-
公开(公告)号：KR1020130136812A
公开(公告)日：2013-12-13
申请号：KR1020120060507
申请日：2012-06-05
Applicant: 목포대학교산학협력단
IPC: G10L19/02
Abstract: The present invention relates to a support vector machine (SVM)-based speech and music classifier of a hierarchical structure for a selectable mode vocoder (SMV) codec, a SVM-based speech and music classifier according to the present invention facilitates implementation of a classifier into an embedded system by reducing the amount of operation while maintaining the classification performance of an SVM-based speech/music classifier. According to the present invention the amount of operation accompanied by an SVM-based classifier using an existing SMV codec classifier in front of the SVM-based speech/music classifier for an SMV codec, in particular, the implementation of a classifier into an embedded system is easily done.
Abstract translation: 本发明涉及一种用于可选模式声码器（SMV）编解码器的分级结构的基于支持向量机（SVM）的语音和音乐分类器，根据本发明的基于SVM的语音和音乐分类器有助于实现分类器 通过在保持基于SVM的语音/音乐分类器的分类性能的同时减少操作量而进入嵌入式系统。 根据本发明，使用SMV编解码器的基于SVM的语音/音乐分类器前面的现有SMV编解码器分类器的基于SVM的分类器的操作量，特别是将分类器实现为嵌入式系统 很容易完成
-
公开(公告)号：KR1020130037921A
公开(公告)日：2013-04-17
申请号：KR1020110102460
申请日：2011-10-07
Applicant: 목포대학교산학협력단
Abstract: PURPOSE: A method for implementing an efficient embedded system of an SVM-based sound classifier proper for a marine environment is provided to allocate different transmission rates depending on a voice signal type by accurately distinguishing the voice signal type proper for the marine environment. CONSTITUTION: An embedded system obtains two or more support vectors by learning. The embedded system checks a contribution level of each support vector. The embedded system performs discriminant computation by an SVM(Support Vector Machine) algorithm except for a support vector having a lower contribution level than a preset level.
Abstract translation: 目的：提供一种实现适用于海洋环境的基于SVM的声音分类器的高效嵌入式系统的方法，以通过准确区分适合海洋环境的语音信号类型，根据语音信号类型分配不同的传输速率。 构成：嵌入式系统通过学习获得两个或更多的支持向量。 嵌入式系统检查每个支持向量的贡献级别。 嵌入式系统通过SVM（支持向量机）算法执行判别式计算，除了具有比预设级别低的贡献级别的支持向量。
-
公开(公告)号：KR1020130037050A
公开(公告)日：2013-04-15
申请号：KR1020110101371
申请日：2011-10-05
Applicant: 목포대학교산학협력단
CPC classification number: H04W72/087 , H04L47/14 , H04L2012/6448 , H04W84/12
Abstract: PURPOSE: An SoQ(Service of Quality) based cooperative communication protocol implementation method is provided to connect an installed system by having compatibility with a WiMedia D-MAC(Distributed-Medium Access Control) standard technology without modifying a specific system. CONSTITUTION: In case it is possible to reserve MAS S-R and MAS R-T for random relay nodes, a reservation host node sets up a reservation reason code of a DRP IE for MAS S-T as relay approval and transmits the reservation reason code to a reservation target node. The reservation host node sets up a reservation detail state code of the DRP IE for MAS S-R as relay request and transmits the reservation detail state code to the relay node. The relay node sets up the reservation detail state code of the DRP IE for MAS R-T as relay notification and transmits the reservation detail code to the reservation target node. In case the reservation detail code of the DRP IE received from the relay node and the reservation target node is relay approval, the reservation host node transmits data through the relay node.
