Transfer Inference Learning

Techniques are described that combine machine learning with an edge network that includes IoT devices to yield an effective and efficient method of assessing a condition of an environment. An inference module that includes a machine-learning algorithm, installed and executing on the IoT devices, assesses a condition detected from multiple, different geographic locations. The IoT devices transfer sets of data and inferences as well as respective sets of confidence levels to converge on a verified set of inferences. The verified set of inferences is arrived at quickly and with a high confidence level