Big data Minimization and Optimization Systems

Machine Learning Platforms  often require  hundreds or thousands of features, some of which can be irrelevant, redundant, or both. However, selecting the minimum number of optimal features  from a large feature space in a Big Data DB is considered as an NP-complete problem.

Our proprietary MinOptDB^© System (Minimal Optimal DataBase) is a low-energy, ML-augmented, and human-understandable technology that improves the performance and reduces the storage & computational cost of Machine Learning Platforms  by: 1) eliminating the irrelevant and redundant features; 2) accelerating the model training and prediction speed; 3)reducing the monitoring and maintenance workload for the Big Data pipeline, and; 4) providing better model interpretation and diagnosis capability.