公开(公告)号：US20190332951A1

公开(公告)日：2019-10-31

申请号：US16475540

申请日：2017-11-09

Applicant: SONY CORPORATION

Inventor： Ryo NAKAHASHI ,  Hirotaka SUZUKI ,  Takuya NARIHIRA ,  Atsushi NODA

IPC: G06N5/02 ,  G06F16/906 ,  A63F13/46

Abstract: Provided are an apparatus and a method enabling efficient reinforcement learning to be performed by input of an annotation. Included are a database configured to store respective pieces of information of a state, an action, and a reward of a processing execution unit, a learning execution unit configured to execute learning processing in accordance with a reinforcement learning algorithm to which the information stored in the database is applied, and an annotation input unit configured to input annotation information including sub reward setting information and store the annotation information in the database. The learning execution unit executes learning processing to which the respective pieces of information of the state, the action, and the reward input from the processing execution unit and the sub reward setting information are applied, derives an action determination rule, and determines an action which is caused to be executed in accordance with the action determination rule.

公开(公告)号：US20190272477A1

公开(公告)日：2019-09-05

申请号：US16340843

申请日：2017-11-15

Applicant: SONY CORPORATION

Inventor： Naoki IDE ,  Akira FUKUI ,  Ryo NAKAHASHI ,  Hideyuki MATSUNAGA

IPC: G06N20/00 ,  G06F3/01

Abstract: There is provided an information processing apparatus and an information processing method each enabling a reward to be properly imparted for an action. The information processing apparatus includes a reward estimating part executing estimation of a reward for an action on a basis of a user input for the action and a presentation control part executing control for presentation of an estimated reward. The present technique is applicable to an agent such as a robot, an electronic device, software, or the like that can, for example, assist a user, communicate with the user, and the like.

公开(公告)号：US20190317514A1

公开(公告)日：2019-10-17

申请号：US16473461

申请日：2018-01-04

Applicant: SONY CORPORATION

Inventor： Ryo NAKAHASHI ,  Naoki IDE ,  Hideyuki MATSUNAGA ,  Akira FUKUI

IPC: G05D1/02 ,  A47L9/28

Abstract: There is provided an information processing apparatus and information processing method that enables a user to easily teach an action with regard to action learning. A touchscreen outputs unknown state element information that indicates an unknown state element, and teaching request information that requests teaching of an action corresponding to an ambient state, in a case where the ambient state includes the unknown state element. For example, it is possible to apply the present disclosure to a cleaner robot or the like that controls actions on the basis of an action model for finding a probability P (a|s) that the cleaner robot performs an action a in a state s.