unsupervised learning造句

• The former belongs to supervised learning and the latter belongs to unsupervised learning
  它们分属于有监督学习与无监督学习。
• The distinct difference between supervised learning and unsupervised learning lies in whether the example consists of the pre - processed output value
  这两种方法最大的区别就在于学习样本是否包含有预先规定好的输出值。
• Decision theory , statistical classification , maximum likelihood and bayesian estimation , non - parametric methods , unsupervised learning and clustering
  决策理论，统计分类，最大似然和贝叶斯估计，非参数方法，非监督的学习与聚类。
• Due to its unsupervised learning ability , clustering has been widely used in numerous applications , such as pattern recognition , image processing , market research and so on
  聚类具有无监督学习能力,被广泛应用于多个领域中,如模式识别、数据分析、图像处理以及市场调研等。
• Both theories are combined to classify the documents by unsupervised learning and discuss the method in which new rules , applied to new unclassified documents , can be formed after classifying the training documents
  本文利用文档聚类和粗糙集约简相结合的方法，对训练文档进行分类，形成规则后对新加入的未分类文档进行归类。
• 3 ) semantic classification model based som network we use the classification model to combines attributes within a database . this is done using an unsupervised learning algorithm . the output is used as training data for the next stage
  3 )基于som网络的语义分类模型设计建立som网络模型，将元数据特征向量进行分类，形成bp网络的目标向量，用于匹配规则的提取。
• The supervised and unsupervised learning diagnosis methods are discussed and several improvements have been presented in the learning algorithms . the simulation results show that the proposed method can perforfti correct diagtioals iii the linear analog circuits with tolerances
  本文对模拟故障诊断的有监督学习和无监督学习方法分别进行了研究，通过对实现过程的分析，对经典的学习算法进行深入研究，并提出若干改进。
• Refer to chinese automatic word segmentation based on statistics , this paper imports the mechanism of open learning , and uses the method of supervised and unsupervised learning . the word segmentation model includes credibility revising and partial tri - gram information
  本文在基于统计的汉语自动分词的基础上，引入开放学习机制，通过有监督和无监督相结合的学习方法，建立包含可信度修正和部分三元语法信息的多元分词模型。
• Based on unsupervised learning , sparse coding is suitable to describe images with non - gaussian distribution and can get rid of the high order redundancy among the image pixels . since the basis function of sparse coding has build - in clustering property , it increases the inter - class variations of the features
  稀疏编码是一种基于非监督学习的算法，它适合描述具有非高斯分布的数据对象，能够有效地消除图像象素点之间的冗余，并具有内在的聚类特性。
• 1 , q 3 , and at last prove the exisitence of ( q , m + n , n , m ) resilient functions when n > q ? 1 . intelligentized ids methods , which can make the system more adaptability and self - studying , are important research directions of ids so far . in order to make the ids systems have better identifying ability and efficiency against new intrusions , we propose the intrusion feature extra - ction algorithm based on ikpca by studying the different kinds of intrusion detection feature extraction algorithm based on unsupervised learning , and then theoretically analysis the conver - gence of the algorithm . in addition , we validate the validity of the algorithm by means of experim - ents ; at the same time , through studying ica and neural networks , we propose fastica - nn ids , and then test the kddcup99 10 % date set to make comparison of kpca 、 ikpca and fastica algorithms in intrusion detection advantages and disadvantages
  为了使入侵检测系统对新的入侵行为有更好的识别能力和识别效率,本文在研究了各种基于无监督学习的入侵检测特征提取方法的基础上,提出了基于增量核主成份分析( ikpca )的入侵检测特征提取方法,并对该方法进行了收敛性分析,同时结合仿真试验对其正确性进行了验证;另外,本文通过研究独立成份分析和神经网络,提出了基于快速独立成份分析和神经网络的入侵检测方法( fastica - nnids ) ,并通过对kddcup99的10 %数据集的检测比较了核主成份分析( kpca ) 、增量核主成份分析( ikpca )和快速独立成份分析( fastica )在入侵检测特征提取方面的优缺点。
• Aiming at the large spatial data sets whose qualities are complex and the situation that non - linearity , continuity and noises exist commonly , the spatial data mining method based on fuzzy neural network is put forward . an improved nearest neighboring clustering algorithm is used to construct the structure of fuzzy neural network , and thus fuzzy rules are extracted from large amounts of data to go on unsupervised learning , and only one dimension parameter needs to be adjusted by bp algorithm . so the method is speeded up , high efficient , accurate precision and has an extensive and promising application
  针对庞大空间数据集性质复杂且非线性、持续性及噪音普遍存在的情况提出了基于模糊神经网络的空间数据挖掘方法，并采用一种改进的最近邻聚类算法用于构建模糊神经网络结构，可从大量的数据中自提取模糊规则进行无导师自学习，采用网络bp算法只调整一维参数，故计算速度较快并更好的保证了精度，经算例分析，证明了该方法快速、高效、精度高，具有广泛的应用前景。
• The main factors of probabilistic neural network including the hidden neuron size , hidden central vector and the smoothing parameter , to influence the pnn classification , are analyzed ; the xor problem is implemented by using pnn . a new supervised learning algorithm for the pnn is developed : the learning vector quantization is employed to group training samples and the genetic algorithms ( ga ’ s ) is used for training the network ’ s smoothing parameters and hidden central vector for determining hidden neurons . simulations results show that , the advantage of our method in the classification accuracy is over other unsupervised learning algorithms for pnn
  本文主要分析了pnn隐层神经元个数,隐中心矢量,平滑参数等要素对网络分类效果的影响,并用pnn实现了异或逻辑问题;提出了一种新的pnn有监督学习算法:用学习矢量量化对各类训练样本进行聚类,对平滑参数和距离各类模式中心最近的聚类点构造区域,并采用遗传算法在构造的区域内训练网络,实验表明:该算法在分类效果上优于其它pnn学习算法
• One method was supervised recognition , which was to take advantage of some known information to determine a given sequence whether contained some specific functional elements ; the other way was unsupervised learning , which was to utilize some measures of comparability and some search algorithm to discovery some potential signals in biosequences
  一种是有指导的识别方法,即利用已知的信息判读一段未知的序列中是否含有某种功能元件;另一种是无指导的学习方法,即利用一些相似性指标,通过搜索算法发现序列中可能蕴含的信号。
• Since the instinct unsupervised learning of the rbf network and blind signal processing are in essence unsupervised learning procedures , therefore the algorithm based on rbf seems rational . 2 ) according to signals high order cumulant . bss problems are transformed into diagnolization of special matrix
  2 )在研究随机过程理论的基础上，从信号的高阶累积量出发，将线性西安理工大学硕士学位论文混合盲分离问题转化为矩阵的对角化问题，大大简化了bss算法的复杂度。
• The other is that when the extending areas of the samples overcross , wrong classification of the samples will occur . as for the first problem a genetic algorithm is used to improve the process of the best parameters " finding . and as for the latter a kind of improved hamming net which uses supervised and unsupervised learning method is employed
  针对模糊hamming网络在应用中存在的参数调整效率低下以及难以保证参数最优的问题，提出了应用遗传算法进行参数调整的改进方法；针对该网络在样本离散范围发生交叠情况下导致归类错误的问题，研究了对于不同模式采用不同的警戒参数的有监督无监督混合学习的改进算法。