The Adaptive Suffix Tree: A Space Efficient Sequence Learning Algorithm

The Adaptive Suffix Trie algorithm was previously proposed by the authors as a sequence learning algorithm for capturing frequent sub sequences of variable length. This algorithm builds up a suffix trie data structure, capturing repetitive patterns in a given set of sequences. Its application has been demonstrated in bioinformatics and text clustering. Suffix trees are the space efficient variant of suffix tries and are thus more widely used in the current literature. In this paper we propose the Adaptive Suffix Tree algorithm, which is based on the same learning principles as the Adaptive Suffix Trie, but has the advantage that it is more space efficient. We discuss the new algorithm in detail and demonstrate that the same set of sub sequences can be learnt by the proposed algorithm while utilizing less than 50% of the space used by its predecessor.