Patents by Inventor Junichi Shigezumi
Junichi Shigezumi has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Patent number: 11619915Abstract: A computer-implemented reinforcement learning method includes determining, based on a target probability of satisfaction of a constraint condition related to a state of a control object and a specific time within which a controller causes the state of the control object not satisfying the constraint condition to be the state of the control object satisfying the constraint condition, a parameter of a reinforcement learner that causes, in a specific probability, the state of the control object to satisfy the constraint condition at a first timing following a second timing at which the state of control object satisfies the constraint condition; and determining a control input to the control object by either the reinforcement learner or the controller, based on whether the state of the control object satisfies the constraint condition at a specific timing.Type: GrantFiled: February 21, 2020Date of Patent: April 4, 2023Assignee: FUJITSU LIMITEDInventors: Hidenao Iwane, Junichi Shigezumi, Yoshihiro Okawa, Tomotake Sasaki, Hitoshi Yanami
-
Publication number: 20210109491Abstract: A policy improvement method for reinforcement learning using a state value function, the method including: calculating, when an immediate cost or immediate reward of a control target in the reinforcement learning is defined by a state and an input, an estimated parameter that estimates a parameter of the state value function for the state of the control target; contracting a state space of the control target using the calculated estimated parameter; generating a TD error for the estimated state value function that estimates the state value function in the contracted state space of the control target by perturbing each parameter that defines the policy; generating an estimated gradient that estimates the gradient of the state value function with respect to the parameter that defines the policy, based on the generated TD error and the perturbation; and updating the parameter that defines the policy using the generated estimated gradient.Type: ApplicationFiled: September 29, 2020Publication date: April 15, 2021Applicant: FUJITSU LIMITEDInventors: Junichi Shigezumi, Tomotake Sasaki, Hidenao Iwane, Hitoshi Yanami
-
Publication number: 20200285204Abstract: A computer-implemented reinforcement learning method includes determining, based on a target probability of satisfaction of a constraint condition related to a state of a control object and a specific time within which a controller causes the state of the control object not satisfying the constraint condition to be the state of the control object satisfying the constraint condition, a parameter of a reinforcement learner that causes, in a specific probability, the state of the control object to satisfy the constraint condition at a first timing following a second timing at which the state of control object satisfies the constraint condition; and determining a control input to the control object by either the reinforcement learner or the controller, based on whether the state of the control object satisfies the constraint condition at a specific timing.Type: ApplicationFiled: February 21, 2020Publication date: September 10, 2020Applicant: FUJITSU LIMITEDInventors: Hidenao IWANE, Junichi Shigezumi, Yoshihiro Okawa, Tomotake Sasaki, Hitoshi Yanami
-
Publication number: 20200234123Abstract: A reinforcement learning method executed by a computer includes calculating, in reinforcement learning of repeatedly executing a learning step for a value function that has monotonicity as a characteristic of a value according to a state or an action of a control target, a contribution level of the state or the action of the control target used in the learning step, the contribution level of the state or the action to the reinforcement learning being calculated for each learning step and calculated using a basis function used for representing the value function; determining whether to update the value function, based on the value function after each learning step and the calculated contribution level calculated in each learning step; and updating the value function when the determining determines to update the value function.Type: ApplicationFiled: January 16, 2020Publication date: July 23, 2020Applicant: FUJITSU LIMITEDInventors: Junichi Shigezumi, Hidenao Iwane, Hitoshi Yanami
-
Patent number: 10401185Abstract: An apparatus acquires a first piece of trajectory information from among plural pieces of trajectory information, and acquires a first planar graph from among one or more planar graphs. The apparatus generates a second planar graph, based on the first planar graph and plural pieces of position information included in the first piece of trajectory information, and extracts, from among the plural pieces of trajectory information, second pieces of trajectory information indicating trajectories passing a difference portion between the first and second planar graphs. For each of candidate graphs each obtained by excluding a reduction set of edges from the second planar graph, the apparatus calculates optimality of the each candidate graph with which an addition set of trajectories indicated by the first and second pieces of trajectory information are associated, and outputs one of the candidate graphs determined based on the calculated optimality.Type: GrantFiled: October 19, 2016Date of Patent: September 3, 2019Assignee: FUJITSU LIMITEDInventors: Hiroya Inakoshi, Hiroaki Morikawa, Tatsuya Asai, Junichi Shigezumi
