Perda de Círculo
Circle Loss é uma especializada função de perda designed to enhance the quality of embeddings in various aprendizado de máquina tasks, particularly in classification problems. It was introduced to address the challenges faced when using traditional funções de perda, such as Perda de Entropia Cruzada, especially in scenarios involving conjuntos de dados desequilibrados ou quando as classes estão intimamente relacionadas.
The main objective of Circle Loss is to minimize the distances between embeddings of samples from the same class while maximizing the distances between embeddings of samples from different classes. This is achieved by creating a ‘circle’ in the espaço de incorporação, where each class is represented as a point. The loss function encourages the embeddings to be positioned around a central point, forming a circular structure that enhances class separation.
Circle Loss operates by utilizing two key components: the angular margin and the radius. The angular margin is a parameter that controls the separation between different classes, while the radius determines the size of the circle in the embedding space. By adjusting these parameters, Circle Loss can be fine-tuned for various applications, making it a versatile choice for tasks such as face recognition, object detection, and reconhecimento de fala.
In summary, Circle Loss provides an innovative approach to optimizing embeddings in machine learning models, focusing on both intra-class compactness and inter-class separability. This results in improved classification performance, especially in complex cenários onde funções de perda tradicionais podem não ser suficientes.