Method and environment controller for validating a predictive model of a neural network through interactions with the environment controller

Abstract

Method and environment controller for validating a predictive model of a neural network. The environment controller receives at least one environmental characteristic value and determines a plurality of input variables. At least one of the plurality of input variables is based on one among the environmental characteristic value(s). The environment controller executes an environment control software module for calculating at least one output variable based on the plurality of input variables. The environment controller transmits the plurality of input variables to a training server executing a neural network training engine using the predictive model; and receives at least one inferred output variable from the training server. Each inferred output variable corresponds to one of the at least one output variable calculated by the environment control software module. The environment controller compares each inferred output variable with the corresponding calculated output variable; and sends a feedback to the training server.

What is claimed is: 
     
         1 . A method for validating a predictive model of a neural network through interactions with an environment controller, the method comprising:
 receiving at least one environmental characteristic value via a communication interface of the environment controller;   determining by a processing unit of the environment controller a plurality of input variables, at least one of the plurality of input variables being based on one among the at least one environmental characteristic value;   executing by the processing unit a legacy environment control software module for calculating at least one output variable based on the plurality of input variables, the legacy environment control software module not implementing a neural network;   transmitting by the processing unit via the communication interface the plurality of input variables to a training server executing a neural network training engine using the predictive model; and   receiving by the processing unit via the communication interface at least one inferred output variable from the training server, each inferred output variable corresponding to one of the at least one output variable calculated by the legacy environment control software module; and   validating by the processing unit the predictive model used by the neural network inference engine executed by the training server by transmitting by the processing unit a feedback to the training server via the communication interface, the feedback comprising for each of the least one inferred output variable received from the training server the corresponding output variable calculated by the legacy environment control software module.   
     
     
         2 . The method of  claim 1 , further comprising displaying by the processing unit each output variable calculated by the legacy environment control software module and each corresponding inferred output variable received from the training server on a display of the environment controller. 
     
     
         3 . The method of  claim 1 , further comprising generating by the processing unit one or more command for controlling a controlled appliance based on one or more of the at least one output variable calculated by the legacy environment control software module. 
     
     
         4 . The method of  claim 3 , further comprising transmitting by the processing unit the one or more command to the controlled appliance via the communication interface. 
     
     
         5 . The method of  claim 1 , wherein the environmental characteristic value comprises one of the following: a current temperature, a current humidity level, a current carbon dioxide (CO2) level, and a current room occupancy. 
     
     
         6 . The method of  claim 1 , wherein at least one of the plurality of input variables is based on one of the following: a set point defined by a user, and a characteristic of an area of a building. 
     
     
         7 . The method of  claim 1 , further comprising receiving by a processing unit of the training server via a communication interface of the training server the plurality of input variables from the environment controller, and executing by the processing unit of the training server the neural network training engine using the predictive model for inferring the at least one inferred output variable based on the plurality of input variables received from the environment controller. 
     
     
         8 . The method of  claim 7 , wherein the predictive model is stored in a memory of the training server and comprises weights used by the neural network training engine. 
     
     
         9 . The method of  claim 7 , further comprising receiving by the processing unit of the training server via the communication interface of the training server the feedback from the environment controller, and comparing by the processing unit of the training server each inferred output variable with the corresponding output variable calculated by the legacy environment control software module and comprised in the received feedback. 
     
     
         10 . The method of  claim 9 , further comprising adapting by the processing unit of the training server the predictive model based on the comparison of each inferred output variable with the corresponding output variable calculated by the legacy environment control software module. 
     
     
         11 . The method of  claim 10 , wherein the adaptation of the predictive model is performed through back-propagation. 
     
     
         12 . A non-transitory computer program product comprising instructions executable by a processing unit of an environment controller, the execution of the instructions by the processing unit providing for validating a predictive model of a neural network through interactions with the environment controller by:
 receiving at least one environmental characteristic value via a communication interface of the environment controller;   determining by the processing unit of the environment controller a plurality of input variables, at least one of the plurality of input variables being based on one among the at least one environmental characteristic value;   executing by the processing unit a legacy environment control software module for calculating at least one output variable based on the plurality of input variables, the legacy environment control software module not implementing a neural network;   transmitting by the processing unit via the communication interface the plurality of input variables to a training server executing a neural network training engine using the predictive model;   receiving by the processing unit via the communication interface at least one inferred output variable from the training server, each inferred output variable corresponding to one of the at least one output variable calculated by the legacy environment control software module; and   validating by the processing unit the predictive model used by the neural network inference engine executed by the training server by transmitting by the processing unit a feedback to the training server via the communication interface, the feedback comprising for each of the least one inferred output variable received from the training server the corresponding output variable calculated by the legacy environment control software module.   
     
     
         13 . An environment controller, comprising:
 a communication interface; and   a processing unit for:
 receiving at least one environmental characteristic value via the communication interface; 
 determining a plurality of input variables, at least one of the plurality of input variables being based on one among the at least one environmental characteristic value; 
 calculating by a legacy environment control software module at least one output variable based on the plurality of input variables, the legacy environment control software module not implementing a neural network; 
 transmitting via the communication interface the plurality of input variables to a training server executing a neural network training engine; 
 receiving via the communication interface at least one inferred output variable from the training server, each inferred output variable corresponding to one of the at least one output variable calculated by the legacy environment control software module; and 
 validating a predictive model used by the neural network inference engine executed by the training server by transmitting a feedback to the training server via the communication interface, the feedback comprising for each of the least one inferred output variable received from the training server the corresponding output variable calculated by the legacy environment control software module. 
   
     
     
         14 . The environment controller of  claim 13 , wherein the processing unit further displays each output variable calculated by the legacy environment control software module and each corresponding inferred output variable received from the training server on a display of the environment controller. 
     
     
         15 . The environment controller of  claim 13 , wherein the processing unit further generates one or more command for controlling a controlled appliance based on one or more of the at least one output variable calculated by the legacy environment control software module. 
     
     
         16 . The environment controller of  claim 15 , wherein the processing unit further transmits the one or more command to the controlled appliance via the communication interface. 
     
     
         17 . The environment controller of  claim 13 , wherein the environmental characteristic value comprises one of the following: a current temperature, a current humidity level, a current carbon dioxide (CO2) level, and a current room occupancy. 
     
     
         18 . The environment controller of  claim 13 , wherein at least one of the plurality of input variables is based on one of the following: a set point defined by a user, and a characteristic of an area of a building.