|
所在平台: Udemy |
课程主页: https://www.udemy.com/course/unsupervised-machine-learning-clustering-algorithms/
课程评论：没有评论
课程名称：无监督机器学习：聚类分析算法 概述：聚类是将数据划分为不同部分的活动，旨在揭示未标记数据的内在信息。通过对相似数据点的分组，聚类为数据提供了结构化的视角。这些聚类算法几乎在我们日常生活的方方面面都能看到。聚类分析是所有机器学习的重要组成部分，广泛应用于标签生成、标签验证、降维、半监督学习、强化学习、计算机视觉和自然语言处理等多个机器学习任务。对于数据科学家而言，聚类分析是他们在探索性分析中最初使用的工具之一，用于识别数据中的自然分区。 在本课程中，您将学习一些用于聚类分析的重要算法。每个数据集和特征空间都是独特的，因此无法采用一种放之四海而皆准的方法来识别数据中的模式。每种算法都有其特定的用途。通过深入学习核心概念，详细编写每个算法的代码，您将能够根据具体场景识别出正确的算法。一些算法速度较快，适合快速识别数据模式，而另一些算法虽然较慢，但能够更精确地捕捉模式。 您将深入理解每种算法，从而提升调优其参数和最大化其效用的直觉。课程将涵盖五种聚类算法：KMeans和Meanshift是基于中心的算法，聚类的定义围绕聚类中心进行；DBSCAN和OPTICS是基于密度的算法，旨在寻找数据中的高密度区域，将这些连续的密度区域识别为聚类；Agglomerative Clustering是一种层次聚类算法，提供聚类的层级结构。 对于每种算法，您将理解其核心工作机制、所需参数、选择和调优这些参数的方法，以及如何评估每个算法的结果。为了巩固您的理解，您还将在多个数据集上应用这些学习成果，比较所有算法及其性能。 本课程可以成为您学习各种聚类算法的唯一参考。建议您首次学习时按照给定顺序进行，跟随介绍性讲座。在编码课程中，建议您边听边动手编码。我提供了详细的Jupyter笔记本供您参考，但强烈建议您跟随授课进行编码，并在学习过程中思考工作流程。课程结束后，您将能以此课程作为您的聚类算法查阅资料，并可终身访问，再次回顾这些算法。
Clustering is the activity of splitting the data into partitions that give an insight about the unlabelled data. It gives a structure to the data by grouping similar data pointsWe see these clustering algorithms almost everywhere in our everyday life. Cluster Analysis has and always will be a staple for all Machine Learning. Clustering has its applications in many Machine Learning tasks: label generation, label validation, dimensionality reduction, semi supervised learning, Reinforcement learning, computer vision, natural language processing.For a data scientist, cluster analysis is one of the first tools in their arsenal during exploratory analysis, that they use to identify natural partitions in the data.In this course, you will learn some of the most important algorithms used for Cluster AnalysisEach dataset and feature space is unique. You cannot use a one-size-fits-all method for recognizing patterns in the data. Each algorithm has its own purpose.By studying the core concepts and working in detail and writing the code for each algorithm from scratch, will empower you, to identify the correct algorithm to use for each scenario.Some algorithms are fast and are a good starting point to quickly identify the pattern of the dataAnd some algorithms are slow but more precise, and allow you to capture the pattern very accurately.You will get to understand each algorithm in detail, which will give you the intuition for tuning their parameters and maximizing their utilityIn this course, for cluster analysis you will learn five clustering algorithms:You will learn about KMeans and Meanshift. These are two centroid based algorithms, which means their definition of a cluster is based around the center of the cluster.Next you will study DBSCAN and OPTICS. These are density based algorithms, in which they find high density zones in the data and for such continuous density zones, they identify them as clusters.Another type of algorithm that you will learn is Agglomerative Clustering, a hierarchical style of clustering algorithm, which gives us a hierarchy of clusters.For each algorithm, you will understand the core working of the algorithm. What parameters they use. How to choose and tune these parameters. How to evaluate the results for each algorithm. To consolidate your understanding, you will also apply all these learnings on multiple datasets for each algorithm. You can later compare all the algorithms and their performance.This course can be your only reference that you need, for learning about various clustering algorithms.The correct approach to this course is going in the given order the first time.Follow along the introductory lecture. It is highly recommended that during the coding lessons, you must code along. I have provided detailed jupyter notebooks along the course. You can keep them for reference. But it is highly recommended that you code along.You can pause the lesson. Write the code needed and at the same time think about the working flow.I assure you, there onwards, this course can be your go-to reference to answer all questions about these algorithms. You will have a lifetime of access to this course, and thus you can keep coming back to quickly brush up on these algorithms
