Contextual active dynamic learning with a digital twin system

A non-transitory computer readable medium comprising computer readable program code, which when executed by a computer processor enables the computer processor to perform a method of ranking data for use in training discriminative artificial intelligence (AI) models, the method comprising:

• receiving, at a digital twin system, data from a distributed learning knowledge acquirer, wherein the digital twin system runs concurrently with the distributed learning knowledge acquirer, wherein:
• digital twin system and the distributed learning knowledge acquirer are located in an edge site of a plurality of edge sites, and
• the plurality of edge sites are connected to a centralized system to form an AI hybrid distributed inference system;
• classifying, at a pre-classifier of the digital twin system, the data to form classified data, wherein:
• the data is classified by a frequency of the data in the distributed learning knowledge acquirer, a type frequency of a type to which the data belongs, and an ablation rank of the data;
• transmitting the classified data to a digital twin model of the digital twin system;
• ranking, by the digital twin model, the classified data to form a ranking for the classified data, wherein the digital twin model uses the frequency, the type frequency, and the ablation rank to create the ranking;
• storing the classified data and the ranking in a data cache based on frequency of observance and digital twin model ranked relevance;
• and
• transmitting the classified data and the ranking to the distributed learning knowledge acquirer and a training dataset repository of the centralized system, wherein:
• the ranking is used by the distributed learning knowledge acquirer to assist in using the classified data to train discriminative AI models, and
• the discriminative AI models are used at endpoint management systems to perform classifications on input data.

A system with a digital twin, comprising:

• a processor;
• storage comprising instructions, which when executed by the processor perform a method, the method comprising:
• receiving, at a digital twin system, data from a distributed learning knowledge acquirer, wherein the digital twin system runs concurrently with the distributed learning knowledge acquirer, wherein:
• digital twin system and the distributed learning knowledge acquirer are located in an edge site of a plurality of edge sites, and
• the plurality of edge sites are connected to a centralized system to form an AI hybrid distributed inference system;
• classifying, at a pre-classifier of the digital twin system, the data to form classified data, wherein:
• the data is classified by a frequency of the data in the distributed learning knowledge acquirer, a type frequency of a type to which the data belongs, and an ablation rank of the data;
• transmitting the classified data to a digital twin model of the digital twin system;
• ranking, by the digital twin model, the classified data to form a ranking for the classified data, wherein the digital twin model uses the frequency, the type frequency, and the ablation rank to create the ranking;
• storing the classified data and the ranking in a data cache based on frequency of observance and digital twin model ranked relevance;
• and
• transmitting the classified data and the ranking to the distributed learning knowledge acquirer and a training dataset repository of the centralized system, wherein:
• the ranking is used by the distributed learning knowledge acquirer to assist in using the classified data to train discriminative AI models, and
• the discriminative AI models are used at endpoint management systems to perform classifications on input data.