Plenary lecture 2: a fundamental survey and an efficient soft library for several basic models of artificial neural networks
Nicolae Popoviciu · International Conference on Mathematical methods, Computational techniques and Intelligent systems · 2011
The neural networks algorithms are related with main chapters of neural networks: supervised learning and unsupervised learning. We use some explicit notations for learning sets, called, the input data: I(x,d),x ∈ Rn, d ∈ RM for supervised learning and I(x),x ∈ Rn for unsupervised learning. The input set contains the vectors x = x(t),d = d(t), where t = 1, N. The variable t designates the time when the vector x add/or d arrive in the learning network. All algorithmic are written (coded) in C++ computer language. The algorithm presentation is based on matrix utilization method. This approach (in comparison with element by element method) creates an easier algorithm understanding and makes on easier utilization So, the resulting computer program is more efficient. All algorithms are explained and described in the author's monograph Artificial Neural Networks. Mathemetical Foundation, Algorithms and Applications (Nicolae Popoviciu, Floarea Baicu), Bucharest, PRINTECH Editor, 2009, in romanian language). For each algorithm we have written a C++ computer programs in one or two versions. Each algorithm has a name and the computer program has the same name.