Machine learning with model filtering and model mixing for edge devices in a heterogeneous environment

Abstract

Machine learning with model filtering and model mixing for edge devices in a heterogeneous environment is disclosed. In an example embodiment, an edge device includes a communication module, a data collection device, a memory, a machine learning module, and a model mixing module. The edge device analyzes collected data with a model for a first task, outputs a result, and updates the model to create a local model. The edge device communicates with other edge devices in a heterogeneous group, transmits a request for local models to the heterogeneous group, and receives local models from the heterogeneous group. The edge device filters the local models by structure metadata, including second local models, which relate to a second task. The edge device performs a mix operation of the second local models to generate a mixed model which relates to the second task, and transmits the mixed model to the heterogeneous group.

What is claimed is: 
     
         1 . A model generating device comprising:
 a memory; and   one or more processors configured to:
 cause transmission of a request for model to each of a plurality of devices; 
 receive the models from the plurality of devices, the respective model received from each device being updated based on data collected by said each device; 
 generate a mixed model from the received models based on at least one of structural metadata, context metadata, or data distribution; and 
 cause transmission of the mixed model to at least one of the plurality of devices. 
   
     
     
         2 . The model generating device according to  claim 1 , wherein the one or more processors are configured to generate the mixed model based on the structural metadata, the structural metadata including at least one of model type, model structure, variables, parameters, or data type. 
     
     
         3 . The model generating device according to  claim 1 , wherein the one or more processors are configured to generate the mixed model based on the context metadata, the context metadata including at least one of a contextual situation in which a model has been or is operating, data collection environment, model history, model stability, or geographic location. 
     
     
         4 . The model generating device according to  claim 1 , wherein the one or more processors are configured to generate the mixed model based on the data distribution, the data distribution including similarity of collected data. 
     
     
         5 . The model generating device according to  claim 1 , wherein the one or more processors are configured to generate the mixed model based on the data distribution, the data distribution including clustering of the received models. 
     
     
         6 . The model generating device according to  claim 1 , wherein the one or more processors are further configured to generate the mixed model based on at least one of averaging operation, genetic algorithm operation, enumeration operation, or ensemble operation. 
     
     
         7 . The model generating device according to  claim 1 , wherein the one or more processors are further configured to switch operations for generating the mixed model based on a quantity of models to be mixed, a level of accuracy, or a computing power of the model generating device. 
     
     
         8 . The model generating device according to  claim 1 , wherein the one or more processors are further configured to:
 collect data and store the data in the memory; and   analyze, using a local model, the collected data.   
     
     
         9 . The model generating device according to  claim 8 , wherein the one or more processors are further configured to:
 update the local model based on the analysis result using the local model; and   generate the mixed model from the received models and the updated local model.   
     
     
         10 . The model generating device according to  claim 8 , wherein the one or more processors are further configured to replace the local model with the mixed model. 
     
     
         11 . A model generating system comprising:
 a first device configured to analyze data collected by the first device using a first model;   a second device configured to analyze data collected by the second device using a second model; and   a third device configured to transmit a request for model to each of the first device and the second device, respectively, wherein   the first device is further configured to transmit the first model to the third device, the second device is further configured to transmit the second model to the third device, and   the third device is further configured to:
 receive the first model and the second model; 
 generate a third model from the first model and the second model based on at least one of structural metadata, context metadata, or data distribution; and 
 transmit the third model to the first device and the second device. 
   
     
     
         12 . The model generating system according to  claim 11 , wherein a communication between the first device and the third device is routed through the second device. 
     
     
         13 . The model generating system according to  claim 12 , wherein the first device is further configured to receive the request for model via the second device. 
     
     
         14 . The model generating system according to  claim 12 , wherein the third device is further configured to receive the first model via the second device. 
     
     
         15 . A model generating device comprising:
 a memory; and   one or more processors configured to:
 update a local model based on data collected by the model generating device; 
 cause transmission of a request for model to each of a plurality of devices; 
 receive the models from the plurality of devices, the model received from each device being updated based on data collected by said each device; and 
 generate a mixed model from the received models and the local model based on at least one of structural metadata, context metadata, or data distribution. 
   
     
     
         16 . The model generating device according to  claim 15 , wherein the one or more processors are further configured to cause transmission of the mixed model to the plurality of devices. 
     
     
         17 . The model generating device according to  claim 15 , wherein the one or more processors are further configured to replace the local model with the mixed model. 
     
     
         18 . The model generating device according to  claim 15 , wherein the one or more processors are configured to generate the mixed model based on the structural metadata, the structural metadata including at least one of model type, model structure, variables, parameters, or data type. 
     
     
         19 . The model generating device according to  claim 15 , wherein the one or more processors are configured to generate the mixed model based on the context metadata, the context metadata including at least one of a contextual situation in which a model has been or is operating, data collection environment, model history, model stability, or geographic location. 
     
     
         20 . The model generating device according to  claim 15 , wherein the one or more processors are configured to generate the mixed model based on the data distribution, the data distribution including similarity of collected data. 
     
     
         21 . A model transmitting device comprising:
 a memory; and   one or more processors configured to:
 collect data and store the data in the memory; 
 analyze, using a local model, the collected data; 
 update the local model based on a result of the analysis; 
 receive a request for model from a model generating device; and 
 in response to receiving the request, cause transmission of the updated local model to the model generating device. 
   
     
     
         22 . The model transmitting device according to  claim 21 , wherein the one or more processors are further configured to receive a generated model from the model generating device after transmitting the updated local model, the generated model has been generated based on the updated local model. 
     
     
         23 . The model transmitting device according to  claim 22 , wherein the one or more processors are further configured to replace the updated local model with the generated model. 
     
     
         24 . A method for generating models, comprising:
 causing, by one or more processors, transmission of a request for model to each of a plurality of devices;   receiving, by the one or more processors, the models from the plurality of devices, the respective model received from each device being updated based on data collected by said each device;   generating, by the one or more processors, a mixed model from the received models based on at least one of structural metadata, context metadata, or data distribution; and   causing, by the one or more processors, transmission of the mixed model to at least one of the plurality of devices.   
     
     
         25 . A method for generating models, comprising:
 updating, by one or more processors of a model generating device, a local model based on data collected by the model generating device;   causing, by the one or more processors, transmission of a request for model to each of a plurality of devices;   receiving, by the one or more processors, the models from the plurality of devices, the model received from each device being updated based on data collected by said each device; and   generating, by the one or more processors, a mixed model from the received models and the local model based on at least one of structural metadata, context metadata, or data distribution.   
     
     
         26 . A method for transmitting models, comprising:
 collecting, by one or more processors, data and storing, by the one or more processors, the collected data in a memory;   analyzing, by the one or more processors, the collected data using a local model;   updating, by the one or more processors, the local model based on a result of the analysis;   receiving, by the one or more processors, a request for model from a model generating device; and   in response to receiving the request, by the one or more processors, causing transmission of the updated local model to the model generating device.