Intrinsic Embedded Hardware Evolution of Block-based Neural Networks

- An intrinsic embedded online evolution system has been designed using Block-based neural networks and implemented on Xilinx VirtexIIPro FPGAs. The designed network can dynamically adapt its structure and parameters to input data pattern variations without any FPGA reconfiguration overheads, overcoming a major bottleneck for online evolution systems. With increasing speeds of silicon hardware and availability of faster embedded processors in the near future, a wide spread deployment of these platforms is becoming possible.

• Type: Conference
• Year: 2006
• Where: ERSA
• Authors: Saumil Merchant, Gregory D. Peterson, Seong Kong
• Keywords: Block-based Neural Networks, ERSA 2006, ERSA 2007, Intrinsic Embedded Online, Online Evolution