WO2015090062A1 - 终端终生学习处理方法、装置和*** - Google Patents

终端终生学习处理方法、装置和*** Download PDF

Info

Publication number
WO2015090062A1
WO2015090062A1 PCT/CN2014/082698 CN2014082698W WO2015090062A1 WO 2015090062 A1 WO2015090062 A1 WO 2015090062A1 CN 2014082698 W CN2014082698 W CN 2014082698W WO 2015090062 A1 WO2015090062 A1 WO 2015090062A1
Authority
WO
WIPO (PCT)
Prior art keywords
terminal
task
functional module
server
function module
Prior art date
Application number
PCT/CN2014/082698
Other languages
English (en)
French (fr)
Inventor
杨强
宋阳秋
何秀强
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Publication of WO2015090062A1 publication Critical patent/WO2015090062A1/zh
Priority to US15/181,664 priority Critical patent/US10078509B2/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F8/00Arrangements for software engineering
    • G06F8/60Software deployment
    • G06F8/65Updates
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04BTRANSMISSION
    • H04B1/00Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
    • H04B1/38Transceivers, i.e. devices in which transmitter and receiver form a structural unit and in which at least one part is used for functions of transmitting and receiving
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/34Network arrangements or protocols for supporting network services or applications involving the movement of software or configuration parameters 

