Catatan: Terjemahan untuk entri ini sedang dalam peninjauan kualitas, sehingga sebagian konten untuk sementara hanya ditampilkan dalam bahasa Inggris.
Entri ini belum diterjemahkan ke bahasa Anda, sehingga versi aslinya ditampilkan di bawah.
cross entropy
This term is a technical specialization from information theory and machine learning. It describes the average number of bits needed to identify an event from a set of possibilities if a coding scheme is based on an estimated probability distribution rather than the true distribution. In practical application, it serves as a proxy for the divergence between the predicted output of a neural network and the ground truth.
Because it is used as a loss function, the goal during model training is to minimize this value. A lower cross entropy indicates that the predicted distribution is closer to the actual distribution, whereas a high value suggests a significant mismatch. It is almost exclusively used in academic, mathematical, and data science contexts and is rarely encountered in general conversation.
Meanings
A measure from information theory used to quantify the difference between two probability distributions, typically used as a loss function in machine learning to measure the performance of a classification model.
The model's loss was calculated using cross entropy to determine how closely the predicted probability distribution matched the actual labels.