Search - precision recall

Search - precision recall - List

[Software Engineering] Text-Retrieval DL : 0: 信息检索系统从最初的纯手工检索系统业已发展到现在的以信息技术为支撑的检索系统，在这一过程中，适应新的信息资源、信息技术这些检索环境，提高信息检索系统的查全率、查准率和系统响应时间是不变的主题，在众多文本中掌握最有效的信息始终是信息处理的一大目标。围绕向量空间模型设计了一个文本检索系统，介绍向量空间模型的基础上给出了基于它的信息检索系统的一般结构框架和各部分的功能，探讨了系统中所涉及到的关键技术。用向量空间模型进行特征表达，用TF-IDF(Term-Frequency Inverse-Document-Frequency)进行特征项赋权，用倒排文档进行索引，用余弦夹角进行距离度量，用查全率和查准率评价检索系统性能，并以向量空间模型及相关理论为基础对中文信息检索进行了一些探讨。向量空间模型需要解决特征项的生成和加权、相似度的计算（检索运算）等一系列问题。由于向量检索中采用的向量叫某种距离度量来反映文档的满足程度，所以相似度的值最好能与真实情况相符，计算简便。-Information retrieval system to retrieve from the first hand to the present system has been developed using information technology to support the retrieval system, in the process and adapt to new information resources, information technology, the search environment, improve information retrieval system recall , precision and system response time is the constant theme in many text information is always the most effective control is a major goal of information processing. Vector space model around a text retrieval system is designed to introduce the vector space model is given on the basis of its information retrieval system based on the general framework and functions of each part, of the system, the key technologies involved. The feature vector space model using the expression, with the TF-IDF (Term-Frequency Inverse-Document-Frequency) for feature items empowerment, with the inverted file indexing, with the cosine angle between the distance measurement, with recall and precision evalu
Date : 2025-12-28 Size : 697kb User : Peng Jin
[Software Engineering] labinithom DL : 0: We will use MatCont to start homoclinic orbits that emanate BT points in several models 1 . Recall that in the GUI of MatCont k is denoted as TTolerance. In all examples, we will set TTolerance=10 − 5 (k). As a rule, the amplitude (A) should always be larger than TTolerance given the geometric meaning of both variables, cf. Figure 1. In all cases, ε 0 and ε 1 are the free homoclinic parameters since this appears to be the most stable choice. As another rule, the BT point itself should be computed to a geometric precision signiﬁ cantly smaller than TTolerance. This can be achieved in MatCont by decreasing the tolerances VarTolerance and TestTolerance for the curve on which the BT points are detected.-We will use MatCont to start homoclinic orbits that emanate BT points in several models 1 . Recall that in the GUI of MatCont k is denoted as TTolerance. In all examples, we will set TTolerance=10 − 5 (k). As a rule, the amplitude (A) should always be larger than TTolerance given the geometric meaning of both variables, cf. Figure 1. In all cases, ε 0 and ε 1 are the free homoclinic parameters since this appears to be the most stable choice. As another rule, the BT point itself should be computed to a geometric precision signiﬁ cantly smaller than TTolerance. This can be achieved in MatCont by decreasing the tolerances VarTolerance and TestTolerance for the curve on which the BT points are detected.
Date : 2025-12-28 Size : 706kb User : fouzirock
[Software Engineering] Robust_Face_Landmark DL : 0: 在现实世界条件下获取人脸存在较大的变化在形状和遮挡由于不同在姿态、表情、附属品的使用，例如，太阳镜和帽子以及与目标体（e.g. 食物）的交。当前的人脸界标估计方法在这种条件下努力但由于缺乏一种有效的理论方法用于处理局外点。我们提供了一个新奇的方法，称为Robust Cascaded Pose Regression (RCPR)，通过检测显式的遮挡且使用鲁棒的形状索引的特征可以减少exposure对于局外点。我们证明RCPR改进先前的界标估计方法在3个通用的人脸数据集上(LFPW, LFW and HELEN)。我们进一步探讨RCPR的性能通过引入一个新奇的人脸数据集集中于遮挡，共由1007幅人脸图像组成，表示了大范围的遮挡模式。RCPR减少失败的案例通过在所有4个数据集上的half，同时，它检测人脸遮挡具有一个80/40 的准确率/召回率。-We propose a novel method, called Robust Cascaded Pose Regression (RCPR) which reduces exposure to outliers by detecting occlusions explicitly and using robust shape-indexed features. We show that RCPR improves on previous landmark estimation methods on three popular face datasets (LFPW, LFW and HELEN). We further explore RCPR’s performance by introducing a novel face dataset focused on occlusion, composed of 1,007 faces presenting a wide range of occlusion patterns. RCPR reduces failure cases by half on all four datasets, at the same time as it detects face occlusions with a 80/40 precision/recall.
Date : 2025-12-28 Size : 1008kb User : 郭继东