Definitions

  • Embodiments of the present invention relate to the technical field of computer science, and in particular, to a terminal lifelong learning processing method, apparatus, and system. Background technique
  • the intelligent terminal has a separate operating system and a separate operating space. Therefore, the smart terminal has the function of intelligently analyzing and processing problems. For example, the smart terminal can intelligently analyze the user's usage scene according to the current usage configuration of the user, and can also perform intelligent analysis on the user's preference of the smart terminal.
  • the embodiments of the present invention provide a terminal lifetime learning processing method, apparatus, and system, which overcome the problem that the smart terminal cannot perform function expansion and update in the prior art.
  • a first aspect of the present invention provides a terminal lifetime learning processing method, including: sending a download request of a function module to a server, where the download request includes description information of the function module;
  • the function module is used to expand and/or update local functions.
  • the method before the sending the download request of the function module to the server, the method further includes:
  • a retrieval query request is sent to the server, where the retrieval query request includes task requirement information;
  • the method further includes: processing the task by using an extended and/or updated local function.
  • the operational data obtained by the task processing is fed back to the server, so that the server updates the functional module according to the task and the operational data corresponding to the task.
  • a second aspect of the present invention provides a terminal lifetime learning processing method, including: receiving a download request of a function module sent by a terminal, where the download request includes description information of the function module;
  • the method before the downloading the function module sent by the terminal, the method further includes:
  • the function module is updated according to the task and the running data.
  • the method before the download request of the function module sent by the receiving terminal, the method further includes:
  • the search result is sent to the terminal, and the search result includes description information of the function module corresponding to the task requirement information.
  • a third aspect of the present invention provides a terminal device, including: a sending unit, configured to send a download request of the function module to the server, where the download request includes description information of the function module;
  • a receiving unit configured to receive a function module that is sent by the server and corresponding to the description information
  • a local update unit configured to expand and/or update the local function by using the function module
  • the method further includes:
  • a search query unit configured to send a search query request to the server after the task is established, before the sending unit sends the download request of the function module to the server, where the search query request includes task requirement information;
  • a processing unit configured to: after the local update unit adopts the function module, expand and/or update the local function, and process the task by using an extended and/or updated local function.
  • the method further includes:
  • a feedback unit configured to: after the processing unit processes the task by using an extended and/or updated local function, feed back, to the server, the operation data obtained by the task processing, so that the server is configured according to the task and
  • the function module corresponding to the task updates the function module.
  • a fourth aspect of the present invention provides a server, including:
  • a receiving unit configured to receive a download request of a function module sent by the terminal, where the download request includes description information of the function module;
  • a sending unit configured to send, to the terminal, a function module corresponding to the description information, so that the terminal uses the function module to expand and/or update a local function.
  • the method further includes:
  • a creating unit configured to: before the receiving unit receives the download request of the function module sent by the terminal, create a function module according to the task and the running data corresponding to the task;
  • a storage unit configured to store, after the receiving unit receives the download request of the function module sent by the terminal, a correspondence between the function module and the description information of the function module.
  • the method further includes:
  • an update unit configured to: after the storage unit stores a correspondence between the function module and the description information of the function module, receive a task sent by the terminal and operation data corresponding to the task;
  • the function module is updated according to the task and the running data.
  • the method further includes:
  • a search processing unit configured to receive, before the downloading request of the function module sent by the terminal, the search query request sent by the terminal, where the search query request includes task requirement information;
  • the search result is sent to the terminal, and the search result includes description information of the function module corresponding to the task requirement information.
  • a fifth aspect of the present invention provides a terminal life learning processing system, including the terminal device according to any one of the third aspect to the second aspect, and the fourth aspect to the fourth aspect A server as claimed in any one of the third possible implementations.
  • the terminal life learning processing method, device and system according to the embodiment of the present invention, the terminal downloading the function module continuously updated by the server according to the task and the operation data corresponding to the task, and using the function module to implement local function expansion and/or update, The problem that the terminal cannot perform function expansion and update in the prior art is solved.
  • the terminal has the ability to have lifelong learning.
  • Embodiment 1 is a flowchart of Embodiment 1 of a terminal life learning processing method of the present invention
  • Embodiment 2 is a flowchart of Embodiment 2 of a terminal life learning processing method of the present invention
  • FIG. 3 is a schematic structural diagram of Embodiment 1 of a terminal device according to the present invention
  • 4 is a schematic structural diagram of Embodiment 2 of a terminal device according to the present invention
  • FIG. 5 is a schematic structural diagram of Embodiment 1 of a server according to the present invention.
  • Embodiment 2 of a server according to the present invention is a schematic structural diagram of Embodiment 2 of a server according to the present invention.
  • FIG. 7 is a schematic structural diagram of an embodiment of a terminal life learning processing system according to the present invention
  • FIG. 8 is a schematic diagram of a mechanism of a third embodiment of the terminal device according to the present invention.
  • FIG. 9 is a schematic structural diagram of Embodiment 3 of a server according to the present invention. detailed description
  • FIG. 1 is a flowchart of Embodiment 1 of a terminal life learning processing method of the present invention.
  • the method in this embodiment may include: Step 101: Send a download request of a function module to a server, where the download request includes Describe the description of the function module;
  • Step 102 Receive a function module that is sent by the server and corresponding to the description information.
  • Step 103 Use the function module to expand and/or update the local function.
  • the terminal may send a request containing the corresponding function module description information to the server according to its own function requirement, and request to download the function module. It is also possible to periodically send a download request containing the corresponding function module description information to the server by setting the terminal.
  • the description information of the function module may be information describing the function module, such as the name of the function module and the type of the function module.
  • the terminal receives the function module corresponding to the description information sent by the server, and uses the downloaded function module to expand the local function or update the existing function module locally.
  • the terminal has the ability to analyze the behavior of the user, image recognition and other lifelong learning. In addition to obtaining updated function modules from the server side, the terminal can also independently update the function modules according to the data collected by the terminal to improve their intelligence. Further, before the sending the download request of the function module to the server, the method further includes: After the task is established, sending a search query request to the server, where the search query request includes task requirement information; receiving a search result sent by the server, where the search result includes a function module corresponding to the task demand information Descriptive information; after the function module is used to expand and/or update the local function, the method further includes: processing the task by using an extended and/or updated local function.
  • the terminal wants to query the function module about its function requirement, and can determine whether the server has a corresponding function module by sending a search query to the server, and if so, the server can also be obtained from the server.
  • the description result of the function module is obtained by the feedback result, and the terminal sends a download request of the function module to the server.
  • the function module is used to complete local function expansion and/or update, and the task is processed by the extended and/or updated local function module.
  • the personalized recommendation function in the smart phone that does not match the user of the smart phone or the personalized recommendation function module in the smart phone needs to be updated, and the smart phone may send a “recommended” keyword to the server for searching.
  • the query is used to retrieve a query about the recommended function module, and the smart phone receives the search result sent by the server, and the search result includes the name of the functional module and the corresponding type.
  • the terminal downloads the function module to the local area, and uses the function module to implement the function of recommending movies, books, music, and the like to the smartphone user.
  • the method further includes: feeding back, to the server, the running data obtained by the task processing, so that the server according to the task and The operation data corresponding to the task updates the function module.
  • the terminal processes the task by using an extended and/or updated local function module, and in the process of processing the task, the terminal needs to provide the function module operation data, and the terminal may The processing data is uploaded to the server after the task is processed, so that the server updates the function module according to the task and its corresponding running data.
  • the functional modules that are updated by the terminal's uploading to the server's operational data are more targeted to the terminal.
  • the terminal sends a function module retrieval query and a download request to the server, and uses the downloaded function module to expand and/or update the local function, and may also upload the operation data pair obtained by the task processing to the server through the terminal.
  • the function module on the server side is updated to solve the problem that the terminal cannot perform function expansion and/or update in the prior art.
  • the terminal has the ability to have lifelong learning.
  • Embodiment 2 is a flowchart of Embodiment 2 of a method for a lifetime learning process of a terminal according to the present invention. As shown in FIG. 2, the method in this embodiment may include:
  • Step 201 Receive a download request of a function module sent by a terminal, where the download request includes description information of the function module;
  • Step 202 Send, to the terminal, a function module corresponding to the description information, so that the terminal uses the function module to expand and/or update a local function.
  • the server sends a function module corresponding to the description information to the terminal according to the description information of the function module included in the function module download request sent by the terminal, so that the terminal expands and/or updates the local function.
  • the method further includes: creating a function module according to the task and the operation data corresponding to the task; storing between the function module and the description information of the function module Correspondence.
  • the server needs to create a function module according to the task and the corresponding operation data, and the function module is an initial function module.
  • the initial functional modules can be created through lifelong learning algorithms such as classification, clustering, regression, and rules.
  • the type of function module can be determined according to the task, and the corresponding one is to construct a whole model, a group model or an individual model.
  • the learned function modules are stored in the function module library by index. The index corresponds to the description of the function module.
  • a large number of functional modules are stored and retrieved, and the data structures of different types of functional modules need to be analyzed, and the effective arching is performed separately.
  • dense vector, sparse vector, and tree classifiers require different indexing mechanisms. Search by id, group, and similarity.
  • the functional module storage used in this embodiment involves storage and retrieval of large-scale dense vectors, sparse vectors, trees, graphs.
  • dense vectors methods such as KDTree, ConeTree, BBTree, sparse vectors, inverted tables, PPJoin, Spilltree, etc.
  • For tree-type functional modules you can convert the tree into rules and then use string matching. You can also think of the tree as a simple graph and directly match the graph.