Abstract translation: 目的：提供基于协作通信协议实现方法的SoQ（质量服务），通过与WiMedia D-MAC（分布式媒体访问控制）标准技术兼容而不修改特定系统来连接已安装的系统。 规定：如果可以为随机中继节点保留MAS SR和MAS RT，则预约主机节点将用于MAS ST的DRP IE的预留原因代码设置为中继许可，并将预约原因代码发送到预约目标节点 。 预留主机节点为MAS S-R作为中继请求建立DRP IE的预留详细状态码，并将预约详细状态码发送给中继节点。 中继节点将用于MAS R-T的DRP IE的预留详细状态码设置为中继通知，并将预约详细代码发送到预约目标节点。 在从中继节点和预约目标节点接收的DRP IE的预留详细代码为中继许可的情况下，预约主机节点通过中继节点发送数据。
-
公开(公告)号：KR1020130022784A
公开(公告)日：2013-03-07
申请号：KR1020110085689
申请日：2011-08-26
Applicant: 목포대학교산학협력단
CPC classification number: H04L47/56 , H04L47/522
Abstract: PURPOSE: A resource allocation method for a network in a ship is provided to relatively divide delay time guaranteed by each sub-class according to a buffer size required by differentiating the amount of a bandwidth. CONSTITUTION: The amount of a bandwidth allocated in each AS(Assured Service) sub-class is differentiated. Delay time guaranteed by each AS sub-class according to a requested buffer size is divided. An RIO(Random early detection with In and Out) variable value is set based on a buffer size determined according to a network topology and a ratio of the bandwidth. The size of an output link bandwidth, which is reference for access control, is set by receiving the RIO variable value.
Abstract translation: 目的：提供一种用于船舶网络的资源分配方法，以根据区分带宽量所需的缓冲区大小相对分割每个子类保证的延迟时间。 规定：每个AS（Assured Service）子类中分配的带宽量是有差异的。 每个AS子类根据请求的缓冲区大小保证的延迟时间被划分。 基于根据网络拓扑和带宽比率确定的缓冲区大小设置RIO（随机早期检测，具有In和Out）变量值。 通过接收RIO变量值来设置作为访问控制参考的输出链路带宽的大小。
-
5.发明授权나이브 베이스 분류자와 퍼지 추론을 이용한 적조 발생 예측 방법 有权使用FUZZY推理和NAIVE BAYES CLASSIFIER的RED TIDE BLOOMS PREDICING METHOD
公开(公告)号：KR101145397B1
公开(公告)日：2012-05-15
申请号：KR1020110055571
申请日：2011-06-09
Applicant: 목포대학교산학협력단
IPC: G06N7/02
Abstract: PURPOSE: A red tide generation predicting method which uses a naive Bayes classifier and a fuzzy inference are provided to increase the accuracy of prediction rate of a red tide by using a naive Bayes classifier. CONSTITUTION: A marine environment data is normalized as learning data which is suitable for a fuzzy inference. A fuzzy reasoning rule is generated by using learning material(230). A red tide plankton density is predicted for input data. A red tide generation is predicted by estimating probability value based on a naive Bayes method(300). The red tide plankton density is controlled based on the prediction result.
