I-Corps: Translation Potential of Bitwise Dependence Detection

This I-Corps project is focused on the development of algorithms that are used for solving a problem or performing a computation. The goal is to minimize model assumptions in data analysis making the procedure widely applicable. Large scale, high-dimensional data sets are becoming ubiquitous in modern society, particularly in physical, biomedical, and commercial applications. While there has been a significant increase in the amount of data generated by these applications, existing scientific statistical models are often lacking in accurately and quickly analyzing these kinds of data. This solution may provide more robust data interpretations with large amounts of data. The technology uses efficient software packages that are about 100 times faster than existing methods. This speed may help to enhance the replicability and reliability of studies in various areas, such as economics, engineering, genetics, and neuroscience. This I-Corps project utilizes experiential learning coupled with a first-hand investigation of the industry ecosystem to assess the translation potential of distribution-free modeling and inference for complex dependency in data science. The algorithms use atomic information in data bits applied to fundamental problems in modeling, testing, and computing by connecting model-free inference to important concepts in statistics and computer science. This method of bitwise dependence detection utilizes hierarchical decompositions and compressive networks tailored to the compressible features of high-dimensional data. Bitwise technology offers significant improvements in computational efficiency and interpretability over existing methods, making it a powerful tool for various applications. In addition, this technology may enable a deeper understanding of nonparametric inference for dependency and for enhancing powerful, robust, interpretable, and computationally efficient nonparametric procedures. The technology also may help build a stronger connection between statistics and computer science, as the algorithms work directly with binary digits, which are fundamental to computing. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.