  • the function module corresponds to a server-side statistics or a life-study result model for the client to perform functions such as intelligent prediction. It should be noted that, in this embodiment, the specific implementation of the function module creation and storage is not limited.
  • the receiving terminal further includes: The sent task and the running data corresponding to the task; updating the function module according to the task and the running data.
  • the server update function module may be based on the task sent by the terminal and the running data corresponding to the task, and the data may be generated in batches over time, or may be fed back to the server through the terminal word. of.
  • the update algorithm of the function module is learned through statistics and lifelong learning.
  • the function module update corresponding to the same task can be completed by online learning, and the function module corresponding to different tasks can be completed by using migration learning. It also includes multi-tasking learning, lifelong learning and other mechanisms that can be used to complete the update of functional modules.
  • Statistics and lifelong learning can include relevant statistical and lifelong learning algorithms such as kmeans, svm, naive Bayes, decision trees, and more. Methods for migrating learning design may include multitasking based learning, TrAdaboost, and the like. Lifelong learning can include deep learning, ELLA algorithms, and more.
  • the update of function modules through the lifelong learning mechanism makes the function modules more and more accurate and intelligent. It should be noted that this embodiment does not limit the specific implementation of the function module update.
  • the method before receiving the download request of the function module sent by the terminal, the method further includes: receiving a search query request sent by the terminal, where the search query request includes a task requirement
  • the server sends a search result to the terminal, the search result including description information of the function module corresponding to the task requirement information.
  • the server receives the search query request sent by the terminal, and sends the description information of the function module corresponding to the task requirement information retrieved by the terminal to the terminal.
  • the description information of the function module may be information for describing the function module, such as the name of the function module and the type of the function module.
  • the server creates a function module according to the task and the operation data corresponding to the task, and stores the function module to the function module library, according to the task and the operation data corresponding to the task.
  • the function module is updated, and the search query and the download request sent by the terminal are received, and the corresponding function module is sent to the terminal.
  • the local function is expanded and/or updated.
  • the terminal has the ability to have lifelong learning. The problem that the terminal cannot perform function expansion and/or update in the prior art is solved.
  • the terminal device in this embodiment may include:
  • the sending unit 101 is configured to send a download request of the function module to the server, where the download request includes description information of the function module;
  • the receiving unit 102 is configured to receive a function module that is sent by the server and corresponding to the description information, and the local update unit 103 is configured to expand and/or update the local function by using the function module.
  • the terminal in this embodiment may be used to implement the technical solution of the method embodiment shown in FIG. 1.
  • the principle and the technical effect are similar, and details are not described herein again.
  • the terminal device in this embodiment may further include: a search query unit 104, based on the device structure shown in FIG. And sending, by the sending unit, a search query request to the server, after the task is sent to the server, before the task is established, the search query request includes task requirement information; and receiving the search sent by the server As a result, the search result includes description information of the function module corresponding to the task requirement information; the processing unit 105 is configured to: after the local update unit adopts the function module, expand and/or update the local function Processing the task with extended and/or updated local functions;
  • the feedback unit 106 is configured to: after the processing unit processes the task by using an extended and/or updated local function, feed back, to the server, the operation data obtained by the task processing, so that the server according to the task
  • the function module is updated with operational data corresponding to the task.
  • the device of this embodiment can be used to implement the technical solution of the method embodiment shown in FIG.
  • the current principle and technical effect are similar, and will not be described here.
  • FIG. 5 is a schematic structural diagram of Embodiment 1 of a server according to the present invention. As shown in FIG. 5, the server in this embodiment may include:
  • the receiving unit 201 is configured to receive a download request of the function module sent by the terminal, where the download request includes description information of the function module;
  • the sending unit 202 is configured to send, to the terminal, a function module corresponding to the description information, so that the terminal uses the function module to expand and/or update a local function.
  • the device in this embodiment can be used to implement the technical solution of the method embodiment shown in FIG. 2, and the principle and the technical effect are similar, and details are not described herein again.
  • FIG. 6 is a schematic structural diagram of a second embodiment of a server according to the present invention.
  • the server in this embodiment further includes: a creating unit 203, and a server, based on the device structure shown in FIG.
  • the receiving unit Before receiving the download request of the function module sent by the terminal, the receiving unit creates a function module according to the task and the running data corresponding to the task;
  • the storage unit 204 is configured to store a correspondence between the function module and the description information of the function module before the receiving unit receives the download request of the function module sent by the terminal.
  • the updating unit 205 is configured to: after the storage unit stores the correspondence between the function module and the description information of the function module, receive a task sent by the terminal and operation data corresponding to the task;
  • the retrieval processing unit 206 before the receiving unit receives the download request of the function module sent by the terminal, receives a search query request sent by the terminal, where the search query request includes task requirement information;
  • the search result is transmitted to the terminal, and the search result includes description information of the function module corresponding to the task demand information.
  • the device in this embodiment can be used to implement the technical solution of the method embodiment shown in FIG. 2, and the principle and the technical effect are similar, and details are not described herein again.
  • FIG. 7 is a schematic structural diagram of an embodiment of a terminal life learning processing system according to the present invention, as shown in FIG.
  • the terminal life learning processing system of this embodiment may include: a terminal device 301 and a server 302.
  • the terminal device 301 of the present embodiment has the same principle as the embodiment of the terminal life learning processing method of the present invention shown in FIG. 1 and will not be described again.
  • the server of the present embodiment and the terminal lifetime learning processing method of the present invention shown in FIG. 2 are implemented.
  • the principle of the example is the same and will not be described again.
  • FIG. 8 is a schematic diagram of a mechanism of a third embodiment of a terminal device according to the present invention. As shown in FIG.
  • the terminal device of the embodiment includes: a processor 401 and an interface circuit 402, and further includes a memory 403 and a bus 404.
  • the processor 401, the interface circuit 402, and the memory 403 are connected by a bus 404 and complete communication with each other.
  • the bus 404 can be an industry standard architecture (English: Industry Standard Architecture, ISA for short) bus, external device interconnection (English: Peripheral Component Interconnect, referred to as PCI) bus or internal integrated circuit (English: Inter-Integrated Circuit, abbreviation : I2C) Bus, etc.
  • the bus 404 can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one thick line is shown in Figure 4, but it does not mean that there is only one bus or one type of bus.
  • Memory 403 is for storing executable program code, the program code including computer operating instructions.
  • the memory 403 may be a volatile memory, such as a random-access memory (English: random-access memory, RAM for short) or a non-volatile memory (NVM). , read-only memory (ROM), flash memory (English: flash memory), hard disk (English: hard disk drive, HDD) or solid state drive (English: solid-state drive , referred to as SSD).
  • the processor 401 can be a central processing unit (English: central processing unit, abbreviated as: CPU).
  • the processor 401 can execute the terminal life learning processing method provided by the embodiment of the present invention by using an operation instruction or a program code stored in the memory 403, where the method includes: the processor 401 sends a download request of the function module to the server, where The download request includes description information of the function module;
  • the processor 401 receives, by the interface circuit 402, a function module that is sent by the server and corresponds to the description information.
  • the processor 401 employs the functional modules to extend and/or update local functions.
  • the processor 401 sends a search query request to the server after the task is established, and the search query
  • the request request contains the task requirement information
  • the processor 401 receives the search result sent by the server through the interface circuit 402, where the search result includes description information of the function module corresponding to the task requirement information;
  • the processor 401 employs the function module to expand and/or update the local function, and further includes: the processor 401 processes the task by using an extended and/or updated local function.
  • the processor 401 feeds back to the server the operational data obtained by the task processing, so that the server updates the functional module according to the task and the operational data corresponding to the task.
  • FIG. 9 is a schematic diagram of the mechanism of the third embodiment of the server of the present invention.
  • the terminal of the embodiment includes: a processor 501 and an interface circuit 52.
  • the memory 503 and the bus 504 are also shown.
  • the interface circuit 502 and the memory 503 are connected by a bus 504 and complete communication with each other.
  • the bus 504 can be an industry standard architecture (English: Industry Standard Architecture, ISA for short) bus, external device interconnection (English: Peripheral Component Interconnect, PCI) bus or internal integrated circuit (English: Inter-Integrated Circuit, abbreviation : I2C) Bus, etc.
  • the bus 504 can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one thick line is shown in Figure 9, but it does not mean that there is only one bus or one type of bus.
  • Memory 503 is for storing executable program code, the program code including computer operating instructions.
  • the memory 503 may be a volatile memory, such as a random-access memory (English: random-access memory, abbreviation: RAM), or a non-volatile memory (NVM). , read-only memory (ROM), flash memory (English: flash memory), hard disk (English: hard disk drive, HDD) or solid state drive (English: solid-state drive , referred to as SSD).
  • the processor 501 can be a central processing unit (English: central processing unit, abbreviated as: CPU).
  • the processor 501 can execute the terminal life learning processing method provided by the embodiment of the present invention by using an operation instruction or a program code stored in the memory 503, where the method includes:
  • the processor 501 receives, by using the interface circuit 502, a download request of the function module sent by the terminal, where the download request includes description information of the function module;
  • the processor 501 sends a function module corresponding to the description information to the terminal, so that the terminal uses the function module to expand and/or update the local function.
  • the processor 501 creates a function module according to the task and the operation data corresponding to the task; the memory 503 stores a correspondence relationship between the function module and the description information of the function module.
  • the processor 501 receives, through the interface circuit 502, a task sent by the terminal and operation data corresponding to the task;
  • the processor 501 updates the function module according to the task and the operational data.
  • the processor 501 receives the retrieval query request sent by the terminal through the interface circuit 502, where the retrieval query request includes task requirement information;
  • the processor 501 transmits a search result to the terminal, where the search result includes description information of the function module corresponding to the task requirement information.
  • the aforementioned program can be stored in a computer readable storage medium.
  • the program when executed, performs the steps including the foregoing method embodiments; and the foregoing storage medium includes: a medium that can store program codes, such as a ROM, a RAM, a magnetic disk, or an optical disk.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Security & Cryptography (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Information Transfer Between Computers (AREA)
  • Stored Programmes (AREA)
  • Electrically Operated Instructional Devices (AREA)