-
Patent number: 10191928Abstract: A planar graph generation device that includes a processor that executes a process. The process includes: computing a specific value, including components of a value representing complexity of a track of the given track data, and a value representing a non-nearness between the given track data and each of all the other track data; selecting the track data with the smallest specific value among the collection; a first portion of the first track or a second portion of the second track positioned within the specific distance of each other, or a combination of the first portion and the second portion, is approximated to a specific portion such that a track of the addition target track data after addition passes through the specific portion in cases in which there are portions positioned within the specific distance of each other in a combination of the first track with the second track.Type: GrantFiled: October 9, 2014Date of Patent: January 29, 2019Assignee: FUJITSU LIMITEDInventors: Hiroya Inakoshi, Tatsuya Asai, Hiroaki Morikawa, Junichi Shigezumi
-
Patent number: 10049567Abstract: A traffic flow rate calculation method includes, by using a road network produced by representing a road system with a plurality of nodes and a plurality of edges including a stationary sensor edge in which a stationary sensor measures the number of moving bodies; obtaining the first number of observations corresponding to the number of trajectories measured by mobile sensors for each path, the each path including the at least one edge, the each of trajectories corresponding to a movement trajectory of the moving body, and the second number of observations corresponding to the number of moving bodies measured by the stationary sensor; estimating an observation rate by using the first number of observations and the second number of observations; calculating a traffic flow rate for the each path by using the estimated observation rate and the first number of observations for each path.Type: GrantFiled: January 17, 2017Date of Patent: August 14, 2018Assignee: FUJITSU LIMITEDInventors: Tatsuya Asai, Junichi Shigezumi, Hiroya Inakoshi
-
Publication number: 20180018362Abstract: A data processing apparatus includes a processor. The processor selects candidate tables corresponding to a first table. The respective candidate tables include a first data item included in the first table. The processor acquires a first coincidence degree of the first table for the respective candidate tables. The processor selects third tables corresponding to one of the candidate tables. The respective third tables include a second data item included in the one of the candidate tables. The processor acquires a second coincidence degree of the one of the candidate tables for the respective third tables. The processor acquires a reliability of the one of the candidate tables on basis of the first coincidence degree of the first table for the one of the candidate tables and the second coincidence degree of the one of the candidate tables for the respective third tables.Type: ApplicationFiled: May 18, 2017Publication date: January 18, 2018Applicant: FUJITSU LIMITEDInventors: Tatsuya Asai, TAKASHI KATOH, Junichi Shigezumi, Hiroya INAKOSHI, Yuiko OHTA
-
Patent number: 9727662Abstract: A data output method includes: generating, by a computer, (n?1) first conditions (n is an integer number of three or more) on a relationship between two data by dividing, in a sequential order, a common element condition in which an attribute of each of n data includes a common element, the attribute of each of the two data including the common element; extracting first data corresponding to each of n data to set the first data as a node under a condition; creating a first graph in which nodes are coupled with links based on the first condition; creating a second graph by repeatedly performing a first process, a second process and a third process; determining candidates of a combination of data from the second graph; and outputting a combination of data satisfying the common element condition from the candidates of a combination of data.Type: GrantFiled: June 19, 2014Date of Patent: August 8, 2017Assignee: FUJITSU LIMITEDInventors: Takashi Katoh, Shinichiro Tago, Tatsuya Asai, Hiroaki Morikawa, Junichi Shigezumi, Hiroya Inakoshi
-
Patent number: 9720959Abstract: A data output method, includes: extracting, by a computer, data satisfying a first condition on one data among a plurality of conditions, from a plurality of data, as a node; creating a first graph by coupling two nodes satisfying a second condition on a relationship between two data among the plurality of conditions with a link; creating a second graph by deleting a node that does not satisfy the second condition from the first graph; determining candidates of data combination in the second graph; and outputting a data combination satisfying the plurality of conditions from the candidates of data combination.Type: GrantFiled: May 21, 2014Date of Patent: August 1, 2017Assignee: FUJITSU LIMITEDInventors: Takashi Katoh, Shinichiro Tago, Tatsuya Asai, Hiroaki Morikawa, Junichi Shigezumi, Hiroya Inakoshi
