no existe una relación inherente entre ellos.
To convert these logits into probabilities that sum to one—making them interpretable as the likelihood of each class occurring—we typically apply the softmax function. The softmax function exponentiates each logit and normalizes the results to produce a probability distribution. In our example, applying softmax to the logits [2.0, -1.5, 0.5] will yield a set of probabilities that indicates how confident the model is about each class.
Logits are essential for understanding how a model makes predictions, as they provide insight into the decision-making process of the network. By analyzing the logits, researchers and practitioners can better diagnose model behavior, tune hyperparameters, and improve y fiabilidad de los servicios modernos de telecomunicaciones y datos..