A machine learning device performs machine learning with respect to a servo motor controller that converts a three-phase current to a two-phase current of the d- and q-phase. The machine learning device includes: a state information acquisition unit configured to acquire, from the servo motor controller, state information including velocity or a velocity command, reactive current, and an effective current command and effective current or a voltage command; an action information output unit configured to output action information including a reactive current command to the servo motor controller; a reward output unit configured to output a value of a reward of reinforcement learning based on the voltage command or the effective current command and the effective current; and a value function updating unit configured to update a value function on the basis of the output value of the reward, the state information, and the action information.

Laundry appliances use machine learning models and/or personalization to provide better treatments. As one example, a laundry appliance has a chamber in which laundry items are placed for treatment. Sensor(s) are positioned to sense contents of the chamber or to sense the laundry items as they are loaded into the chamber. The machine learning model uses data from these sensors to determine various attributes of the laundry items and/or the treatment, such as the type of fabric and/or how dirty the items are, and the treatment process is controlled accordingly. Personalized data, such as an individual's preferences for laundry treatments or his sensitivities and allergies, may also be used to personalize the treatment process.

Methods and apparatus relating to autonomous vehicle neural network optimization techniques are described. In an embodiment, the difference between a first training dataset to be used for a neural network and a second training dataset to be used for the neural network is detected. The second training dataset is authenticated in response to the detection of the difference. The neural network is used to assist in an autonomous vehicle/driving. Other embodiments are also disclosed and claimed.

A control device for performing optimal control by path integral includes a neural network section including a machine-learned dynamics model and cost function, an input section that inputs a current state of a control target and an initial control sequence for the control target into the neural network section, and an output section that outputs a control sequence for controlling the control target, the control sequence being calculated by the neural network section by path integral from the current state and the initial control sequence by using the dynamics model and the cost function. Here, the neural network section includes a second recurrent neural network incorporating a first recurrent neural network including the dynamics model.

Disclosed is a real-time decision support system for facilitating optimization of in a distributed environment. A data capturing module for capturing data from a plurality of data sources at a predefined interval of time. The data is captured pertaining to a specific domain and a specific geographical area. A forecasting module for forecasting demand of energy to be consumed by energy utilities in the specific domain and the geographical area upon analyzing the data. A position gap determination module for determining a position gap indicating a difference between the demand of energy and supply of energy. An energy optimization module for identifying at least one energy pool, from a plurality of energy pools, for retrieving a deficit of the demand of energy and providing the deficit of the demand of energy retrieved from the at least one energy pool in order to bridge the position gap.

A method for assisting maneuver planning for a transportation vehicle driving by automation or for a robot; wherein a state space of an environment of the transportation vehicle or the robot is discretely described by a Markov decision process; wherein optimal action values for discretized actions are determined by dynamic programming, based on discrete states in the state space; wherein a mapping with states in the state space as input values, and with action values for actions in the state space as output values, is learned by a reinforcement learning method; wherein a reinforcement learning agent is initialized based on the optimal action values determined by the dynamic programming; and wherein the learned mapping is provided for maneuver planning. Also disclosed is a device for assisting maneuver planning for a transportation vehicle driving by automation or for a robot.

In a motor control device, a velocity feed-forward control portion includes a velocity-side acceleration input portion that outputs received high-order command acceleration as a velocity-side acceleration output; a velocity-side velocity input portion that outputs a received high-order command velocity as a velocity-side velocity output; velocity-side boundary-velocity input portions which are prepared so as to respectively correspond to boundary velocities, and to output velocity-side boundary velocity outputs from the velocity-side boundary-velocity input portions corresponding to the high-order command velocity, the boundary velocities being velocities at boundaries of preset adjacent velocity ranges obtained by dividing a limited velocity range; a velocity-side first weight learning portion that changes velocity-side first learning weights in accordance with a velocity deviation, the velocity-side first learning weights respectively corresponding to velocity-side first outputs; and a velocity-side output portion that outputs, as a second tentative command current, a value obtained by summing velocity-side first multiplication values.

Tuning model structures of dynamic systems are described herein. One method for tuning model structures of a dynamic system includes predicting a variable for each of a number of models associated with a number of model structures of a dynamic system, calculating a rate of error of the predicted variable for each of the number of models compared to an observed variable, determining a best model structure among the number of model structures based on the calculated rate of error, and creating a revised model structure using the best model structure to tune the number of model structures of the dynamic system.

In a motor control device, a velocity feed-forward control portion includes a velocity-side acceleration input portion that outputs received high-order command acceleration as a velocity-side acceleration output; a velocity-side velocity input portion that outputs a received high-order command velocity as a velocity-side velocity output; velocity-side boundary-velocity input portions which are prepared so as to respectively correspond to boundary velocities, and to output velocity-side boundary velocity outputs from the velocity-side boundary-velocity input portions corresponding to the high-order command velocity, the boundary velocities being velocities at boundaries of preset adjacent velocity ranges obtained by dividing a limited velocity range; a velocity-side first weight learning portion that changes velocity-side first learning weights in accordance with a velocity deviation, the velocity-side first learning weights respectively corresponding to velocity-side first outputs; and a velocity-side output portion that outputs, as a second tentative command current, a value obtained by summing velocity-side first multiplication values.

Tuning model structures of dynamic systems are described herein. One method for tuning model structures of a dynamic system includes predicting a variable for each of a number of models associated with a number of model structures of a dynamic system, calculating a rate of error of the predicted variable for each of the number of models compared to an observed variable, determining a best model structure among the number of model structures based on the calculated rate of error, and creating a revised model structure using the best model structure to tune the number of model structures of the dynamic system.