Fujitsu Corporation

Haruyasu Ueda

Kenichi Kobayashi

Akira Ura

Masahiro Kataoka

Kiichi Yamada

Yuichi Matsuda

Emeric Viel

Nobuyuki Kuromatsu

Shinichiro Tago

Toshihiro Shimizu

Takamitsu Maeda

Hirotaka Shibata

A search apparatus causes a first learning process using a first sample size and a first hyperparameter value to be executed, and causes a second learning process using the first sample size and a second hyperparameter value to be executed. When a third learning process using a second sample size larger than the first sample size and the first hyperparameter value has not been executed, the search apparatus calculates total resources associated with the first sample size based on resources used by the first and second learning processes. If the total resources exceed a threshold, the search apparatus allows the third learning process to be executed. If the total resources are equal to or less than the threshold, the search apparatus withholds the execution of the third learning process, and allows a fourth learning process using the first sample size or smaller and a third hyperparameter value to be executed.

Search method and apparatus

A relevance searching method performed by a computer, the relevance searching method includes generating a combined database by combining a plurality of databases each including a plurality of elements and relevance information indicating direct relevance between two elements in the plurality of elements; and searching for relevance between two elements that do not have direct relevance by using the combined database.

Relevance searching method, relevance searching apparatus, and storage medium

A machine learning management apparatus identifies a maximum prediction performance score amongst a plurality of prediction performance scores corresponding to a plurality of models generated by executing each of a plurality of machine learning algorithms. As for a first machine learning algorithm having generated a model corresponding to the maximum prediction performance score, the machine learning management apparatus determines a first training dataset size to be used when the first machine learning algorithm is executed next time based on the maximum prediction performance score, first estimated prediction performance scores, and first estimated runtimes. As for a second machine learning algorithm different from the first machine learning algorithm, the machine learning management apparatus determines a second training dataset size to be used when the second machine learning algorithm is executed next time based on the maximum prediction performance score, second estimated prediction performance scores, and second estimated runtimes.

Apparatus and method for managing machine learning with plurality of learning algorithms and plurality of training dataset sizes

A processor obtains a first measurement value representing prediction performance of a model that has been learned by using a first parameter value and training data of first size. The processor calculates a first expected value and a first variance degree of prediction performance of a model that would be learned by using the first parameter value and training data of second size. The processor also obtains a second measurement value representing prediction performance of a model that has been learned by using a second parameter value and training data of the first size. The processor calculates a second expected value and a second variance degree of prediction performance of a model that would be learned by using the second parameter value and training data of the second size.

Method and apparatus for managing machine learning process

An operation control method includes: storing data used by an application according to a first definition with respect to an interface of data operation when defining a way of starting a container, the container representing a region in which an execution environment of the application is disposed; storing an image file for the execution environment of the application; and registering a second definition into definition information, the second definition indicating the way of starting the container and being obtained by adding data information to information regarding the image file for the execution environment of the application in accordance with a third definition with respect to an interface of container operation, the data information being information regarding data used by the application in the container.

Operation control method, and apparatus for operation control, and non-transitory computer-readable storage medium for storing program

A searching device converts a searching data into an encoded searching data in a first encoding format according to a first encoding rule; and searches the encoded searching data in a target encoded data that has been encoded into the first encoding format by the first encoding rule, or that has been encoded in a second encoding format wherein when a specific encoded data encoded in the second encoding format is detected in the searching, the searching compares the encoded searching data and a converted encoded data, the converted encoded data being the specific encoded data converted into the first encoding format based on a predetermined conversion rule.

Non-transitory computer-readable recording medium, searching method, and searching device

A non-transitory computer-readable storage medium stores a data stream processing program that causes a computer to execute a process. The process includes extracting a relationship between properties included in events in a data stream from the data stream itself, definitions of queries for processing the data stream, or both of the data stream and the definitions of the queries; specifying a common partitioning key common to a subset of the queries to be grouped into a partitioning group based on partitioning keys in the definitions of the queries, the relationship between the properties, and a probability of the relationship; and assigning the subset of the queries belonging to the partitioning group to each of parallel nodes provided for parallel distributed processing based on the specified common partitioning key.

