arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 0710.4182cs.NE
收藏

Beyond Feedforward Models Trained by Backpropagation:A Practical Training Tool for a More Efficient Universal Approximator

Beyond Feedforward Models Trained by Backpropagation: a Practical Training Tool for a More Efficient Universal Approximator

Roman Ilin, Robert Kozma, Paul J. Werbos

摘要:

细胞同步递归神经网络(SRN)已被证明是一个更强大的函数逼近器比MLP。这意味着MLP的复杂性对于某些问题来说是非常大的,而SRN可以在可接受的计算约束下实现所需的映射。复杂递归网络的训练速度是其成功应用的关键。本文通过扩展卡尔曼滤波器(EKF)训练网络,改进了已有的结果。我们实现了一个通用的细胞SRN,并将其应用于解决两个具有挑战性的问题:二维迷宫导航和连通性问题的一个子集。的收敛速度已提高了几个数量级的迷宫导航的情况下,与早期的结果相比,和优越的推广已被证明在连通性的情况下。这种改进的影响进行了讨论。

英文摘要:

Cellular Simultaneous Recurrent Neural Network (SRN) has been shown to be a function approximator more powerful than the MLP. This means that the complexity of MLP would be prohibitively large for some problems while SRN could realize the desired mapping with acceptable computational constraints. The speed of training of complex recurrent networks is crucial to their successful application. Present work improves the previous results by training the network with extended Kalman filter (EKF). We implemented a generic Cellular SRN and applied it for solving two challenging problems: 2D maze navigation and a subset of the connectedness problem. The speed of convergence has been improved by several orders of magnitude in comparison with the earlier results in the case of maze navigation, and superior generalization has been demonstrated in the case of connectedness. The implications of this improvements are discussed.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