This repo contains a collection of links to free courses and educational material related to machine learning. While there are many excellent paid courses out there, such as platforms like Udacity, Coursera and DataCamp, everything you need to know can be learned completely for free.

The resources are organized into four levels:

• Essential: The main pre-requisit to starting your machine learning journey is being able to use the Python programming language. Without a solid understanding of Python, you'll find it challenging to make significant progress in machine learning.
• Foundation: The Foundation level material covers mathematical and computer science topics that underpin many machine learning algorithms. While it's beneficial to have some background knowledge in these areas, you don't necessarily need to start by diving deep into these foundations*.
• Core: Includes general resources for learning the core concepts of both machine and deep learning.
• Specialization: More advanced and focused resources covering topics such as computer vison, natural language processing (NLP), and machine learning model deployment.

*This doesn't mean that these topics are not important, they are. The level of importance may depend on your role (e.g. applied machine learning engineer vs. researcher), but for sure building your foundational knowledge will help you in the long run.

The approach to learning is entirely up to you. Many learners find a top-down approach effective (assuming you already have some Python skills), as advocated by Jeremy Howard in the fast.ai courses. This method starts with the big picture, applying machine learning models to real data first, and then gradually delving into the details once you understand the concept and why it's important. Over time, you can supplement this practical knowledge by delving deeper into the foundational concepts.

Once you have a good understanding of machine learning and deep learning, you can move onto more specialized courses, such as the deployment of machine learning models.

• Essential: Python
• Foundation: Calculus
• Foundation: Computer Science
• Foundation: General Maths for ML
• Foundation: Linear Algebra
• Foundation: Probability and Statistics
• Core: Machine Learning
• Core: Deep Learning
• Specialization: Computer Vision
• Specialization: Generative Deep Learning
• Specialization: ML Deployment
• Specialization: Natural Language Processing
• Specialization: Reinforcement Learning
• Other: Applied ML/DL
• Other: Misc Advanced Topics

Level key: 🟢 Beginner, 🟡 Intermediate, 🔴 Advanced (These are relative levels but take with a pinch of salt)

Resources legend: 🎥 Videos, 📖 Book, 📙 Jupyter Notebooks, 💻 General code, 📊 Slides, 📝 Notes, 🌐 Website

Python is a critical language for anyone interested in machine learning. Its simplicity and readability make it an excellent choice for beginners, while its powerful libraries like NumPy, Pandas, and Scikit-learn make data manipulation, analysis, and machine learning accessible and straightforward.

Example learning outcomes from studying Python for machine learning:

• Understand the Python programming language, its syntax, and semantics.
• Be proficient in using Python's data structures like lists, tuples, dictionaries, and sets.
• Be able to manipulate and analyze data using Python libraries such as NumPy and Pandas.
• Understand how to visualize data using libraries like Matplotlib and Seaborn.
• Be able to perform feature engineering and feature selection in Python.
• Learn to implement machine learning algorithms using Scikit-learn.

Calculus, particularly matrix calculus, is a fundamental branch of mathematics that plays a vital role in machine learning, especially in the optimization of machine learning models. The article Matrix Calculus for Deep Learning provides a comprehensive overview of what is required for deep learning practitioners. Here are some example learning outcomes for studying calculus in the context of machine learning:

• Understand the concept of a function: Functions are used to represent mathematical relationships between quantities, and are fundamental to machine learning algorithms.
• Master the principles of differential calculus: This includes understanding concepts such as limits, continuity, derivatives, and the rules for differentiation. These are used in machine learning for optimization and understanding how changes in input affect the output.
• Learn about integral calculus: Understanding concepts such as integrals and the fundamental theorem of calculus is important for areas of machine learning such as probabilistic models.
• Understand multivariable calculus: Many machine learning algorithms involve functions of many variables, so understanding partial derivatives and multiple integrals is crucial.
• Learn about matrix calculus: As explained in the article, matrix calculus extends the concepts of calculus to higher dimensions. It is used extensively in deep learning for backpropagation and other optimization algorithms.
• Understand the concept of a gradient: Gradients are used in machine learning to update the parameters of models and minimize error functions.
• Understand optimization: Techniques for finding the maxima and minima of functions are used extensively in machine learning for training models.

