Seventh Sense Software actively takes part in multivariate statistical research. The company has developed several state-of-the-art statistical algorithms and has made a number of important mathematical discoveries in pure statistics. Amongst them are the following:
Derivation of a rigorous entropic approach to the general multivariate density estimation problem, namely that of finding the population probability distribution which best fits a given set of multivariate data samples.
Development of the world's fastest single-link clustering algorithm for Euclidean data sets. The algorithm is both storage efficient, coping easily with enormous data sets, and parallelisable for time-critical applications.
Development of advanced multiresolution edge detection techniques for removal of mixed pixels from multispectral imagery to enhance results from density-based hierarchical clustering techniques such as the single-link method.
Discovery of an important one-to-one relationship between the minimal spanning tree resulting from single-link clustering of a data set and its sample density distribution, leading to the development of the so-called "bubble model" density estimator.
Development of improved parameter estimators and hypothesis testing methods for spatial Poisson processes.
Development of a fast, self-consistent and iterative approach to locally-adaptive multivariate kernel density estimation which allows rapid convergence to a final solution even with high-dimensional data.