Abstract

 本发明实施例提供一种终端终生学习处理方法、装置和***。本发明终端终生学习处理方法,包括:向服务器发送功能模块的下载请求,所述下载请求包含所述功能模块的描述信息;接收服务器发送的与所述描述信息对应的功能模块;采用所述功能模块,对本地功能进行扩展和/或更新。本发明实施例实现了终端的终生学习,解决现有技术中终端无法进行功能扩展和更新的问题。

Description

终端终生学习处理方法、 装置和***
技术领域
本发明实施例涉及计算机科学的技术领域, 尤其涉及一种终端终生学 习处理方法、 装置和***。 背景技术
科技的快速发展, 智能终端随之而诞生。 智能终端具有独立的操作系 统和独立的运行空间。因此,智能终端具备智能地分析和处理问题的功能。 例如, 智能终端可以根据用户当前的使用配置智能地分析用户的使用场 景、 也可以对智能终端的用户的使用偏好进行智能分析。
目前市场上的智能终端, 在出厂时由厂商将相应的***移植到其操作 ***中。 因此, 智能终端相应的功能是被限定的、 不可更改的, 对于所有 用户而言智能终端相应的功能都是相同的。
因此, 现有技术中, 智能终端无法进行功能扩展和更新。 发明内容
本发明实施例提供一种终端终生学习处理方法、 装置和***, 以克服 现有技术中, 智能终端无法进行功能扩展和更新的问题。
本发明的第一方面提供了一种终端终生学习处理方法, 包括: 向服务器发送功能模块的下载请求, 所述下载请求包含所述功能模块 的描述信息;
接收服务器发送的与所述描述信息对应的功能模块;
采用所述功能模块, 对本地功能进行扩展和 /或更新。
在第一方面的第一种可能的实现方式中, 所述向服务器发送功能模块 的下载请求之前, 还包括:
在任务建立之后, 向服务器发送检索査询请求, 所述检索査询请求中 包含任务需求信息;
接收所述服务器发送的检索结果, 所述检索结果中包含与所述任务需 求信息对应的功能模块的描述信息;
所述采用所述功能模块,对本地功能进行扩展和 /或更新之后,还包括: 采用扩展和 /或更新后的本地功能对所述任务进行处理。
结合第一方面的第一种可能的实现方式, 在第一方面的第二种可能的 实现方式中,所述采用扩展和 /或更新后的本地功能对所述任务进行处理之 后, 还包括:
向服务器反馈所述任务处理得到的运行数据, 以使所述服务器根据所 述任务和与所述任务对应的运行数据对所述功能模块进行更新。
本发明的第二方面提供了一种终端终生学习处理方法, 包括: 接收终端发送的功能模块的下载请求, 所述下载请求包含所述功能模 块的描述信息;
向所述终端发送与所述描述信息对应的功能模块, 以使所述终端采用 所述功能模块, 对本地功能进行扩展和 /或更新。
在第二方面的第一种可能的实现方式中, 所述接收终端发送的功能模 块的下载请求之前, 还包括:
根据任务和与所述任务对应的运行数据创建功能模块;
存储所述功能模块与所述功能模块的描述信息之间的对应关系。
结合第二方面的第一种可能的实现方式, 在第二方面的第二种可能的 实现方式中, 所述存储所述功能模块与所述功能模块的描述信息之间的对 应关系之后, 还包括:
接收终端发送的任务和与所述任务对应的运行数据;
根据所述任务和所述运行数据, 对所述功能模块进行更新。
结合第二方面至第二方面的第二种任一种可能的实现方式, 在第二方 面的第三种可能的实现方式中, 所述接收终端发送的功能模块的下载请求 之前, 还包括:
接收所述终端发送的检索査询请求, 所述检索査询请求中包含任务需 求信息;
向所述终端发送检索结果, 所述检索结果中包含与所述任务需求信息 对应的功能模块的描述信息。
本发明的第三方面提供了一种终端设备, 包括: 发送单元, 用于向服务器发送功能模块的下载请求, 所述下载请求包 含所述功能模块的描述信息;
接收单元, 用于接收服务器发送的与所述描述信息对应的功能模块; 本地更新单元, 用于采用所述功能模块,对本地功能进行扩展和 /或更 新。
在第三方面的第一种可能的实现方式中, 还包括:
检索査询单元, 用于在所述发送单元向服务器发送功能模块的下载请 求之前, 在任务建立之后, 向服务器发送检索査询请求, 所述检索査询请 求中包含任务需求信息;
接收所述服务器发送的检索结果, 所述检索结果中包含与所述任务需 求信息对应的功能模块的描述信息;
处理单元, 用于在所述本地更新单元采用所述功能模块, 对本地功能 进行扩展和 /或更新之后, 采用扩展和 /或更新后的本地功能对所述任务进 行处理。
结合第三方面的第一种可能的实现方式, 在第三方面的第二种可能的 实现方式中, 还包括:
反馈单元,用于在所述处理单元采用扩展和 /或更新后的本地功能对所 述任务进行处理之后, 向服务器反馈所述任务处理得到的运行数据, 以使 所述服务器根据所述任务和与所述任务对应的运行数据对所述功能模块 进行更新。
本发明的第四方面提供了一种服务器, 包括:
接收单元, 用于接收终端发送的功能模块的下载请求, 所述下载请求 包含所述功能模块的描述信息;
发送单元, 用于向所述终端发送与所述描述信息对应的功能模块, 以 使所述终端采用所述功能模块, 对本地功能进行扩展和 /或更新。
在第四方面的第一种可能的实现方式中, 还包括:
创建单元, 用于所述接收单元接收终端发送的功能模块的下载请求之 前, 根据任务和与所述任务对应的运行数据创建功能模块;
存储单元, 用于所述接收单元接收终端发送的功能模块的下载请求之 前, 存储所述功能模块与所述功能模块的描述信息之间的对应关系。 结合第四方面的第一种可能的实现方式, 在第四方面的第二种可能的 实现方式中, 还包括:
更新单元, 用于所述存储单元存储所述功能模块与所述功能模块的描 述信息之间的对应关系之后, 接收终端发送的任务和与所述任务对应的运 行数据;
根据所述任务和所述运行数据, 对所述功能模块进行更新。
结合第四方面至第四方面的第二种任一种可能的实现方式, 在第四方 面的第三种可能的实现方式中, 还包括:
检索处理单元, 用于所述接收单元接收终端发送的功能模块的下载请 求之前, 接收所述终端发送的检索査询请求, 所述检索査询请求中包含任 务需求信息;
向所述终端发送检索结果, 所述检索结果中包含与所述任务需求信息 对应的功能模块的描述信息。
本发明第五方面提供了一种终端终生学习处理***, 包括所述第三方 面至第三方面第二种可能实现的方式中任一项所述的终端设备以及所述 第四方面至第四方面第三种可能实现的方式中任一项所述的服务器。
本发明实施例终端终生学习处理方法、 装置和***, 通过终端下载服 务器根据任务和所述任务对应的运行数据持续进行更新的功能模块, 并采 用所述功能模块实现本地功能扩展和 /或更新,解决现有技术中终端无法进 行功能扩展和更新的问题。 实现了终端具有终生学习的能力。 附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案, 下面将对 实施例或现有技术描述中所需要使用的附图作一简单地介绍, 显而易见 地, 下面描述中的附图是本发明的一些实施例, 对于本领域普通技术人员 来讲, 在不付出创造性劳动性的前提下, 还可以根据这些附图获得其他的 附图。
图 1为本发明终端终生学习处理方法实施例一的流程图;
图 2为本发明终端终生学习处理方法实施例二的流程图;
图 3为本发明终端设备实施例一的结构示意图; 图 4为本发明终端设备实施例二的结构示意图;
图 5为本发明服务器实施例一的结构示意图;
图 6为本发明服务器实施例二的结构示意图;
图 7为本发明终端终生学习处理***实施例的结构示意图; 图 8为本发明终端设备实施例三的机构示意图;
图 9为本发明服务器实施例三的结构示意图。 具体实施方式
为使本发明实施例的目的、 技术方案和优点更加清楚, 下面将结合本发 明实施例中的附图, 对本发明实施例中的技术方案进行清楚、 完整地描述, 显然, 所描述的实施例是本发明一部分实施例, 而不是全部的实施例。 基于 本发明中的实施例, 本领域普通技术人员在没有作出创造性劳动前提下所获 得的所有其他实施例, 都属于本发明保护的范围。 图 1为本发明终端终生学习处理方法实施例一的流程图, 如图 1所示, 本实施例的方法可以包括: 歩骤 101、 向服务器发送功能模块的下载请求, 所述下载请求包含所述 功能模块的描述信息;