The field of machine learning is deeply rooted in computer science, which provides the tools and frameworks that allow us to implement and optimize machine learning algorithms efficiently. It covers a wide range of topics including algorithms and data structures, which are essential for writing efficient code; computer architecture, which can be important for understanding and optimizing computational performance; and software engineering principles, which are crucial for building robust, maintainable, and scalable systems.

The level of computer science knowledge required can vary significantly depending on the role. For instance, roles such as Machine Learning Engineer or MLOps Engineer will require a deeper understanding of computer science principles as they involve designing, implementing, and maintaining machine learning systems. On the other hand, a Data Scientist might focus more on statistical analysis and data interpretation, and thus might not need as extensive a computer science background.

While this repository provides a few links to general material that may be useful in developing code, particularly in Python, there are other resources that provide a more comprehensive study of computer science. One such resource is the Open Source Society University's Computer Science curriculum, which provides a great list of resources to study computer science in detail. Regardless of your role, having a solid foundation in computer science will undoubtedly be beneficial in your machine learning journey.

This section provides general resources covering the mathematical foundations essential for machine learning. These resources include books, courses, and other materials that cover topics such as calculus, linear algebra, statistics, and probability.

Linear Algebra is a fundamental area of mathematics that is important for understanding the underlying structure of machine learning algorithms. Here are some key learning outcomes for studying linear algebra in the context of machine learning:

• Understand the concept of vectors and matrices: These are the basic building blocks of linear algebra and are used extensively in machine learning.
• Master operations with vectors and matrices: This includes addition, subtraction, multiplication, division, and finding the determinant.
• Learn about vector spaces: Understanding the properties of vector spaces and subspaces is crucial for many machine learning algorithms.
• Understand eigenvalues and eigenvectors: These concepts are used in many machine learning algorithms including Principal Component Analysis (PCA).
• Learn about matrix factorizations: Techniques such as Singular Value Decomposition (SVD) and LU decomposition are used in machine learning for solving systems of linear equations, among other things.
• Understand linear transformations: These are functions between vector spaces that preserve the operations of addition and scalar multiplication.
• Learn about orthogonality: The concept of orthogonality is used in machine learning in the context of minimizing error functions.

Probability and statistics are key to understanding the theories behind machine learning algorithms and making informed decisions based on data. Here are some example key learning outcomes for studying probability and statistics in the context of machine learning:

• Understand the basics of probability: This includes understanding concepts such as probability distributions, conditional probability, and Bayes' theorem. These are fundamental to many machine learning algorithms.
• Master descriptive statistics: This includes measures of central tendency (mean, median, mode), measures of dispersion (variance, standard deviation), and understanding distributions (normal, binomial, etc.).
• Learn about inferential statistics: This includes hypothesis testing, confidence intervals, and p-values. These concepts are used to make inferences from data and evaluate machine learning models.
• Understand correlation and covariance: These measures are used to understand the relationship between variables, which is crucial in feature selection in machine learning.
• Learn about regression analysis: Linear regression and logistic regression are fundamental machine learning algorithms.
• Understand the concept of maximum likelihood estimation: This is a method used to estimate the parameters of a probability distribution or statistical model.
• Learn about Bayesian Statistics: Bayesian methods can be utilized in machine learning to update probabilities as more data becomes available.

