Levenberg–Marquardt multi-classification using hinge loss function

Incorporating higher-order optimization functions, such as Levenberg–Marquardt (LM) have revealed better generalizable solutions for deep learning problems. However, these functions suffer from very large processing time and training complexity especially datasets become large, in multi-view classification problems, where finding global optima is a costly problem. To solve this issue, we develop solution LM-enabled with, to the best of knowledge first-time implementation hinge loss, multiview classification. Hinge loss allows neural network converge faster perform than other logistic or square rates. We prove our method by experimenting with various multiclass challenges varying data size. The empirical results show accuracy rates achieved, highlighting how outperforms all cases, when limited. Our paper presents important relationship between can impact