歩骤 102、 接收服务器发送的与所述描述信息对应的功能模块; 歩骤 103、 采用所述功能模块, 对本地功能进行扩展和 /或更新。 具体来说, 当终端在有新功能需求时, 可以根据自身的功能需求向服务 器发送包含有对应的功能模块描述信息的请求, 请求下载所述功能模块。 也 可以通过设置终端, 定时向服务器发送包含有对应的功能模块描述信息的下 载请求。 功能模块的描述信息可以是功能模块的名称和功能模块的类型等用 于描述功能模块的信息。 终端接收服务器发送的与所述描述信息对应的功能 模块, 采用下载后的功能模块对本地功能进行扩展或者是对本地已有的功能 模块进行更新。 使得终端具备对用户的行为进行分析、 图像识别等终生学习 的能力。 并且除从服务器端获取更新的功能模块外, 终端还可以根据自身收 集的数据对功能模块进行独立更新, 提升自身智能性。 进一歩地, 所述向服务器发送功能模块的下载请求之前, 还包括: 在任务建立之后, 向服务器发送检索査询请求, 所述检索査询请求中包 含任务需求信息; 接收所述服务器发送的检索结果, 所述检索结果中包含与所述任务需求 信息对应的功能模块的描述信息; 所述采用所述功能模块, 对本地功能进行扩展和 /或更新之后, 还包括: 采用扩展和 /或更新后的本地功能对所述任务进行处理。 具体来说, 在根据功能需求建立任务后, 终端想査询关于其功能需求的 功能模块, 可以通过向服务器发送检索査询, 来确定服务器是否有相应的功 能模块, 若有, 还可从服务器反馈的检索结果中获取所述功能模块的描述信 息, 用于终端向服务器发送功能模块的下载请求。 终端下载所述功能模块之 后, 用该功能模块完成本地功能扩展和 /或更新, 用扩展和 /或更新后的本地功 能模块对所述任务进行处理。 举例说明, 智能手机中不具有和该智能手机的 用户相匹配的个性化推荐功能或者该智能手机中的个性化推荐功能模块需要 更新, 所述智能手机可向服务器发送 "推荐"关键词进行检索査询, 用以检 索査询有关于推荐的功能模块, 智能手机接收服务器发送的检索结果, 该检 索结果包含有该类功能模块的名称和相对应的类型。 终端将功能模块下载到 本地, 并采用所述功能模块实现对该智能手机用户推荐电影、 书籍、 音乐等 功能。
进一歩地, 所述采用扩展和 /或更新后的本地功能对所述任务进行处理之 后, 还包括: 向服务器反馈所述任务处理得到的运行数据, 以使所述服务器根据所述 任务和与所述任务对应的运行数据对所述功能模块进行更新。 具体来说, 终端在采用扩展和 /或更新后的本地功能模块对所述任务进行 处理, 在对所述任务进行处理的过程当中相应地需要终端提供所述功能模块 运行数据, 终端可将在处理任务之后把所述运行数据上传给服务器, 以使所 述服务器根据所述任务及其对应的运行数据对所述功能模块进行更新。 通过 终端上传到服务器的运行数据进行更新的功能模块对于终端而言更具有针对 性。 本实施例, 通过终端向服务器发送功能模块检索査询和下载请求, 并采 用下载后的功能模块对本地功能进行扩展和 /或更新, 并且还可以通过终端向 服务器上传任务处理得到的运行数据对服务器端的功能模块进行更新, 解决 现有技术中终端无法进行功能扩展和 /或更新的问题。 实现了终端具有终生学 习的能力。
图 2为本发明终端终生学习处理方法实施例二的流程图, 如图 2所示, 本实施例的方法可以包括:
歩骤 201、 接收终端发送的功能模块的下载请求, 所述下载请求包含所 述功能模块的描述信息;
歩骤 202、 向所述终端发送与所述描述信息对应的功能模块, 以使所述 终端采用所述功能模块, 对本地功能进行扩展和 /或更新。 具体来说, 服务器根据终端发送的功能模块下载请求中所包含的功能模 块的描述信息向终端发送与描述信息对应的功能模块, 以使所述终端对本地 功能进行扩展和 /或更新。
进一歩地, 所述接收终端发送的功能模块的下载请求之前, 还包括: 根据任务和与所述任务对应的运行数据创建功能模块; 存储所述功能模块与所述功能模块的描述信息之间的对应关系。 具体来说, 服务器需要根据任务以及与所述对应的运行数据创建功能模 块, 该功能模块为初始功能模块。创建初始功能模块可以通过终生学习算法, 如分类、 聚类、 回归、 以及规则等。 功能模块的类型可根据任务确定, 对应 为构建整体模型、 组别模型或者个体模型等。 学习得到的功能模块通过索引 存储到功能模块库中。 索引对应功能模块的描述信息。 大量的功能模块进行 存储和检索, 需要分析不同类型的功能模块的数据结构, 分别进行有效地索 弓 1。 例如对稠密向量、 稀疏向量和树状分类器需要不同的索引机制。 做到按 id, 组群, 相似度进行检索。 本实施例中使用的功能模块存储涉及到大规模 稠密向量、 稀疏向量、 树、 图的存储和检索。 对于稠密向量, 涉及 KDTree、 ConeTree, BBTree等方法, 对稀疏向量, 涉及倒排表、 PPJoin、 Spilltree等方 法; 对于树形式的功能模块, 可以把树转化为规则、 然后使用字符串匹配, 也可以把树看成简单的图, 直接进行图的匹配。 所述功能模块对应于服务器 端统计或终生学习的结果模型, 用于客户端完成智能预测等功能。 需要说明的是, 本实施例并不限功能模块创建和存储的具体实现 进一歩地, 所述存储所述功能模块与所述功能模块的描述信息之间的对 应关系之后, 还包括: 接收终端发送的任务和与所述任务对应的运行数据; 根据所述任务和所述运行数据, 对所述功能模块进行更新。 具体来说, 服务器更新功能模块可根据终端发送的任务和该任务对应的 运行数据, 针对于同一个任务, 该数据可以是随时间的推移, 批量增长的, 也可以是通过终端单词反馈到服务器的。 功能模块的更新算法通过统计和终 生学习。例如: 对于相同任务所对应的功能模块更新可采用在线学习来完成, 对于不同的任务所对应的功能模块可采用迁移学习来完成。 还包括多任务学 习、 终身学习等机制都可用于完成功能模块的更新。 统计和终生学习可包括 相关的统计和终生学习算法, 如 kmeans、 svm、 朴素贝叶斯、 决策树等。 迁 移学习设计的方法可包括基于多任务的学习、 TrAdaboost等方法。 终身学习 可包括深度学习、 ELLA 算法等。 通过终身学习机制对功能模块的更新, 使 得功能模块越来越准确、 越来越智能。 需要说明的是, 本实施例并不限定功 能模块更新的具体实现。 进一歩地, 所述接收终端发送的功能模块的下载请求之前, 还包括: 接收所述终端发送的检索査询请求, 所述检索査询请求中包含任务需求
I Ή自、 ., 向所述终端发送检索结果, 所述检索结果中包含与所述任务需求信息对 应的功能模块的描述信息。 具体来说, 服务器接收终端发送的检索査询请求, 向终端发送终端所检 索的任务需求信息所对应的功能模块的描述信息。 所述功能模块的描述信息 可以是功能模块的名称和功能模块的类型等用于描述功能模块的信息。 本实施例, 通过服务器根据任务和所述任务对应的运行数据创建功能模 块, 存储所述功能模块至功能模块库, 根据任务和所述任务对应的运行数据 对所述功能模块进行更新, 并接收终端的发送的检索査询和下载请求, 向终 端发送相应的功能模块。 以使所述终端采用所述功能模块, 对本地功能进行 扩展和 /或更新。 实现了终端具有终生学习的能力。 解决了现有技术中终端无 法进行功能扩展和 /或更新的问题。
图 3为本发明终端设备实施例一的结构示意图, 如图 3所示, 本实施例 的终端设备可以包括:
发送单元 101, 用于向服务器发送功能模块的下载请求, 所述下载请求 包含所述功能模块的描述信息;
接收单元 102, 用于接收服务器发送的与所述描述信息对应的功能模块; 本地更新单元 103, 用于采用所述功能模块, 对本地功能进行扩展和 /或 更新。
本实施例的终端, 可以用于执行图 1所示方法实施例的技术方案, 其实 现原理和技术效果类似, 此处不再赘述。
图 4为本发明终端设备实施例二的结构示意图, 如图 4所示, 本实施例 的终端设备在图 3所示装置结构的基础上, 进一歩地, 还可以包括: 检索査询单元 104、 用于在所述发送单元向服务器发送功能模块的下载 请求之前, 在任务建立之后, 向服务器发送检索査询请求, 所述检索査询请 求中包含任务需求信息; 接收所述服务器发送的检索结果, 所述检索结果中包含与所述任务需求 信息对应的功能模块的描述信息; 处理单元 105、 用于在所述本地更新单元采用所述功能模块, 对本地功 能进行扩展和 /或更新之后,采用扩展和 /或更新后的本地功能对所述任务进行 处理;
反馈单元 106、 用于在所述处理单元采用扩展和 /或更新后的本地功能对 所述任务进行处理之后, 向服务器反馈所述任务处理得到的运行数据, 以使 所述服务器根据所述任务和与所述任务对应的运行数据对所述功能模块进行 更新。
