CMU Artificial Intelligence Repository
Home INFO Search FAQs Repository Root

GRADSIM: Connectionist Network Optimization Package

GRADSIM is a connectionist network optimization package, a neural network simulator that supports recurrent time-delay network optimization. GRADSIM was designed for simulation experiments with the temporal flow model and was developed for research in speech recognition. The temporal flow model is characterized by delay links and unrestricted network connectivity. The simulator accepts network descriptors and experiment descriptors in a very simple format, and efficiently computes the complete gradient of recurrent networks. It can be configured for several gradient descent methods.

Version: 2.0 (8-OCT-90) Requires: C Copying: Copyright (C) 1988, Trustees of the University of Pennsylvania. Copyright (C) 1990, Trustees of the University of Toronto. Use, copying, and distribution permitted, provided it is not sold for profit. CD-ROM: Prime Time Freeware for AI, Issue 1-1 Author(s): Raymond Watrous Siemens Corporate Research 755 College Road East Princeton, NJ 08540 Tel: (609) 734-6596 Keywords: Authors!Watrous, GRADSIM, Gradient Descent, Machine Learning!Neural Networks, Neural Networks!Simulators, Recurrent Neural Networks, Speech Recognition, Temporal Flow Model, Time Delay Neural Networks, Univ. of Pennsylvania, Univ. of Toronto References: ?
Last Web update on Mon Feb 13 10:25:20 1995