Pitfalls and Best Practices in Algorithm Configuration
This work helps researchers and practitioners in AI fields like machine learning and planning by improving the reliability of automated parameter tuning, though it is incremental as it builds on existing configuration methods.
The paper addresses common pitfalls in experimental design for automated algorithm configuration, which can hinder performance improvements, and proposes best practices and a tool called GenericWrapper4AC to mitigate these issues.
Good parameter settings are crucial to achieve high performance in many areas of artificial intelligence (AI), such as propositional satisfiability solving, AI planning, scheduling, and machine learning (in particular deep learning). Automated algorithm configuration methods have recently received much attention in the AI community since they replace tedious, irreproducible and error-prone manual parameter tuning and can lead to new state-of-the-art performance. However, practical applications of algorithm configuration are prone to several (often subtle) pitfalls in the experimental design that can render the procedure ineffective. We identify several common issues and propose best practices for avoiding them. As one possibility for automatically handling as many of these as possible, we also propose a tool called GenericWrapper4AC.