Position paper on appropriate audio/visual Turing test
NeMLaP3/CoNLL '98 Proceedings of the Joint Conferences on New Methods in Language Processing and Computational Natural Language Learning
Information Sciences: an International Journal
| Hi-index | 0.00 |
Much of computational science simulates the physical world using models that are based on classical first-principles mathematical reasoning. Neural networks and related techniques are showing up more and more as alternatives to formal mathematical models. This issue contains a number of articles that describe how neural computing techniques, including fuzzy methods, are playing a role in computational science and engineering. Neural computing techniques are motivated by biological models but, in most cases, neural computational methods bear little resemblence to real biological systems. They do however provide an interesting and potentially useful class of alternative modeling approaches, using large numbers of simple computational elements--the artificial neurons. Neural networks work by fitting data to a parameterized family of nonlinear functions, so that neural computing has many relationships with optimization and approximation theory among other areas. This issue's theme articles related to neural networks complement the extensive coverage of neurocomputing in the March issue of Computer, the general membership magazine of the IEEE Computer Society.