Machine Learning is a vast field with a variety of techniques and methodologies. Understanding the fundamentals, such as supervised and unsupervised learning, is crucial for anyone looking to work in this field. Here are some of the key learning outcomes for studying the fundamentals of machine learning:

• Understand the difference between supervised and unsupervised learning: Supervised learning involves learning from labeled data, while unsupervised learning involves finding patterns in unlabeled data.
• Learn about different types of supervised learning algorithms: This includes linear regression, logistic regression, decision trees, random forests, support vector machines, and neural networks.
• Understand the concept of training and testing data: Learn how to split your data into training and testing sets to evaluate the performance of your machine learning model.
• Learn about different types of unsupervised learning algorithms: This includes clustering algorithms like K-means, hierarchical clustering, and DBSCAN, and dimensionality reduction techniques like PCA.
• Understand the concept of overfitting and underfitting: Learn how to recognize when a model is too complex (overfitting) or too simple (underfitting) and how to address these issues.
• Learn about evaluation metrics: Understand how to evaluate the performance of your machine learning models using metrics like accuracy, precision, recall, F1 score for classification problems, and mean squared error, mean absolute error for regression problems.
• Understand the basics of feature engineering and selection: Learn how to create new features and select the most relevant ones for your machine learning model.

Deep Learning is a subset of machine learning that focuses on algorithms inspired by the structure and function of the brain called artificial neural networks. Here are the key learning outcomes for studying the fundamentals of deep learning:

• Understand the concept of artificial neural networks: Learn about the structure of neural networks, including concepts such as neurons, layers, weights, biases, and activation functions.
• Learn about different types of neural networks: This includes feedforward neural networks, convolutional neural networks (CNNs), recurrent neural networks (RNNs), Long Short-Term Memory (LSTM) networks and Transformers.
• Understand the process of training neural networks: Learn about forward propagation, cost functions, backpropagation, and gradient descent.
• Learn about optimization algorithms: Understand different optimization techniques used in training neural networks, such as Stochastic Gradient Descent (SGD), RMSprop, Adam, and learning rate scheduling.
• Understand the concept of overfitting and underfitting in the context of deep learning: Learn about techniques to prevent overfitting, such as dropout, early stopping, and regularization.
• Learn about handling sequential data: Understand how RNNs, LSTMs, and GRUs are used to handle time-series data or any data where sequence matters.
• Understand the basics of Convolutional Neural Networks: Learn how CNNs are used for image classification, object detection, and other computer vision tasks.
• Learn about transfer learning: Understand how pre-trained models can be used to improve performance, especially when data is limited

Computer Vision is a field of artificial intelligence that trains computers to interpret and understand the visual world. Here are some key learning outcomes for studying the fundamentals of computer vision:

• Understand the use of convolutional neural networks in computer vision: Learn how CNNs are used for tasks like image classification, object detection, and semantic segmentation.
• Understand the principles of object detection and recognition: Learn about techniques such as R-CNN, Fast R-CNN, and YOLO.
• Learn about segmentation: Understand techniques for partitioning an image into multiple segments, including thresholding, clustering methods (like k-means), and more advanced techniques like Mask R-CNN.
• Learn about transfer learning in the context of computer vision: Understand how pre-trained models on large datasets can be used to improve performance on computer vision tasks.

Generative Deep Learning is a subset of machine learning that focuses on creating new data instances that resemble your training data. This is achieved by training models to generate data that can pass as "real" data, thereby learning the true data distribution of the training set. Generative models have a wide range of applications, including image synthesis, semantic image editing, style transfer, data augmentation and much more.

Key learning objectives for studying Generative Deep Learning include:

• Understanding the basics of generative models: This includes learning about the difference between generative and discriminative models and the use cases for generative models.
• Learning about Generative Adversarial Networks (GANs): This involves understanding the architecture of GANs, how they work, and how to train them.
• Grasping the principles of Variational Autoencoders (VAEs): This includes learning about the architecture of VAEs, how they work, and how to train them.
• Learning about other generative models: This involves understanding other types of generative models such as autoregressive models and normalizing flows.
• Understanding the use of generative models for data augmentation: This includes learning how to use generative models to augment your training data and improve the performance of your models.
• Learning about the ethical considerations of generative models: This involves understanding the ethical implications of creating realistic synthetic data and how to use generative models responsibly.

