Abstract:
Capabilities of enumeration optimization in the intelligent data analysis by the DSM method of automated hypothesis formation are discussed. Some options for enumeration control by using specifically generated combinatory objects called pseudo-trees are considered. Algorithms for targeted restoration of pseudo-trees according to their pseudomodels are proposed. The concept of the approximate DSM method is developed. The capabilities of additional acceleration of DSM data processing by using parallel algorithms, specialized cloud computing, and some problem-oriented hardware-circuit solutions are described.
Keywords:DSM method of automated hypothesis generation, computing complexity, enumeration optimization, methods of decomposition in reduction of enumeration, approximate computing.