Abstract translation: 目的：提出一种使用朴素贝叶斯分类器和模糊推理的红潮生成预测方法，通过使用朴素贝叶斯分类器提高红潮预测率的准确性。 规定：海洋环境数据被归一化为适合于模糊推理的学习数据。 通过使用学习材料（230）生成模糊推理规则。 预测输入数据的红潮浮游生物密度。 通过基于朴素贝叶斯方法估计概率值来预测红潮一代（300）。 基于预测结果控制红潮浮游生物密度。
-
Patent Agency Ranking6.发明公开선택 모드 보코더 코덱을 위한 서포트 백터의 기여도에 따른 서포트 백터 머신 기반 음성 및 음악 분류기의 간략화 방법 无效基于支持向量的贡献简化基于SVM的语音和音乐分类器的方法，用于可选择模式VOCODER编解码器
公开(公告)号：KR1020130136813A
公开(公告)日：2013-12-13
申请号：KR1020120060508
申请日：2012-06-05
Applicant: 목포대학교산학협력단
IPC: G10L19/02
Abstract: The present invention relates to a simplification method of a support vector machine-based speech and music classifier, a simplification method of a support vector machine-based speech and music classifier according to the present invention facilitates implementation of a classifier into an embedded system by reducing the amount of calculation while maintaining the performance of classification of a support vector machine (SVM)-based speech/music classifier for a selectable mode vocoder (SMV). According to the present invention is able to reduce the amount of operation which is followed by an SVM-based classifier by simplifying a support vector of the SVM-based speech/music classifier for an SMV codec, and in particular, the implementation of the classifier into an embedded system is easy. [Reference numerals] (AA) Calculating the average contribution of the support vector;(BB) Calculating the relativce contribution of the support vector;(CC) Removing the support vector having the smaller contribution compared to the threshold value
Abstract translation: 本发明涉及基于支持向量机的语音和音乐分类器的简化方法，根据本发明的基于支持向量机的语音和音乐分类器的简化方法有利于通过减少将分类器实现到嵌入式系统中 在维持用于可选模式声码器（SMV）的支持向量机（SVM）的语音/音乐分类器的分类的性能的同时计算的量。 根据本发明，通过简化用于SMV编解码器的基于SVM的语音/音乐分类器的支持向量，能够减少基于SVM的分类器所遵循的操作量，特别地，分类器的实现 嵌入式系统很容易。 （AA）计算支持向量的平均贡献;（BB）计算支持向量的相对贡献;（CC）去除与阈值相比具有较小贡献的支持向量
-
公开(公告)号：KR1020130044044A
公开(公告)日：2013-05-02
申请号：KR1020110108300
申请日：2011-10-21
Applicant: 목포대학교산학협력단
Abstract: PURPOSE: A cluster-based routing method for sensing an error in a vessel network is provided to continuously perform a data collection and delivery function when an unexpected error in a main cluster head causes suspension of operation. CONSTITUTION: A backup cluster head to replace a main cluster head is selected. The main cluster head and the backup cluster head receive data from a sensor node in a cluster in order to prepare for an error of the main cluster head. When an error occurs in the main cluster head, a base station makes the backup cluster head to take a role of the main cluster head. The backup cluster head collects the data from the sensor node. The backup cluster head merges the data. The backup cluster head transmits the merged data to the base station. [Reference numerals] (AA) Base station checks the failure of a main cluster head; (BB) Base station instructs a backup cluster head to act as the main cluster head; (CC) Backup cluster receive data from a sensor node as before; (DD) Merge the data when all sensor data are received; (EE) Backup cluster head transmits the merged data to the base station
Abstract translation: 目的：提供一种用于感测船舶网络中的错误的基于群集的路由方法，以便在主群集头中的意外错误导致操作暂停时，持续执行数据收集和传递功能。 规定：选择用于替换主集群头的备份集群头。 主集群头和备份集群头从集群中的传感器节点接收数据，以准备主集群头的错误。 当主群集头发生错误时，基站使备份簇头起到主簇头的作用。 备份集群头从传感器节点收集数据。 备份集群头合并数据。 备份簇头将合并的数据发送到基站。 （附图标记）（AA）基站检查主簇头的故障; （BB）基站指示备份簇头作为主簇头; （CC）备份集群像以前一样从传感器节点接收数据; （DD）接收所有传感器数据时合并数据; （EE）备份簇头将合并的数据发送到基站
-
公开(公告)号：KR1020130043960A
公开(公告)日：2013-05-02
申请号：KR1020110108173
申请日：2011-10-21
Applicant: 목포대학교산학협력단
IPC: G06F9/06
CPC classification number: G06F9/06