-
Publication number: 20170206782Abstract: A traffic flow rate calculation method includes, by using a road network produced by representing a road system with a plurality of nodes and a plurality of edges including a stationary sensor edge in which a stationary sensor measures the number of moving bodies; obtaining the first number of observations corresponding to the number of trajectories measured by mobile sensors for each path, the each path including the at least one edge, the each of trajectories corresponding to a movement trajectory of the moving body, and the second number of observations corresponding to the number of moving bodies measured by the stationary sensor; estimating an observation rate by using the first number of observations and the second number of observations; calculating a traffic flow rate for the each path by using the estimated observation rate and the first number of observations for each path.Type: ApplicationFiled: January 17, 2017Publication date: July 20, 2017Applicant: FUJITSU LIMITEDInventors: Tatsuya Asai, Junichi Shigezumi, Hiroya Inakoshi
-
Publication number: 20170122760Abstract: An apparatus acquires a first piece of trajectory information from among plural pieces of trajectory information, and acquires a first planar graph from among one or more planar graphs. The apparatus generates a second planar graph, based on the first planar graph and plural pieces of position information included in the first piece of trajectory information, and extracts, from among the plural pieces of trajectory information, second pieces of trajectory information indicating trajectories passing a difference portion between the first and second planar graphs. For each of candidate graphs each obtained by excluding a reduction set of edges from the second planar graph, the apparatus calculates optimality of the each candidate graph with which an addition set of trajectories indicated by the first and second pieces of trajectory information are associated, and outputs one of the candidate graphs determined based on the calculated optimality.Type: ApplicationFiled: October 19, 2016Publication date: May 4, 2017Inventors: Hiroya Inakoshi, Hiroaki Morikawa, Tatsuya Asai, Junichi Shigezumi
-
Publication number: 20170092120Abstract: A common information output method including acquiring first trajectory information and second trajectory information, the first trajectory information indicating a first node sequence corresponding to nodes at which a first object has arrived in order, the second trajectory information indicating a second node sequence corresponding to nodes at which a second object has arrived in order, specifying a number of common node based on the first trajectory information and the second trajectory information when at least a part of the first node sequence and at least a part of the second node sequence are overlapped, the specifying being executed for each of a plurality of patterns of overlap of the first node sequence and the second node sequence, and outputting information in accordance with the specified number of common node.Type: ApplicationFiled: September 21, 2016Publication date: March 30, 2017Applicant: FUJITSU LIMITEDInventors: Junichi Shigezumi, Tatsuya Asai, Hiroaki Morikawa, Hiroya INAKOSHI
-
Patent number: 9581454Abstract: A route information processing apparatus includes: a processor configured to execute a program; and a memory configured to store the program, wherein, based on the program, the processor performs operations to: calculate maximum values corresponding to a Frechet distance between one line segment of one or more line segments between observation points included in a trajectory and each of sets of two nodes corresponding to the one line segment, the nodes being included in a route graph including linked paths which link nodes; and acquire a path corresponding to the trajectory having a smallest value among the maximum values of the one or more line segments.Type: GrantFiled: September 25, 2015Date of Patent: February 28, 2017Assignee: FUJITSU LIMITEDInventors: Junichi Shigezumi, Tatsuya Asai, Hiroaki Morikawa, Hiroya Inakoshi
-
Publication number: 20170045363Abstract: A path graph generation method executed by a processor included in a path graph generation device, the method includes acquiring a plurality of pieces of locus information regarding a plurality of moving bodies, the locus information including each information on a plurality of positions acquired at predetermined time intervals, each of the plurality of positions being associated with a label indicating a belonging area; determining, for each of the plurality of pieces of locus information, whether to integrate two or more positions having an each other's distance less than or equal to a predetermined distance among the plurality of positions into one position, based on the label associated with each of the two or more positions; and setting the one position to a node position and thereby generating the path graph, when it is determined to integrate the two or more positions into the one position.Type: ApplicationFiled: June 10, 2016Publication date: February 16, 2017Applicant: FUJITSU LIMITEDInventors: Tatsuya Asai, Junichi Shigezumi, Hiroya INAKOSHI
