# US 11,442,458 B2 · Machine learning system > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,442,458 B2 - Document ID: US11442458B2 - Status: Granted - Kind: B2 - Application number: 16/729,483 - Filed: 2019-12-30 - Published: 2020-07-23 - Granted: 2022-09-13 - Earliest claimed: 2019-01-23 - Assignees: Toyota Jidosha Kabushiki Kaisha - Applicants: Toyota Jidosha Kabushiki Kaisha - Inventors: Keisuke Fukuoka; Shunsuke Kobuna; Eiki Kitagawa; Daiki Yokoyama - Web page: https://patentgrep.com/patent/US11442458B2 ## CPC - G06N 3/084: Backpropagation, e.g. using gradient descent - G05D 1/0219: {ensuring the processing of the whole working surface} - G05D 1/0221: {involving a learning process} - G05D 1/0223: {involving speed control of the vehicle (vehicle fittings for automatically controlling, i.e. preventing speed from exceeding an arbitrarily established velocity or maintaining speed at a particular velocity, as selected by the vehicle operator B60K31/00)} - G06F 18/214: Generating training patterns; Bootstrap methods, e.g. bagging or boosting - G06F 18/24143: {Distances to neighbourhood prototypes, e.g. restricted Coulomb energy networks [RCEN]} - G06N 3/045: Combinations of networks - G06N 3/0499: Feedforward networks - G06N 3/08: Learning methods - G06N 3/09: Supervised learning - G06V 10/82: using neural networks - G06V 20/56: exterior to a vehicle by using sensors mounted on the vehicle - G06V 30/194: References adjustable by an adaptive method, e.g. learning - G06N 3/048: Activation functions - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G05D: SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract A machine learning system comprises a learning planning part 81 configured to create a learning plan of a neural network model, a data set creating part 82 configured to create training data sets, a learning part 83 configured to perform learning of the neural network model using the training data sets when the vehicle is stopped, and a stopping period predicting part 84 configured to predict stopping periods of the vehicle. The data set creating part is configured to allocate the training data sets to a plurality of learning regions determined based on respective ranges of the plurality of input parameters. The learning part is configured to perform learning of the neural network model for each learning region. The learning planning part is configured to allocate the plurality of learning regions to the stopping periods so that learning of the neural network model is not suspended during the stopping periods. ## Claims ### Claim 1 A machine learning system provided in a vehicle, comprising: a learning planning part configured to create a learning plan of a neural network model outputting at least one output parameter from a plurality of input parameters, a data set creating part configured to acquire measured values of the plurality of input parameters and a measured value of the at least one output parameter corresponding to the measured values to create training data sets, a learning part configured to perform learning of the neural network model using the training data sets when the vehicle is stopped based on the learning plan, and a stopping period predicting part configured to predict stopping periods of the vehicle, wherein the data set creating part is configured to allocate the training data sets to a plurality of learning regions determined based on respective ranges of the plurality of input parameters, the learning part is configured to perform learning of the neural network model for each learning region, and the learning planning part is configured to allocate each of the plurality of learning regions to a corresponding stopping period of the stopping periods, and a control part configured to control a drive system of the vehicle based on the neural network model. ### Claim 2 (depends on 1) The machine learning system according to claim 1, wherein the plurality of learning regions includes a small learning region and at least one large learning region encompassing the small learning region, the small learning region and the at least one large learning region being determined so that the small learning region become higher in data density than the large learning region, and the learning planning part is configured to selectively allocate the small learning region and the at least one large learning region to the stopping periods. ### Claim 3 (depends on 1) The machine learning system according to claim 1, wherein priority orders of learning are given to the plurality of learning regions. ### Claim 4 (depends on 3) The machine learning system according to claim 3, wherein the learning planning part is configured to make the priority orders of learning of learning regions including values of the input parameters with relatively high frequencies of appearance higher than the priority orders of learning of learning regions including values of the input parameters with relatively low frequencies of appearance. ### Claim 5 (depends on 3) The machine learning system according to claim 3, wherein the learning planning part is configured to change the priority orders of learning in accordance with information on a road over which the vehicle is being driven or a warmup state of an internal combustion engine provided in the vehicle. ### Claim 6 A electronic control unit provided in a vehicle, configured to: create a learning plan of a neural network model outputting at least one output parameter from a plurality of input parameters, acquire measured values of the plurality of input parameters and a measured value of the at least one output parameter corresponding to the measured values to create training data sets, perform learning of the neural network model using the training data sets when the vehicle is stopped based on the learning plan, predict stopping periods of the vehicle, allocate the training data sets to a plurality of learning regions determined based on respective ranges of the plurality of input parameters, perform learning of the neural network model for each learning region, allocate each of the plurality of learning regions to a corresponding stopping period of the stopping periods, and control a drive system of the vehicle based on the neural network model.