本实施例的装置, 可以用于执行图 1所示方法实施例的技术方案, 其实 现原理和技术效果类似, 此处不再赘述。
图 5为本发明服务器实施例一的结构示意图, 如图 5所示, 本实施例的 服务器可以包括:
接收单元 201, 用于接收终端发送的功能模块的下载请求, 所述下载请 求包含所述功能模块的描述信息;
发送单元 202, 用于向所述终端发送与所述描述信息对应的功能模块, 以使所述终端采用所述功能模块, 对本地功能进行扩展和 /或更新。
本实施例的装置, 可以用于执行图 2所示方法实施例的技术方案, 其实 现原理和技术效果类似, 此处不再赘述。
图 6为本发明服务器实施例二的结构示意图, 如图 6所示, 本实施例的 服务器在图 5所示装置结构的基础上, 进一歩地, 还可以包括: 创建单元 203、 用于所述接收单元接收终端发送的功能模块的下载请求 之前, 根据任务和与所述任务对应的运行数据创建功能模块;
存储单元 204、 用于所述接收单元接收终端发送的功能模块的下载请求 之前, 存储所述功能模块与所述功能模块的描述信息之间的对应关系。
更新单元 205、 用于所述存储单元存储所述功能模块与所述功能模块的 描述信息之间的对应关系之后, 接收终端发送的任务和与所述任务对应的运 行数据;
根据所述任务和所述运行数据, 对所述功能模块进行更新
检索处理单元 206、 用于所述接收单元接收终端发送的功能模块的下载 请求之前, 接收所述终端发送的检索査询请求, 所述检索査询请求中包含任 务需求信息;
向所述终端发送检索结果, 所述检索结果中包含与所述任务需求信息对 应的功能模块的描述信息。
本实施例的装置, 可以用于执行图 2所示方法实施例的技术方案, 其实 现原理和技术效果类似, 此处不再赘述。
图 7为本发明终端终生学习处理***实施例的结构示意图,如图 7所示, 本实施例的终端终生学习处理***可以包括: 终端设备 301、服务器 302。本实施例的终端设备 301与图 1所示的本发 明终端终生学习处理方法实施例的原理相同, 不再赘述; 本实施例的服务器 与图 2所示的本发明的终端终生学习处理方法实施例的原理相同,不再赘述。 图 8为本发明终端设备实施例三的机构示意图, 如图 8所示, 本实施例 的终端设备, 包括: 处理器 401 和接口电路 402, 图中还示出了存储器 403 和总线 404, 该处理器 401、接口电路 402和存储器 403通过总线 404连接并 完成相互间的通信。
该总线 404可以是工业标准架构 (英文: Industry Standard Architecture, 简称: ISA) 总线、 外部设备互连(英文: Peripheral Component Interconnect, 简称: PCI)总线或内部整合电路(英文: Inter-Integrated Circuit, 简称: I2C) 总线等。 该总线 404可以分为地址总线、 数据总线、 控制总线等。 为便于表 示, 图 4中仅用一条粗线表示, 但并不表示仅有一根总线或一种类型的总线。 存储器 403用于存储可执行程序代码,该程序代码包括计算机操作指令。 存储器 403可以是易失性存储器(英文: volatile memory) ,例如随机存储器 (英 文: random-access memory, 简称: RAM), 也可以是非易失性存储器 (英文: non-volatile memory,简称: NVM) , ί列 卩只读存储器(英文: read-only memory, 简称 ROM) ,快闪存储器(英文: flash memory) ,硬盘(英文: hard disk drive, 简称 HDD) 或固态硬盘 (英文: solid-state drive, 简称 SSD)。 处理器 401可以是中央处理器(英文: central processing unit,简称: CPU)。 其中,处理器 401可以调用存储器 403中存储的操作指令或者程序代码, 来执行本发明实施例提供的终端终生学习处理方法, 所述方法包括: 处理器 401 向服务器发送功能模块的下载请求, 所述下载请求包含所述 功能模块的描述信息;
处理器 401通过接口电路 402接收服务器发送的与所述描述信息对应的 功能模块;
处理器 401采用所述功能模块, 对本地功能进行扩展和 /或更新。 处理器 401在任务建立之后, 向服务器发送检索査询请求, 所述检索査 询请求中包含任务需求信息;
处理器 401通过接口电路 402接收所述服务器发送的检索结果, 所述检 索结果中包含与所述任务需求信息对应的功能模块的描述信息;
处理器 401所述采用所述功能模块,对本地功能进行扩展和 /或更新之后, 还包括: 处理器 401采用扩展和 /或更新后的本地功能对所述任务进行处理。 处理器 401 向服务器反馈所述任务处理得到的运行数据, 以使所述服务 器根据所述任务和与所述任务对应的运行数据对所述功能模块进行更新。
图 9为本发明服务器实施例三的机构示意图, 如图 9所示, 本实施例的 终端,包括:处理器 501和接口电路 52,图中还示出了存储器 503和总线 504, 该处理器 501、 接口电路 502和存储器 503通过总线 504连接并完成相互间 的通信。
该总线 504可以是工业标准架构 (英文: Industry Standard Architecture, 简称: ISA) 总线、 外部设备互连(英文: Peripheral Component Interconnect, 简称: PCI)总线或内部整合电路(英文: Inter-Integrated Circuit, 简称: I2C) 总线等。 该总线 504可以分为地址总线、 数据总线、 控制总线等。 为便于表 示, 图 9中仅用一条粗线表示, 但并不表示仅有一根总线或一种类型的总线。 存储器 503用于存储可执行程序代码,该程序代码包括计算机操作指令。 存储器 503可以是易失性存储器(英文: volatile memory) ,例如随机存储器 (英 文: random-access memory, 简称: RAM), 也可以是非易失性存储器(英文: non-volatile memory,简称: NVM) , ί列 卩只读存储器(英文: read-only memory, 简称 ROM) ,快闪存储器(英文: flash memory) ,硬盘(英文: hard disk drive, 简称 HDD) 或固态硬盘 (英文: solid-state drive, 简称 SSD)。 处理器 501可以是中央处理器(英文: central processing unit,简称: CPU)。 其中,处理器 501可以调用存储器 503中存储的操作指令或者程序代码, 来执行本发明实施例提供的终端终生学习处理方法, 所述方法包括:
处理器 501通过接口电路 502接收终端发送的功能模块的下载请求, 所 述下载请求包含所述功能模块的描述信息; 处理器 501 向所述终端发送与所述描述信息对应的功能模块, 以使所述 终端采用所述功能模块, 对本地功能进行扩展和 /或更新。
处理器 501根据任务和与所述任务对应的运行数据创建功能模块; 存储器 503存储所述功能模块与所述功能模块的描述信息之间的对应关 系。 处理器 501通过接口电路 502接收终端发送的任务和与所述任务对应的 运行数据;
处理器 501根据所述任务和所述运行数据, 对所述功能模块进行更新。 处理器 501通过接口电路 502接收所述终端发送的检索査询请求, 所述 检索査询请求中包含任务需求信息;
处理器 501 向所述终端发送检索结果, 所述检索结果中包含与所述任务 需求信息对应的功能模块的描述信息。
本领域普通技术人员可以理解: 实现上述各方法实施例的全部或部分歩 骤可以通过程序指令相关的硬件来完成。 前述的程序可以存储于一计算机可 读取存储介质中。 该程序在执行时, 执行包括上述各方法实施例的歩骤; 而 前述的存储介质包括: ROM、 RAM, 磁碟或者光盘等各种可以存储程序代码 的介质。 最后应说明的是: 以上各实施例仅用以说明本发明的技术方案, 而非对 其限制; 尽管参照前述各实施例对本发明进行了详细的说明, 本领域的普通 技术人员应当理解:其依然可以对前述各实施例所记载的技术方案进行修改, 或者对其中部分或者全部技术特征进行等同替换; 而这些修改或者替换, 并 不使相应技术方案的本质脱离本发明各实施例技术方案的范围。

