A Deep Neural Network Augmented Approach for Fixed Polarity AND-XOR Network Synthesis
Autor: | Hillol Maity, Parthajit Bhattacharya, Girish Patankar, Kaushik Khatua, Santanu Chattopadhyay, Indranil Sengupta |
---|---|
Rok vydání: | 2019 |
Předmět: |
Optimization problem
Artificial neural network Computer science Particle swarm optimization 02 engineering and technology 010501 environmental sciences 01 natural sciences 020202 computer hardware & architecture Computer engineering 0202 electrical engineering electronic engineering information engineering Benchmark (computing) Network synthesis filters Field-programmable gate array Testability Hardware_LOGICDESIGN 0105 earth and related environmental sciences Electronic circuit |
Zdroj: | TENCON |
Popis: | 11This work is partially supported by the research project sponsored by the Synopsys Inc., USAWith the recent advancements of FPGA (Field Programmable Gate Array), circuits in AND-XOR plane gets its fair share of advantages due to the high testability feature of the AND-XOR networks and independence of the logic-gate area as well as delays on FPGA. Minimization of the product terms for such networks is an NP-hard problem. In this paper, we have proposed a Binary Particle Swarm Optimization (BPSO) based technique to solve the optimization problem accurately and accelerate the same using a Deep Neural Network. With the proposed technique, after testing it against various MCNC benchmark circuits, the results were very promising in terms of product terms, while utilizing significantly lesser CPU-time. |
Databáze: | OpenAIRE |
Externí odkaz: |