Target aware adaptive application for anomaly detection at the network edge

A computer-implemented method for generating and deploying customized Deep Learning (DL) anomaly detection models on edge devices, the method executed by one or more computing device processors and comprising:

• communicating with a plurality of edge devices to retrieve information from each edge device that identifies configuration of a corresponding edge device;
• selecting, for each of the edge devices based on the information, at least one primary DL anomaly detection models from amongst one or more primary DL anomaly detection models;
• converting, for each of the edge devices based on the device configuration information, the selected the one or more primary DL anomaly detection models to one or more customized DL anomaly detection models that are specific to a corresponding one of the edge devices;
• translating, for each of the edge devices, the one or more customized DL anomaly detection models to a format executable on the corresponding one of the edge devices;
• and
• deploying, on each of the edge devices, the one or more translated and customized DL anomaly detection models for execution on the corresponding one of the edge devices.

A computer program product comprising a non-transitory computer-readable medium including:

• a first set of codes for causing a computer processor to communicate with a plurality of edge devices to retrieve information from each edge device that identifies configuration of a corresponding edge device;
• a second set of codes for causing a computer processor to select, for each of the edge devices based on the information, at least one primary DL anomaly detection models from amongst one or more primary DL anomaly detection models;
• a third set of codes for causing a computer processor to convert, for each of the edge devices based on the device configuration information, the selected at least one primary DL anomaly detection models to one or more customized DL anomaly detection models that are specific to a corresponding one of the edge devices;
• a fourth set of codes for causing a computer processor to translate, for each of the edge devices, the one or more customized DL anomaly detection models to a format executable on the corresponding one of the edge devices;
• and
• a fifth set of codes for causing a computer processor to deploy, on each of the edge devices, the one or more translated and customized DL anomaly detection models for execution on the corresponding one of the edge devices.