Azure 机器学习设计器机器学习算法备忘单Machine Learning Algorithm Cheat Sheet for Azure Machine Learning designer

“Azure 机器学习算法备忘单”可帮助你从设计器为预测分析模型选择正确的算法。The Azure Machine Learning Algorithm Cheat Sheet helps you choose the right algorithm from the designer for a predictive analytics model.

Azure 机器学习拥有一个大型算法库,包括分类推荐系统聚类分析异常检测回归文本分析系列。Azure Machine Learning has a large library of algorithms from the classification, recommender systems, clustering, anomaly detection, regression and text analytics families. 每一类算法都可用于解决一种类型的机器学习问题。Each is designed to address a different type of machine learning problem.

有关其他指南,请参阅如何选择算法For additional guidance, see How to select algorithms

下载:机器学习算法备忘单Download: Machine Learning Algorithm Cheat Sheet

在此下载备忘单:机器学习算法备忘单(11 x 17 英寸)Download the cheat sheet here: Machine Learning Algorithm Cheat Sheet (11x17 in.)

机器学习算法备忘单:了解如何选择机器学习算法。

下载该机器学习算法备忘单,并将其打印为 Tabloid 大小,既方便携带又可帮助你选择算法。Download and print the Machine Learning Algorithm Cheat Sheet in tabloid size to keep it handy and get help choosing an algorithm.

如何使用机器学习算法备忘单How to use the Machine Learning Algorithm Cheat Sheet

此算法备忘单中提供的建议近似于经验法则。The suggestions offered in this algorithm cheat sheet are approximate rules-of-thumb. 一些可以不完全照做,一些可以大胆地违反。Some can be bent, and some can be flagrantly violated. 此备忘单旨在提供一个起点。This cheat sheet is intended to suggest a starting point. 不要担心几种算法之间对数据运行正面竞争。Don’t be afraid to run a head-to-head competition between several algorithms on your data. 每种算法的原理和生成数据的系统都需要了解,此外别无选择。There is simply no substitute for understanding the principles of each algorithm and the system that generated your data.

每种机器学习算法都有自己的风格或归纳偏差。Every machine learning algorithm has its own style or inductive bias. 对于特定问题,有多种算法可能都合适,但会有一种算法可能比其他算法更合适。For a specific problem, several algorithms may be appropriate, and one algorithm may be a better fit than others. 但并非总是可以预先知道哪种是最合适的。But it's not always possible to know beforehand which is the best fit. 在这些情况下,会在备忘单中列出几种算法。In cases like these, several algorithms are listed together in the cheat sheet. 适当的策略是尝试一种算法,如果结果尚不令人满意,则尝试其他算法。An appropriate strategy would be to try one algorithm, and if the results are not yet satisfactory, try the others.

若要详细了解 Azure 机器学习设计器中的算法,请参阅算法和模块参考To learn more about the algorithms in Azure Machine Learning designer, go to the Algorithm and module reference.

机器学习的种类Kinds of machine learning

有三种主要类别的机器学习:监督式学习非监督式学习强化学习There are three main categories of machine learning: supervised learning, unsupervised learning, and reinforcement learning.

监督式学习Supervised learning

在监督式学习中,将标记每个数据点或将其与某个类别或相关值相关联。In supervised learning, each data point is labeled or associated with a category or value of interest. 分类标签的示例是将图像分配为“猫”或“狗”。An example of a categorical label is assigning an image as either a ‘cat’ or a ‘dog’. 值标签的示例是与二手车关联的销售价格。An example of a value label is the sale price associated with a used car. 监督式学习的目的是研究大量类似这样的标记示例,并能够对未来的数据点进行预测。The goal of supervised learning is to study many labeled examples like these, and then to be able to make predictions about future data points. 例如,识别包含正确动物的新照片或为其他二手车指定准确的销售价格。For example, identifying new photos with the correct animal or assigning accurate sale prices to other used cars. 这是一种常用且有用的机器学习类型。This is a popular and useful type of machine learning.

非监督式学习Unsupervised learning

在非监督式学习中,数据点没有与其关联的标签。In unsupervised learning, data points have no labels associated with them. 相反,非监督式学习算法的目的是以某种方式组织数据或者说明其结构。Instead, the goal of an unsupervised learning algorithm is to organize the data in some way or to describe its structure. 像 K-means 一样,非监督式学习将数据分组到群集中,或者找到不同的方法来查看复杂数据,使其看起来更简单。Unsupervised learning groups data into clusters, as K-means does, or finds different ways of looking at complex data so that it appears simpler.

强化学习Reinforcement learning

在强化学习中,算法需选择响应每个数据点的操作。In reinforcement learning, the algorithm gets to choose an action in response to each data point. 它是机器人学中的常见方法,在此技术中一个时间点的传感器读数集是数据点,并且算法必须选择机器人的下一个动作。It is a common approach in robotics, where the set of sensor readings at one point in time is a data point, and the algorithm must choose the robot’s next action. 它也是物联网应用程序的理想选择。It's also a natural fit for Internet of Things applications. 学习算法还会在短时间后收到奖励信号,指示决策的优秀程度。The learning algorithm also receives a reward signal a short time later, indicating how good the decision was. 基于该信号,该算法会修改其策略以获得最高奖励。Based on this signal, the algorithm modifies its strategy in order to achieve the highest reward.

后续步骤Next steps