Machine Learning Simplified, A Gentle Introduction to Supervised Learning, Wolf A., 2022

Machine Learning Simplified, A Gentle Introduction to Supervised Learning, Wolf A., 2022.

   My goal for this book is to help make machine learning available to as many people as possible whether technical or not. It is easily accessible for a non-technical reader, but also contains way enough mathematical detail to serve as an introduction to machine learning for a technical reader. Nevertheless, some prior knowledge of mathematics, statistics and the Python programming language is recommended to get the most out of this book.




Visualizing Decision Regions.
It is often difficult for us humans to understand what predictions the prediction function f will make from the feature values. To understand the predictions better, it is often useful to make a visualization. For many large-scale ML problems, with thousands or millions of features, visualization is a very difficult task. However, when the input features have only two dimensions, we can simply show how each point in the plane would be classified by f. This visualization is shown in Figure 2.2a for our fruit classification example. Each location in the plane, representing a height and width measurement, is given a color equal to the color of its nearest data point in our training set (dot in figure). Each region of the same color is called a decision region for each class (e.g., the green is the decision region for mandarin), and a curve that makes up the boundary between decision regions is called a decision boundary (e.g., the boundary between green and red).

With this visualization in hand, we can easily see what fruit type our learned classifier would predict given height and width measurements. For example, how would we classify our unlabeled test fruit that is 4cm tall and 5cm wide? We look at the decision regions and see that this point falls in the green regime, corresponding to the mandarin fruit type, and we predict mandarin. This is shown in Figure 2.2b.

Contents.
I FUNDAMENTALS OF SUPERVISED LEARNING.
1 Introduction.
1.1 Machine Learning.
1.1.1 Supervised Learning.
1.1.2 Unsupervised Learning.
1.2 Machine Learning Pipeline.
1.2.1 Data Science.
1.2.2 ML Operations.
1.3 Artificial Intelligence.
1.3.1 Information Processing.
1.3.2 Types of AI.
1.4 Overview of this Book.
2 Overview of Supervised Learning.
2.1 ML Pipeline: Example.
2.1.1 Problem Representation.
2.1.2 Learning a Prediction Function.
2.1.3 How Good is our Prediction Function?.
2.1.4 Controlling Model Complexity.
2.2 ML Pipeline: General Form.
2.2.1 Data Extraction.
2.2.2 Data Preparation.
2.2.3 Model Building.
2.2.4 Model Deployment.
3 Model Learning.
3.1 Linear Regression.
3.1.1 Linear Models.
3.1.2 Goodness-of-Fit.
3.1.3 Gradient Descent Algorithm.
3.1.4 Gradient Descent with More Parameters.
3.2 Gradient Descent in Other ML Models.
3.2.1 Getting Stuck in a Local Minimum.
3.2.2 Overshooting Global Minimum.
3.2.3 Non-differentiable Cost Functions.
4 Basis Expansion and Regularization.
4.1 Basis Expansion.
4.1.1 Polynomial Basis Expansion.
4.1.2 Comparison of Model Weights.
4.2 Regularization.
4.2.1 Ridge Regression.
4.2.2 Choosing Regularization Strength l.
4.2.3 Lasso Regression.
4.2.4 Comparison between L1 and L2 Regularization.
5 Model Selection.
5.1 Bias-Variance Decomposition.
5.1.1 Mathematical Definition.
5.1.2 Diagnosing Bias and Variance Error Sources.
5.2 Validation Methods.
5.2.1 Hold-out Validation.
5.2.2 Cross Validation.
5.3 Unrepresentative Data.
6 Feature Selection.
6.1 Introduction.
6.2 Filter Methods.
6.2.1 Univariate Selection.
6.2.2 Multivariate Selection.
6.3 Search Methods.
6.4 Embedded Methods.
6.5 Comparison.
7 Data Preparation.
7.1 Data Cleaning.
7.1.1 Dirty Data.
7.1.2 Outliers.
7.2 Feature Transformation.
7.2.1 Feature Encoding.
7.2.2 Feature Scaling.
7.3 Feature Engineering.
7.3.1 Feature Binning.
7.3.2 Ratio Features.
7.4 Handling Class Label Imbalance.
7.4.1 Oversampling.
7.4.2 Synthetic Minority Oversampling Technique (SMOTE).
A Appendix Unsupervised Learning.
B Appendix Non-differentiable Cost Functions.
B.0.1 Discontinuous Functions.
B.0.2 Continuous Non-differentiable Functions.



Бесплатно скачать электронную книгу в удобном формате, смотреть и читать:
Скачать книгу Machine Learning Simplified, A Gentle Introduction to Supervised Learning, Wolf A., 2022 - fileskachat.com, быстрое и бесплатное скачивание.

Скачать pdf
Ниже можно купить эту книгу, если она есть в продаже, и похожие книги по лучшей цене со скидкой с доставкой по всей России.Купить книги


Скачать - pdf - Яндекс.Диск.




Теги: учебник по программированию :: программирование :: Wolf