Pattern recognition with block-based neural networks
Sangwoo Moon, Seong-Gon Kong · 2003
This paper presents block-based neural net works (BbNNs) for pattern classification. The BbNN achieves two goals: simultaneous optimization of network structure/weights; and implementation using reconfigurable digital hardware. The BbNN, in a 2-D array of basic blocks with four variable input/output nodes, successfully solve pattern classification problems. Internal structure is optimized using the GA with 2-D encoding scheme.