Software engineering methods for neural networks

摘要: Neural networks have been used to solve a wide range of problems. Unfortunately, many the applications neural reported in literature built an ad-hoc manner, without being informed by techniques and tools software engineering. The problem with developing using "trial error" or "build fix" approach, is that successes are difficult repeat. Building specific problems processes repeatable only if there sufficient culture disciplined practice experienced people organisation facilitate process. We propose set methods for can be systematically repeatably "engineer" explore "design "for networks, validating verifying operation learning algorithms network software. A feature our approach separate generic components from application components.