-
Patent number: 9547983Abstract: An analysis method includes: acquiring first path information indicating a first movement path, first velocity information of a first moving body on the first movement path, second path information indicating a second movement path, and second velocity information of a second moving body on the second movement path; calculating, based on a specific rule, a similarity between the first velocity information and the second velocity information when the first path information and the second path information are extracted as candidates to be integrated; determining, based on the similarity, whether the first path information and second path information are to be integrated; and generating graph information to draw at least a part of the first path information and at least a part of the second path information as a single path when it is determined that the first path information and the second path information are to be integrated.Type: GrantFiled: July 23, 2015Date of Patent: January 17, 2017Assignee: FUJITSU LIMITEDInventors: Junichi Shigezumi, Tatsuya Asai, Hiroaki Morikawa, Hiroya Inakoshi
-
Patent number: 9372087Abstract: A method of providing destination information includes storing route information in which location information identifying locations, coordinate points information, and information of locations adjacent to the locations are associated with each other; storing probability information in which current locations, destinations, and probabilities of moving from the current locations to the corresponding destinations are associated with each other; receiving information indicating a current position of a terminal apparatus; determining a current location corresponding to the current position; determining one or more first destination candidates based on the probability information; determining one or more second destination candidates among the one or more first destination candidates, when a next destination is not uniquely identifiable; determining a nearby location indicating a location adjacent to each second destination candidate and the current location; and determining the next destination among the one or mType: GrantFiled: August 11, 2014Date of Patent: June 21, 2016Assignee: FUJITSU LIMITEDInventors: Tatsuya Asai, Hiroaki Morikawa, Junichi Shigezumi, Hiroya Inakoshi
-
Publication number: 20160161272Abstract: A route information processing apparatus includes: a processor configured to execute a program; and a memory configured to store the program, wherein, based on the program, the processor performs operations to: calculate maximum values corresponding to a Frechet distance between one line segment of one or more line segments between observation points included in a trajectory and each of sets of two nodes corresponding to the one line segment, the nodes being included in a route graph including linked paths which link nodes; and acquire a path corresponding to the trajectory having a smallest value among the maximum values of the one or more line segments.Type: ApplicationFiled: September 25, 2015Publication date: June 9, 2016Inventors: Junichi Shigezumi, Tatsuya Asai, Hiroaki Morikawa, Hiroya INAKOSHI
-
Publication number: 20160091326Abstract: An analysis method includes: acquiring first path information indicating a first movement path, first velocity information of a first moving body on the first movement path, second path information indicating a second movement path, and second velocity information of a second moving body on the second movement path; calculating, based on a specific rule, a similarity between the first velocity information and the second velocity information when the first path information and the second path information are extracted as candidates to be integrated; determining, based on the similarity, whether the first path information and second path information are to be integrated; and generating graph information to draw at least a part of the first path information and at least a part of the second path information as a single path when it is determined that the first path information and the second path information are to be integrated.Type: ApplicationFiled: July 23, 2015Publication date: March 31, 2016Applicant: Fujitsu LimitedInventors: Junichi Shigezumi, Tatsuya Asai, Hiroaki Morikawa, Hiroya Inakoshi
-
Publication number: 20150308851Abstract: A route extraction device includes a processor that executes a process. The process includes, when extracting, for combination with an identified route out of plural routes, a route from the other routes in the plural routes, performing control that increases extraction probability according to distribution density of the other routes.Type: ApplicationFiled: March 11, 2015Publication date: October 29, 2015Inventors: Hiroaki Morikawa, Hiroya INAKOSHI, Tatsuya Asai, Junichi Shigezumi