Claims

权 利 要 求 书
1、 一种终端终生学习处理方法, 其特征在于, 包括: 终端向服务器发送功能模块的下载请求, 所述下载请求包含所述功能模 块的描述信息; 终端接收服务器发送的与所述描述信息对应的功能模块; 终端采用所述功能模块, 对本地功能进行扩展和 /或更新。
2、 根据权利要求 1所述的方法, 其特征在于, 所述向服务器发送功能模 块的下载请求之前, 还包括:
在任务建立之后, 向服务器发送检索査询请求, 所述检索査询请求中包 含任务需求信息; 接收所述服务器发送的检索结果, 所述检索结果中包含与所述任务需求 信息对应的功能模块的描述信息;
所述采用所述功能模块, 对本地功能进行扩展和 /或更新之后, 还包括: 采用扩展和 /或更新后的本地功能对所述任务进行处理。
3、 根据权利要求 2所述的方法, 其特征在于, 所述采用扩展和 /或更新 后的本地功能对所述任务进行处理之后, 还包括: 向服务器反馈所述任务处理得到的运行数据, 以使所述服务器根据所述 任务和与所述任务对应的运行数据对所述功能模块进行更新。
4、 一种终端终生学习处理方法, 其特征在于, 包括: 接收终端发送的功能模块的下载请求, 所述下载请求包含所述功能模块 的描述信息; 向所述终端发送与所述描述信息对应的功能模块, 以使所述终端采用所 述功能模块, 对本地功能进行扩展和 /或更新。
5、 根据权利要求 4所述的方法, 其特征在于, 所述接收终端发送的功能 模块的下载请求之前, 还包括: 根据任务和与所述任务对应的运行数据创建功能模块; 存储所述功能模块与所述功能模块的描述信息之间的对应关系。
6、 根据权利要求 5所述的方法, 其特征在于, 所述存储所述功能模块与 所述功能模块的描述信息之间的对应关系之后, 还包括: 接收终端发送的任务和与所述任务对应的运行数据; 根据所述任务和所述运行数据, 对所述功能模块进行更新。
7、 根据权利要求 4-6中任一项所述的方法, 其特征在于, 所述接收终端 发送的功能模块的下载请求之前, 还包括: 接收所述终端发送的检索査询请求, 所述检索査询请求中包含任务需求
I Ή自、 ., 向所述终端发送检索结果, 所述检索结果中包含与所述任务需求信息对 应的功能模块的描述信息。
8、 一种终端设备, 其特征在于, 包括: 发送单元, 用于向服务器发送功能模块的下载请求, 所述下载请求包含 所述功能模块的描述信息; 接收单元, 用于接收服务器发送的与所述描述信息对应的功能模块; 本地更新单元,用于采用所述功能模块,对本地功能进行扩展和 /或更新。
9、 根据权利要求 8所述的终端设备, 其特征在于, 还包括: 检索査询单元, 用于在所述发送单元向服务器发送功能模块的下载请求 之前, 在任务建立之后, 向服务器发送检索査询请求, 所述检索査询请求中 包含任务需求信息, 接收所述服务器发送的检索结果, 所述检索结果中包含 与所述任务需求信息对应的功能模块的描述信息; 处理单元, 用于在所述本地更新单元采用所述功能模块, 对本地功能进 行扩展和 /或更新之后, 采用扩展和 /或更新后的本地功能对所述任务进行处 理。
10、 根据权利要求 9所述的终端设备, 其特征在于, 还包括: 反馈单元, 用于在所述处理单元采用扩展和 /或更新后的本地功能对所述 任务进行处理之后, 向服务器反馈所述任务处理得到的运行数据, 以使所述 服务器根据所述任务和与所述任务对应的运行数据对所述功能模块进行更 新。
11、 一种服务器, 其特征在于, 包括:
接收单元, 用于接收终端发送的功能模块的下载请求, 所述下载请求包 含所述功能模块的描述信息; 发送单元, 用于向所述终端发送与所述描述信息对应的功能模块, 以使 所述终端采用所述功能模块, 对本地功能进行扩展和 /或更新。
12、 根据权利要求 11所述的服务器, 其特征在于, 还包括: 创建单元,用于所述接收单元接收终端发送的功能模块的下载请求之前, 根据任务和与所述任务对应的运行数据创建功能模块; 存储单元,用于所述接收单元接收终端发送的功能模块的下载请求之前, 存储所述功能模块与所述功能模块的描述信息之间的对应关系。
13、 根据权利要求 12所述的服务器, 其特征在于, 还包括: 更新单元, 用于所述存储单元存储所述功能模块与所述功能模块的描述 信息之间的对应关系之后, 接收终端发送的任务和与所述任务对应的运行数 据; 根据所述任务和所述运行数据, 对所述功能模块进行更新。
14、 根据权利要求 11-13中任一项所述的服务器, 其特征在于, 还包括: 检索处理单元, 用于所述接收单元接收终端发送的功能模块的下载请求 之前, 接收所述终端发送的检索査询请求, 所述检索査询请求中包含任务需 求信息; 向所述终端发送检索结果, 所述检索结果中包含与所述任务需求信息对 应的功能模块的描述信息。
15、 一种终端终生学习处理***, 其特征在于, 包括权利要求 8-10中任 一项所述的终端设备以及权利要求 11-14中任一项所述的服务器。
PCT/CN2014/082698 2013-12-18 2014-07-22 终端终生学习处理方法、装置和*** WO2015090062A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US15/181,664 US10078509B2 (en) 2013-12-18 2016-06-14 Method and system for processing lifelong learning of terminal and apparatus

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201310699881.6 2013-12-18
CN201310699881.6A CN104731607B (zh) 2013-12-18 2013-12-18 终端终生学习处理方法、装置和***

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US15/181,664 Continuation US10078509B2 (en) 2013-12-18 2016-06-14 Method and system for processing lifelong learning of terminal and apparatus

Publications (1)

Publication Number Publication Date
WO2015090062A1 true WO2015090062A1 (zh) 2015-06-25

Family

ID=53402065

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2014/082698 WO2015090062A1 (zh) 2013-12-18 2014-07-22 终端终生学习处理方法、装置和***

Country Status (3)

Country Link
US (1) US10078509B2 (zh)
CN (1) CN104731607B (zh)
WO (1) WO2015090062A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106126269A (zh) * 2016-06-07 2016-11-16 中国建设银行股份有限公司 应用更新方法、装置、***及应用设计方法
EP3402160A4 (en) * 2016-03-15 2018-12-12 Huawei Technologies Co., Ltd. Service processing method and apparatus

