Determining Metaheuristic Similarity Using Behavioral Analysis
提出一套20个行为特征来量化算法搜索行为，通过比较行为特征判断元启发式算法的新颖性，发现大多数算法行为相似，仅少数具有独特行为。
Many nature-inspired metaheuristics have been published, with claims of originality based on the metaphor that inspired the algorithm. Rarely is empirical evidence given to show algorithmic originality. In order to provide an easy and computationally cheap approach to characterise algorithm search behaviour, a suite of 20 behavioural characteristics is proposed. This behavioural characteristic suite allows for the search behaviour of an algorithm to be quantified without manual inspection. By doing so, behavioural novelty of any given algorithm may be determined by comparing the behavioural characteristics to those of well-known metaheuristics. To illustrate this use, and to evaluate whether metaheuristics are behaviourally distinct, a host of metaheuristics is run on various benchmark functions. To evaluate behavioural similarity across all problems, while acknowledging behaviour to be problem dependant, a novel method is proposed. In addition to this method, new behavioural characteristics are also proposed. The behavioural vectors generated for each benchmark function are clustered. The relationships and trends present in the different clusters are summarised by creating a pair-wise matrix for every metaheuristic pair, which tallies the number of times that the pair are found within the same cluster. The tallies are then analysed in order to make inference regarding the distinctness of any metaheuristic’s behaviours, across many different benchmark functions. The analysis finds that the range of unique search behaviours is small and that most metaheuristics share their behaviours with most other metaheuristics. The analysis also identifies both unique algorithms, as well as algorithms which have no unique behaviours.