|
所在平台: Udemy |
课程主页: https://www.udemy.com/course/anomaly-detection/
课程评论：没有评论
**课程概览：异常检测：机器学习、深度学习、AutoML** 本课程深入探讨异常检测的核心概念、算法和实际应用。异常点被定义为数据集中不符合或与其他数据点不协调的数据点。识别这些异常点至关重要，因为它们能揭示潜在的问题，并为组织改进提供宝贵信息。异常检测是人工智能和机器学习领域中最广泛应用的领域之一，即使在数据有限的情况下也适用。 **课程更新亮点：** * **2024年7月：** 新增关于“混合方法”的视频讲座，探讨如何结合聚类与非聚类算法来识别异常。 * **2023年2月：** 增加“可解释AI”的视频讲座，重点关注理解结果驱动因素的新兴领域。 * **2023年1月：** 引入使用深度学习的异常检测算法，包括自编码器（Auto Encoders）、玻尔兹曼机（Boltzmann Machines）和生成对抗网络（Adversarial Networks）。 * **2022年11月：** 以手动计算的方式，通过少量数据点详细解释了“Isolation Forest”算法，提供了独到的算法讲解视角。 * **2022年7月：** 重点介绍了AutoML，即无需编码即可部署机器学习的新趋势，并新增了“使用PowerBI进行异常检测”的内容。 * **2022年6月：** 增加了关于“不平衡数据集处理”的视频讲座。 * **2022年5月：** 提供了“PyOD：10种算法比较”的视频讲座。 **适用领域：** 异常检测可广泛应用于： * 制造业的预测性维护 * 各行业的欺诈检测 * 各行业的监控活动 * 客户服务与零售行业 * 销售分析 **课程内容涵盖：** * **异常检测的三种类型：** 基于时间（time based）、非基于时间（non time based）以及图像异常。其中，图像异常检测是AI的新前沿。 * **机器学习和深度学习概念。** * **监督和无监督学习算法：** 包括DBSCAN和Isolation Forest等。 * **基于深度学习的图像异常检测技术。** * **异常检测的具体应用场景。** 本课程旨在帮助学员掌握异常检测这一极其有价值的技能，无论您是专业人士还是学生，都能在这个迷人的领域中获得深入的理解和实践能力。
Recent UpdatesJuly 2024: Added a video lecture on hybrid approach (combining clustering and non clustering algorithms to identify anomalies)Feb 2023: Added a video lecture on "Explainable AI". This is an emerging and a fascinating area to understand the drivers of outcomes. Jan 2023: Added anomaly detection algorithms (Auto Encoders, Boltzmann Machines, Adversarial Networks) using deep learningNov 2022: We all want to know what goes on inside a library. We have explained isolation forest algorithm by taking few data points and identifying anomaly point through manual calculation. A unique approach to explain an algorithm!July 2022: AutoML is the new evolution in IT and ML industry. AutoML is about deploying ML without writing any code. Anomaly Detection Using PowerBI has been added. June 2022: A new video lecture on balancing the imbalanced dataset has been added.May 2022: A new video lecture on PyOD: A comparison of 10 algorithms has been addedCourse DescriptionAn anomaly is a data point that doesn't fit or gel with other data points. Detecting this anomaly point or a set of anomaly points in a process area can be highly beneficial as it can point to potential issues affecting the organization. In fact, anomaly detection has been the most widely adopted area with in the artificial intelligence - machine learning space in the world of business. As a practitioner of AI, I always ask my clients to start off with anomaly detection in their AI journey because anomaly detection can be applied even when data availability is limited.Anomaly detection can be applied in the following areas:Predictive maintenance in the manufacturing industryFraud detection across industriesSurveillance activities across industriesCustomer Service and retail industriesSalesThe following will be covered in this program:The three types of anomaly detection - time based, non time based and image. Of these, image anomaly is a new frontier for AI. Just like we analyze the numbers, we can now analyze images and identify anomalies.Machine learning and deep learning conceptsSupervised and unsupervised algorithms (DBSCAN, Isolation Forest)Image anomaly detection using deep learning techniquesScenarios where anomaly detection can be appliedAnomaly detection is one area that can be applied in any type of business and hence organizations embarking on AI journey normally first explore anomaly detection area. So, as professionals and students, you can also explore this wonderful field!
