We present a near linear time algorithm for constructing hierarchical nets in finite metric spaces with constant doubling dimension. This data-structure is then applied to obtain...
The Sensitivity-Based Linear Learning Method (SBLLM) is a learning method for two-layer feedforward neural networks, based on sensitivity analysis, that calculates the weights by s...
In this work we study approximation of fuzzy functions on a finite set of distinct points. Two types of approximation are considered, one method based on fuzzy linear programming p...
Abstract. The block ciphers MISTY1 and MISTY2 proposed by Matsui are based on the principle of provable security against differential and linear cryptanalysis. This paper presents...
A method is introduced to learn and represent similarity with linear operators in kernel induced Hilbert spaces. Transferring error bounds for vector valued large-margin classifie...