Drift mode acquisition using successive model training

Abstract

One example method includes defining a time window during which a drift mode of a machine learning model will be evaluated, and the time window begins at t, training, beginning at the time t, a reduced reference model Rwith a data sample dthat is a subset of a training dataset that was used to train the machine learning model; for 'n' iterations: at a time t, when new data samples dare available, train a new version Rof the reduced reference model R; after Ris trained, collect data v; and compare respective inference values generated by the reduced reference model Rand R, using the data samples dand v, and store an absolute difference between the inference values; defining a drift mode curve using the absolute differences; and based on the drift mode curve, determining a drift mode of the machine learning model.