CuPit-2: a portable parallel programming language for artificial neural networks

CuPit-2 is a programming language specifically designed to express neural network learning algorithms. It provides most of the flexibility of general-purpose languages like C/C++, but results in much clearer and more elegant programs due to higher expressiveness, in particular for algorithms that change the network topology dynamically (constructive algorithms, pruning algorithms). Furthermore, CuPit-2 programs can be compiled into efficient code for parallel machines; no changes are required in the source program. This article presents a description of the language constructs and reports performance results for an implementation of CuPit-2 on symmetric multiprocessors (SMPs).

[1]  Martin A. Riedmiller,et al.  A direct adaptive method for faster backpropagation learning: the RPROP algorithm , 1993, IEEE International Conference on Neural Networks.

[2]  Alfredo Weitzenfeld,et al.  NSL: neural simulation language , 1998 .

[3]  Christian Jacob,et al.  The NeuroGraph Neural Network Simulator , 1993, IEEE/ACM International Symposium on Modeling, Analysis, and Simulation On Computer and Telecommunication Systems.

[4]  Holger Hopp,et al.  CuPit-2 - a parallel language for neural algorithms: language reference and tutorial , 1994 .

[5]  A. Linden,et al.  Combining multiple neural network paradigms and applications using SESAME , 1992, [Proceedings 1992] IJCNN International Joint Conference on Neural Networks.

[6]  Andrew A. Chien,et al.  Concurrent Aggregates: Supporting Modularity in Massively Parallel Programs , 1993 .