Aytekin (2022) muestra que una red neuronal con activaciones lineales por tramos, como ReLU, equivale exactamente a un árbol de decisión: cada nodo revisa el signo de un filtro efectivo y cada hoja aplica una función lineal. En su red para y = x², el árbol limpio tiene 5 regiones en vez de 16.
Lee la explicación en español: Redes neuronales como árboles de decisión →Crea tu cuenta gratis para descargar. Por su licencia, la descarga viene del sitio original del paper.
Abstract original
In this manuscript, we show that any neural network with any activation function can be represented as a decision tree. The representation is equivalence and not an approximation, thus keeping the accuracy of the neural network exactly as is. We believe that this work provides better understanding of neural networks and paves the way to tackle their black-box nature. We share equivalent trees of some neural networks and show that besides providing interpretability, tree representation can also achieve some computational advantages for small networks. The analysis holds both for fully connected and convolutional networks, which may or may not also include skip connections and/or normalizations.