|
所在平台: Udemy |
课程主页: https://www.udemy.com/course/demystifying-machine-learning/
课程评论：没有评论
课程名称：掌握机器学习和深度学习与Python 课程概述： 本课程将向您揭示掌握机器学习的秘密。 1. **秘密一**：机器学习的总体秘密是知道不该学习什么。在面对大量机器学习信息时，重要的是要专注于核心概念，避免分心。 2. **秘密二**：对数学和统计的要求相对较低。许多人认为要精通机器学习就必须掌握大量的数学和统计知识，但事实上，实际应用中对这些知识的需求很有限。可以把这种理解比作数据库索引的知识：您需要掌握如何使用数据库索引的最佳实践，而不必了解索引算法的工作原理，机器学习的概念也是如此。 3. **秘密三**：精细调整是掌握机器学习的关键技能。任何有经验的机器学习专家都会告诉你，将机器学习问题投入生产所需的最大时间都花在了优化上。因此，理解过拟合、欠拟合、灵敏度、特异性、精度、ROC和AUC等术语非常重要。课程会花大量时间讲解这些关键基础概念。此外，像Google和亚马逊等公司正在推出AutoML工具，几乎不需要编码，但仍然需要掌握基本概念。未来的数据科学世界将更加注重关键概念，而非编码。 这一千里之行始于第一步。您一直想学习机器学习，但许多因素阻止了您——对数学和统计的恐惧、学科的复杂性。今天是打破这些恐惧的一天。注册本机器学习课程，亲自体验掌握机器学习的简化过程。 课程覆盖以下主题，并使用Google Python Notebooks，您可以立即看到代码结果： - 机器学习基础：成本函数、标记和未标记数据、特征权重、训练与测试、交叉验证 - 特征工程：归一化、标准化 - 线性回归 - 分类：关于真正正例、真正负例、灵敏度、特异性、精度、ROC、AUC和混淆矩阵的概念 - KNN算法 - 过拟合与欠拟合 - 正则化 - 决策树：熵、信息增益 - 集成学习：Bagging和Boosting - 无监督学习：K-Means - 深度学习：权重、偏置、轮次、梯度下降、批量、随机梯度下降、小批量 - 附录课程：Numpy和Pandas 在参加课程之前，请注意以下要点： - 需要具备良好的Python、Numpy和Pandas知识。请确保掌握这些内容后再继续课程。 - 您需要有耐心。如果您对机器学习完全陌生，请准备花费2到4个月的时间来消化这些概念。
Let me begin by telling secrets of mastery of machine learning. # Secret 1 - The overall secret is machine learning is to know what not to learn. Given the amount of information in machine learning it is important to focus on important concepts and not get distracted.#Secret 2 - The requirement of maths and statistics is very shallow. In general people think that to master machine learning one needs to know lot of maths and statistics. That is not true. When it comes to applying machine learning, the knowledge of maths and statistics is limited. The way to think about this to compare with knowledge of database indexes. You need to master the best practices of using database indexes. You don't need to know how databases indexes algorithms work. The same holds for machine learning concepts.#Secret 3 - The key skill to master machine learning is fine tuning. Any experienced ML expert will tell you that the maximum time that goes in taking machine learning problems to production is optimisation. Hence ,is important to understand terms like overfitting ,underfitting sensitivity, specificity, precision, ROC, AUC. The course spends lot of time on these key fundamental concepts. Also the likes of Google and Amazon are producing tools like AutoML where the requirement of coding is close to zero. But what is still required are the fundamental concepts. The world of tomorrow of data science is less of coding but more key concepts.A journey of thousand miles begins with first step. You always wanted to learn machine learning but many factors stopped you - fear of Maths , Statistics , the complexity of subject. Today is the day to break away from those fears. Enrol in the machine learning course and see for yourself that mastering machine learning can be simplified. Following are topics the course covers. The course uses Google Python notebooks. You see the code results immediately.Fundamentals of machine learning - Cost Functions, Labelled and Unlabelled data, Feature weights, Training and Testing Cross Validation.Feature Engineering - Normalization, StandardizationLinear RegressionClassification - Concepts about True Positive, True Negative, Sensitivity, Specificity, Precision, ROC, AUC, Confusion MatrixKNN - AlgorithmOverFitting and UnderFittingRegularizationDecision Trees - Entropy, Information GainBagging and BoostingUnsupervised Learning - K-MeansDeep Learning - Weights, Bias, Epochs, Gradient Descent,Batch, Stochastic Gradient Descent , Mini BatchAppendix course on Numpy and Pandas have also been added.Following are essential points before taking the courseA good knowledge of Python, Numpy and Pandas is required. Please don't proceed with the course unless you master it.You need to be patient. Please be prepared to spend two to four months to digest these concepts if you are completely new to machine learning.
