Notes, code snippets and scripts from "Machine Learning with PyTorch and Scikit-Learn" by Sebastian Raschka

• Official remote url by the author
• Book listing

Here are my personal notes, scripts and code snippets I've collected in the course of reading this book. The publication puts a lot of emphasis on the practical side of machine learning while attempting to cover the theory and stat fundamentals behind machine learning as deeply as possible. A great reading for anyone interested in getting started with machine learning engineering. Below I provide an overview of the chapters as per the book's table of contents linking each specific python file.

NOTE: No python notebooks have been used in here seeing them as unnecessary and an annyoing complication in terms of setup. Instead, I consider a python venv + the use of sys args enough to run the scripts and see what they do. If you really can't do without, refer the author's repository.

Setup

$ python -m venv venv && source ./venv/bin/activate
(venv) $ pip install -r requirements.txt

Usage

$ python ./ch<2..=19>/main.py <function_to_execute_from_file>

 $ python ./ch12/main.py torch_datasets

Overview

• Chapter 1: Giving Computers the Ability to Learn from Data

  • Introduction to supervised, unsupervised, and reinforcement learning

• Chapter 2: Training Simple Machine Learning Algorithms for Classification

  • Implementing and training a perceptron on the Iris dataset
  • Adaptive linear neurons and gradient descent

• Chapter 3: A Tour of Machine Learning Classifiers Using Scikit-Learn

  • Deep dive into logistic regression, SVMs, decision trees, and K-nearest neighbors

• Chapter 4: Building Good Training Datasets – Data Preprocessing

  • Handling missing data, categorical data, and feature scaling
  • Techniques for feature selection and importance

• Chapter 5: Compressing Data via Dimensionality Reduction

  • Unsupervised and supervised data compression methods
  • Principal component analysis and linear discriminant analysis

• Chapter 6: Learning Best Practices for Model Evaluation and Hyperparameter Tuning

  • Streamlining workflows with pipelines
  • Using cross-validation for model evaluation
  • Advanced techniques for tuning machine learning models

• Chapter 7: Combining Different Models for Ensemble Learning

  • Techniques for combining classifiers such as bagging, boosting, and ensemble methods

• Chapter 8: Applying Machine Learning to Sentiment Analysis

  • Preparing text data and training models for sentiment analysis

• Chapter 9: Predicting Continuous Target Variables with Regression Analysis

  • Techniques and challenges in linear and polynomial regression
  • Using decision tree and random forest for regression

• Chapter 10: Working with Unlabeled Data – Clustering Analysis

  • K-means clustering, hierarchical clustering, and DBSCAN
  • Evaluating clustering quality

• Chapter 11: Implementing a Multilayer Artificial Neural Network from Scratch

  • Basics of neural network architecture and training
  • Implementing a multilayer perceptron for classifying handwritten digits

• Chapter 12: Parallelizing Neural Network Training with PyTorch

  • Introduction to PyTorch and its functionalities for neural network training

• Chapter 13: Going Deeper – The Mechanics of PyTorch

  • Deep dive into PyTorch's computation graphs and automatic differentiation

  TODO

• Chapter 14: Classifying Images with Deep Convolutional Neural Networks

  • Building and training CNNs for image classification with PyTorch

  TODO

• Chapter 15: Modeling Sequential Data Using Recurrent Neural Networks

  • Implementing RNNs for tasks like sentiment analysis and sequence modeling

  TODO

• Chapter 16: Transformers – Improving Natural Language Processing with Attention Mechanisms

  • Understanding and implementing transformers for NLP tasks

  TODO

• Chapter 17: Generative Adversarial Networks for Synthesizing New Data

  • Fundamentals of GANs and their applications in generating synthetic data

  TODO

• Chapter 18: Graph Neural Networks for Capturing Dependencies in Graph Structured Data

  • Basics and applications of graph neural networks

  TODO

• Chapter 19: Reinforcement Learning for Decision Making in Complex Environments

  • Introduction to reinforcement learning and its methodologies for complex decision-making tasks

  TODO