MACHINE-LEARNING PROGRAM, METHOD, AND APPARATUS FOR MEASURING, BY PORE ELECTRIC RESISTANCE METHOD, TRANSIENT CHANGE IN ION CURRENT ASSOCIATED WITH PASSAGE OF TO-BE-MEASURED PARTICLES THROUGH PORES AND FOR ANALYZING PULSE WAVEFORM OF SAID TRANSIENT CHANGE

摘要

Disclosed is an apparatus that performs machine-learning using, as teacher data and to-be-analyzed data, feature quantities extracted from a pulse waveform of a transient change in inter-electrode ion current during passage of particles through pores. This apparatus has a machine-learning program, a search device, a host attribute table, and a feature-quantity table, and is configured to: extract a first host ID and a second host ID linked to first host attribute information by searching through the host attribute table by using the first host attribute information as a search key; extract a first teacher feature-quantity group obtained from a first known particle belonging to a first category by searching through the feature-quantity table by using the first host ID as a search key; extract a second teacher feature-quantity group obtained from a second known particle belonging to the first category by searching through the feature-quantity table by using the second host ID as a search key; calculate a machine-learning optimization parameter by performing learning using, as a teacher label, first particle-category information representing the first category and using, as teacher data, the first and second teacher feature-quantity groups; and determine whether or not an unknown particle having a first host attribute belongs to the first category by using the machine-learning optimization parameter and using, as an input value, a to-be-analyzed feature-quantity group obtained from the unknown particle.