Parallel data stream processing method, parallel data stream processing system, and storage medium

A data processing method is executed by a computer to perform distributed processing of a first query and a second query. The data processing method includes calculating a first value indicating a dependency between a first partitioning key and a second partitioning key, the first partitioning key specifying a categorical attribute extracted from syntax of the first query, and the second partitioning key specifying a numerical attribute holding numerical values associated with the second query; generating a query group made of the first query and the second query based on the first value; and performing distributed processing of the first query and the second query based on the query group when executing the first query and the second query.

Data processing method and data processing apparatus

A slave computer reads a plurality of input files that have different formats and generates, for each of the input files, an intermediate file that has added thereto, as a join key, data in a type of column that is common to the input files. The slave computer sorts data in each intermediate file in accordance with the join key. The slave computer extracts, from each intermediate file that has data sorted therein, data that has the format of each of the input files so as to generate a plurality of output files that have data sorted therein with respect to each of the input files. The slave computer inputs the plurality of output files generated to a join program that joins data.

Computer-readable recording medium, execution control method, and information processing apparatus

A data arrangement apparatus calculates evaluation values for individual relocation processes generated based on an access pattern. Each of the relocation processes indicates a relocation of some of a plurality of unit data pieces in a storage device. The data arrangement apparatus selects, amongst influence range levels whose ranges in relation to unit data pieces to be affected by the relocation differ in extent of influence, one influence range level based on the evaluation values and coefficients associated one-to-one with the influence range levels. Based on the extent of influence indicated by the selected influence range level, the data arrangement apparatus determines one or more relocation processes to be executed amongst the generated relocation processes. Based on access status after the determined relocation processes are executed, the data arrangement apparatus updates a coefficient associated with the selected influence range level.

Apparatus and method for data arrangement

A non-transitory computer-readable recording medium storing a program causing a processor to execute a process, the process includes detecting input of data into a memory to which data is inputted in sequence, the data being a processing object of first processing; allocating the first processing, of which a processing object is the data, with respect to any node in a communicable node group; determining whether or not the data is provided with tail information, the tail information indicating tail data of a series of data that are processing objects of the first processing, when detecting input of the data; and allocating second processing, of which a processing object is a processing result of the first processing that is executed with respect to each piece of data of the series of data, to any node of the node group when determining that the data is provided with the tail information.

Storage medium, information processing device, and information processing method

A code converting unit encodes input text data based on an code assignment table stored in a storage device that defines a conversion rule for encoding text data, wherein; the code assignment table being generated by assigning a part of character strings assigned to a 1-byte region of a first code assignment table to a 2-byte region of the code assignment table, and by assigning one or more codes each having two or more bytes to at least a part of character strings assigned to the 2-byte region of the code assignment table.

Non-transitory computer-readable recording medium, encoding method, encoding apparatus, decoding method, and decoding apparatus

A data processing control device performs a MapReduce process. When the data processing control device assigns input data to first Reduce tasks and a second Reduce task performed by using a result of Map processes, the data processing control device assigns input data with smaller amount than any of amounts of the input data which is assigned to the first Reduce tasks to the second Reduce task. The data processing control device assigns the first Reduce tasks and the second Reduce task, to which input data is assigned, to a server that performs Reduce processes in the MapReduce process such that the second Reduce task is started after the assignment of all of the first Reduce tasks.

Data processing control method, computer-readable recording medium, and data processing control device for performing a Mapreduce process

When batch jobs are submitted, the names of the jobs can be arbitrarily designated, so that a user can prepare a configuration file in an easy manner, and the time and effort of the management thereof can be alleviated. Upon submitting batch jobs, a configuration file reading part () reads in a control file (F) in which job names arbitrarily designated by the user are described, and a processing execution part, which is constituted by a job state handling part, an execution-content executing part () and a job submitting part (), determines the sequence of the respective jobs according to the job names by making reference to a job state database (), submits necessary jobs at appropriate timing, and thereafter monitor the states of the jobs thus submitting.

Job control device, job control file, job control method, and job control program

According to an aspect of the embodiment provides a method for controlling a processing device for distributing jobs among a plurality of job processing devices for executing the jobs, respectively. The method comprises the steps of: transmitting a job to one of the job processing devices to have the job executed by the one of the job processing devices; generating a procedure information for transmitting a continuation data from the one of the job processing devices before completion of execution of the job back to the processing device, the continuation data enabling another job processing device to continue execution of the job; and transmitting the procedure information to and receiving the continuation data from the one of the job processing devices.

Growing community of inventors

Ichikawa, Japan

Haruyasu Ueda