Lera G, Pinzolas M
Dept. Automatica y Computacion, Univ. Publica de Navarra, Pamplona, Spain.
IEEE Trans Neural Netw. 2002;13(5):1200-3. doi: 10.1109/TNN.2002.1031951.
Although the Levenberg-Marquardt (LM) algorithm has been extensively applied as a neural-network training method, it suffers from being very expensive, both in memory and number of operations required, when the network to be trained has a significant number of adaptive weights. In this paper, the behavior of a recently proposed variation of this algorithm is studied. This new method is based on the application of the concept of neural neighborhoods to the LM algorithm. It is shown that, by performing an LM step on a single neighborhood at each training iteration, not only significant savings in memory occupation and computing effort are obtained, but also, the overall performance of the LM method can be increased.
尽管Levenberg-Marquardt（LM）算法作为一种神经网络训练方法已被广泛应用，但当待训练的网络具有大量自适应权重时，它在内存和所需操作数量方面都非常昂贵。本文研究了最近提出的该算法变体的性能。这种新方法基于将神经邻域概念应用于LM算法。结果表明，通过在每次训练迭代中对单个邻域执行一次LM步长，不仅可以显著节省内存占用和计算量，而且还可以提高LM方法的整体性能。