Abstract: PURPOSE: An implementation method of an efficient voice/music classifier proper to marine environment is provided to obtain input vectors from a cache memory of the classifier and to convert a source code of the classifier, thereby shortening execution time and saving energy consumption. CONSTITUTION: An input vector is obtained from a cache memory of a classifier. A source code of the classifier is converted. The source code of the classifier is converted by adding an intermediate value storage variable and the last classification value storage variable for processing input vectors at the same time. [Reference numerals] (a) Address of a support vector; (b) Content of a cache block 0 and a register n
Abstract translation: 目的：提供适用于海洋环境的高效语音/音乐分类器的实现方法，以从分类器的高速缓存中获取输入向量，并转换分类器的源代码，从而缩短执行时间并节省能耗。 构成：从分类器的高速缓冲存储器获得输入向量。 转换分类器的源代码。 分类器的源代码通过添加中间值存储变量和最后一个分类值存储变量进行转换，用于同时处理输入向量。 [附图标记]（a）支持向量的地址; （b）缓存块0和寄存器n的内容
-
公开(公告)号：KR1020130037040A
公开(公告)日：2013-04-15
申请号：KR1020110101359
申请日：2011-10-05
Applicant: 목포대학교산학협력단
Abstract: PURPOSE: A voice phishing detection method based on distinctive weighted value learning in marine/voice communication is provided to detect voice phishing without a separate feature vector extraction process by using only a parameter extracted in a process which changes a signal into a voice signal. CONSTITUTION: An SMV(Selectable Mode Vocoder) decoder comprises a new feature vector by using a feature vector which is extracted in a process which changes an LSF(Line Spectral Frequency) into an LPC(Linear Prediction Coding); and a feature vector extracted in a process which changes the obtained LPC into a voice signal. The SMV decoder calculates likelihood for the transmitted bitstream by using a GMM(Gaussian Mixture Model) and selects a flag for a value with the maximum likelihood.
Abstract translation: 目的：提供一种基于海洋/语音通信中特色加权值学习的语音网络钓鱼检测方法，通过仅使用将信号改变为语音信号的过程中提取的参数，无需单独的特征向量提取过程来检测语音网络钓鱼。 构成：SMV（可选模式声码器）解码器通过使用在将LSF（线谱频率）改变为LPC（线性预测编码）的处理中提取的特征向量来包括新特征向量; 以及在将所获得的LPC改变为语音信号的处理中提取的特征向量。 SMV解码器通过使用GMM（高斯混合模型）来计算所发送的比特流的可能性，并且选择具有最大似然度的值的标志。
-
公开(公告)号：KR1020130017134A
公开(公告)日：2013-02-20
申请号：KR1020110079388
申请日：2011-08-10
Applicant: 목포대학교산학협력단
Abstract: PURPOSE: An authentication key distribution method of a sensor network system is provided to reduce the transmission of unnecessary information by monitoring a network state. CONSTITUTION: A base station(110) stores sensor information by collecting the sensor information from one or more cluster heads(120). The base station calculates an authentication key distribution restriction distance by applying a fuzzy algorithm to the stored sensor information. The base station determines the necessity of authentication key distribution using the calculated authentication key distribution restriction distance. When the necessity of the authentication key distribution is requested, the cluster head transmits the authentication key using the received authentication key distribution restriction distance. [Reference numerals] (110) Base station; (120a,120b,120c,120d) Cluster head; (130) Sensor nodes; (130-1) First sensor node; (130-2) Second sensor node; (130-m,130-n) m-th sensor node
Abstract translation: 目的：提供传感器网络系统的认证密钥分发方法，通过监控网络状态来减少不必要信息的传输。 构成：基站（110）通过从一个或多个簇头（120）收集传感器信息来存储传感器信息。 基站通过对存储的传感器信息应用模糊算法来计算认证密钥分发限制距离。 基站使用计算的认证密钥分发限制距离来确定认证密钥分发的必要性。 当请求认证密钥分发的必要性时，簇头使用接收的认证密钥分发限制距离来发送认证密钥。 （附图标记）（110）基站; （120a，120b，120c，120d）簇头; （130）传感器节点; （130-1）第一传感器节点; （130-2）第二传感器节点; （130-m，130-n）第m个传感器节点
-
-
-
-
-
-
-
-
-
Search Results
14
records
(took 0.02 seconds)
