Fig. 5: Results of a NUPACK simulation of using a query image of a cat with a larger database consisting of 5.5 million images.
From: Molecular-level similarity search brings computing to DNA data storage
A Distributions of simulated yields for each target image, categorized by their feature vectors’ Euclidean distances to the query feature vector. The widths of each violin are normalized to be equal. B The top 20 images in the database, sorted by the NUPACK-simulated yield of the reaction between their encoded DNA representation and the reverse complement of the query’s DNA representation.
