Classifying an area as hazardous or non-hazardous based on an operation of a semiconductor device

A system comprising:

• a plurality of semiconductor devices each deployed in a respective area that includes a respective sensor, wherein each of the plurality of semiconductor devices comprise:
• a respective local memory;
• and
• a respective local processor coupled to the respective local memory, wherein the respective local processors of the plurality of semiconductor devices are configured to:
• train a global operation model by federated learning to generate an inference about an environment around a particular semiconductor device;
• and
• update the global operation model by federated learning based on:
• respective test data generated by each of the plurality of semiconductor devices from execution of a golden model, which has a known output for a given input, using the given input;
• respective operating data from each of the plurality of semiconductor devices;
• and
• respective sensor data from the respective sensors;
• and
• a central server coupled to the plurality of semiconductor devices via a wireless network or a physical network, wherein the central server comprises:
• a memory;
• and
• a processor coupled to the memory, wherein the processor is configured to:
• store the global operation model in the memory;
• receive different test data and different operating data from a different semiconductor device based on operation of the different semiconductor device in a different area;
• run the global operation model on the different test data and different operating data to generate an inference about an environment around the different semiconductor device;
• and
• classify the area around the different semiconductor device as hazardous or non-hazardous based on the inference.