The deployment of machine learning models, often referred to as ML Ops or productionizing machine learning, is the process of transforming a machine learning model into an application that can be used in a practical business setting. It involves integrating the model into an existing production environment where it can receive input data, process it, and return output.

Key learning objectives for studying Machine Learning Deployment include:

• Understanding the basics of machine learning deployment: This includes learning about the lifecycle of a machine learning project, from data collection and model training to deployment and monitoring.
• Learning about model versioning: This involves understanding how to manage different versions of a machine learning model and how to handle updates to the model.
• Grasping the principles of model serving: This includes learning about how to set up a server to host the model and handle requests in real-time.
• Learning about containerization: This involves understanding how to use technologies like Docker to package a model and its dependencies into a single, standalone unit that can be run anywhere.
• Understanding the basics of scalability: This includes learning about how to ensure that your model can handle increasing amounts of requests as usage grows.
• Learning about model monitoring: This involves understanding how to track the performance of your model over time and how to detect and handle any degradation in performance.
• Understanding the use of CI/CD in machine learning: This includes learning about how to use continuous integration and continuous deployment to automate the deployment process and ensure that your model is always up to date.
• Learning about data and feature stores: This involves understanding how to manage and store the data that your model needs to make predictions.

Natural Language Processing (NLP) focuses on enabling computers to understand, interpret, and generate human language. This involves teaching machines to understand linguistic structures, meanings, and nuances to effectively process text data in a human-like manner. NLP combines computational linguistics and machine learning to create systems that can perform tasks such as translation, sentiment analysis, and topic extraction.

Key learning objectives for studying NLP include:

• Understanding the basics of text processing: This includes techniques such as tokenization, stemming, lemmatization, and part-of-speech tagging.
• Learning about language modeling: This involves understanding concepts such as n-grams and the use of probabilistic models for predicting the next word in a sentence.
• Grasping the principles of information extraction: This includes learning about named entity recognition, coreference resolution, and relation extraction.
• Learning about text classification: This involves understanding how to use techniques like Naive Bayes and logistic regression for tasks such as spam detection or sentiment analysis.
• Understanding the basics of syntactic analysis: This includes learning about parsing techniques to understand the grammatical structure of sentences.
• Learning about semantic analysis: This involves understanding how to extract meaning from text, including word sense disambiguation and semantic role labeling.
• Understanding the use of embeddings for representing text: This includes learning about techniques like Word2Vec and GloVe for creating dense vector representations of words.
• Learning about sequence-to-sequence models: This involves understanding how to use models like RNNs, LSTMs, and Transformers for tasks like machine translation and text summarization.

Reinforcement Learning (RL) is a type of machine learning where an agent learns to make decisions by taking actions in an environment to achieve a goal. The agent learns from the consequences of its actions, rather than from being explicitly taught, and adjusts its behavior based on the rewards and penalties it receives.

Key learning objectives for studying RL include:

• Understanding the basics of reinforcement learning: This includes concepts such as states, actions, rewards, and the reinforcement learning loop.
• Learning about Markov Decision Processes (MDPs): This involves understanding how MDPs provide a mathematical framework for modeling decision-making in situations where outcomes are partly random and partly under the control of a decision-maker.
• Grasping the principles of dynamic programming: This includes learning about techniques such as policy iteration and value iteration.
• Learning about model-free methods: This involves understanding how to use techniques like Monte Carlo methods and Temporal-Difference learning for estimating value functions without a model of the environment.
• Understanding the basics of function approximation: This includes learning about techniques for dealing with large or continuous state spaces, including linear function approximation and deep learning methods.
• Learning about policy gradient methods: This involves understanding how to optimize policies directly based on the gradient ascent on the expected return.
• Understanding the use of exploration and exploitation: This includes learning about the trade-off between exploiting what you already know to gain reward and exploring the environment to find better actions.

This section includes courses where machine/deep learning is applied to specific fields, such as healthcare and self-driving cars.

This section includes other useful courses covering advanced topics in machine learning.