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105812178B (zh) * 2016-03-08 2019-02-26 青岛海信移动通信技术股份有限公司 一种终端升级方法及终端
CN106020843A (zh) * 2016-05-31 2016-10-12 乐视控股(北京)有限公司 一种***功能的更新方法、***、终端设备及服务器
CN108460523B (zh) * 2018-02-12 2020-08-21 阿里巴巴集团控股有限公司 一种风控规则生成方法和装置
CN109242109B (zh) * 2018-07-06 2022-05-10 网宿科技股份有限公司 深度模型的管理方法及服务器
US11580384B2 (en) 2018-09-27 2023-02-14 GE Precision Healthcare LLC System and method for using a deep learning network over time
CN110020966A (zh) * 2019-04-11 2019-07-16 北京百度网讯科技有限公司 应用于智能电网的数据处理方法和装置
CN113205159B (zh) * 2021-05-31 2022-04-15 西南交通大学 一种知识迁移方法、无线网络设备个体识别方法及***
CN113465925B (zh) * 2021-06-26 2023-10-20 嘉兴学院 一种基于tca的变工况轴承状态诊断方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20080037450A (ko) * 2006-10-26 2008-04-30 웹싱크 주식회사 소프트웨어 실행모듈의 업데이트 시스템 및 방법
CN101616018A (zh) * 2008-06-27 2009-12-30 中兴通讯股份有限公司 网管软件升级的方法及装置
CN101656949A (zh) * 2009-09-21 2010-02-24 中兴通讯股份有限公司 一种固件空中升级的方法及终端
CN102236563A (zh) * 2010-04-28 2011-11-09 腾讯科技(深圳)有限公司 软件升级的方法及***
CN102932441A (zh) * 2012-10-26 2013-02-13 北京小米科技有限责任公司 一种下载应用的方法、终端、服务器及***
CN103067509A (zh) * 2012-12-30 2013-04-24 乐视网信息技术(北京)股份有限公司 一种模块化软件升级***及其方法

Family Cites Families (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1239366B1 (en) * 2000-12-29 2006-03-08 STMicroelectronics S.r.l. Method for expanding in friendly manner the functionality of a portable electronic device and corresponding portable electronic device
US20030171907A1 (en) * 2002-03-06 2003-09-11 Shay Gal-On Methods and Apparatus for Optimizing Applications on Configurable Processors
EP1484716A1 (en) 2003-06-06 2004-12-08 Sony France S.A. An architecture for self-developing devices
US8555273B1 (en) * 2003-09-17 2013-10-08 Palm. Inc. Network for updating electronic devices
US20080148248A1 (en) * 2006-12-15 2008-06-19 Michael Volkmer Automatic software maintenance with change requests
US20080288255A1 (en) 2007-05-16 2008-11-20 Lawrence Carin System and method for quantifying, representing, and identifying similarities in data streams
US8311960B1 (en) 2009-03-31 2012-11-13 Emc Corporation Interactive semi-supervised machine learning for classification
US9003387B2 (en) * 2009-09-25 2015-04-07 Fisher-Rosemount Systems, Inc. Automated deployment of computer-specific software updates
JP5418442B2 (ja) * 2010-08-18 2014-02-19 株式会社リコー 情報処理装置、電子機器、更新処理支援方法、更新処理支援プログラム、及び更新処理支援システム
KR101164999B1 (ko) * 2010-12-07 2012-07-13 주식회사에이메일 모바일 애플리케이션 분석과 대응하는 서비스정보 제공 시스템 및 그 방법
GB2511221B (en) 2011-12-09 2020-09-23 Veracyte Inc Methods and compositions for classification of samples
US9710247B2 (en) * 2012-06-26 2017-07-18 Microsoft Technology Licensing, Llc Identification of host-compatible downloadable applications
CN102857555A (zh) * 2012-08-03 2013-01-02 深圳中兴网信科技有限公司 终端、服务器和软件升级方法
US9208063B1 (en) * 2013-02-21 2015-12-08 Groupon, Inc. Method for testing mobile application and associated apparatus and system
US9361085B2 (en) * 2013-03-18 2016-06-07 Cloudmask Systems and methods for intercepting, processing, and protecting user data through web application pattern detection
US9417866B2 (en) * 2014-12-03 2016-08-16 Verizon Patent And Licensing Inc. Identification and isolation of incompatible applications during a platform update

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20080037450A (ko) * 2006-10-26 2008-04-30 웹싱크 주식회사 소프트웨어 실행모듈의 업데이트 시스템 및 방법
CN101616018A (zh) * 2008-06-27 2009-12-30 中兴通讯股份有限公司 网管软件升级的方法及装置
CN101656949A (zh) * 2009-09-21 2010-02-24 中兴通讯股份有限公司 一种固件空中升级的方法及终端
CN102236563A (zh) * 2010-04-28 2011-11-09 腾讯科技(深圳)有限公司 软件升级的方法及***
CN102932441A (zh) * 2012-10-26 2013-02-13 北京小米科技有限责任公司 一种下载应用的方法、终端、服务器及***
CN103067509A (zh) * 2012-12-30 2013-04-24 乐视网信息技术(北京)股份有限公司 一种模块化软件升级***及其方法

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3402160A4 (en) * 2016-03-15 2018-12-12 Huawei Technologies Co., Ltd. Service processing method and apparatus
US11647108B2 (en) 2016-03-15 2023-05-09 Huawei Technologies Co., Ltd. Service processing method and apparatus
CN106126269A (zh) * 2016-06-07 2016-11-16 中国建设银行股份有限公司 应用更新方法、装置、***及应用设计方法

Also Published As

Publication number Publication date
CN104731607A (zh) 2015-06-24
CN104731607B (zh) 2018-10-30
US10078509B2 (en) 2018-09-18
US20160299755A1 (en) 2016-10-13

Similar Documents

Publication Publication Date Title
WO2015090062A1 (zh) 终端终生学习处理方法、装置和***
US11916764B1 (en) Server-side operations for edge analytics
US9830526B1 (en) Generating image features based on robust feature-learning
US11455306B2 (en) Query classification and processing using neural network based machine learning
WO2019216938A1 (en) Application development platform and software development kits that provide comprehensive machine learning services
US20170161641A1 (en) Streamlined analytic model training and scoring system
Zhang et al. Collective classification via discriminative matrix factorization on sparsely labeled networks
CN107301195B (zh) 生成用于搜索内容的分类模型方法、装置和数据处理***
US11687527B2 (en) System and method for analysis of graph databases using intelligent reasoning systems
EP3210133A1 (en) Tagging personal photos with deep networks
RU2637472C2 (ru) Способ, устройство и терминал для поиска данных
WO2021042763A1 (en) Image searches based on word vectors and image vectors
US20220147878A1 (en) Method for approximate k-nearest-neighbor search on parallel hardware accelerators
US9600540B1 (en) Decision processing applied to data analytics workflow
US11469974B1 (en) Analytics for edge devices to intelligently throttle data reporting
US11960517B2 (en) Dynamic cross-platform ask interface and natural language processing model
US20200065654A1 (en) Neural network fusion apparatus and modular neural network fusion method and matching interface generation method for the same
US9418343B2 (en) Multistage learner for efficiently boosting large datasets
CN117170648A (zh) 一种机器人流程自动化组件推荐方法、装置、设备及存储介质
US11204966B2 (en) Contextual image-assisted search
US10409916B2 (en) Natural language processing system
US20210319050A1 (en) Auto-completion based on content similarities
CN112800181A (zh) 文本检索方法、装置、计算机设备和存储介质
JP2023527188A (ja) 自動機械学習:統合され、カスタマイズ可能、及び拡張可能なシステム
US20190243910A1 (en) Visual Search as a Service

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14871220

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 14871220

Country of ref document: EP

Kind code of ref document: A1