ML Augmented Binary Heap

by Adib Hasan and Angelos Pelecanos

The Binary Heap is a classical data structure commonly used to keep track of the running minimum or running maximum in a stream of insertions and extractions. We tried to augment it using machine learning methods in order to speed up its usual operations. Our augmented heap did up to 55% less comparisons in K-way Merge and gave promising results when solving the 15-puzzle game.

Read more in our Medium post.