The best way to master AI technology is by actually building stuff. Dev-related ressources are in another section. Here I focus on the broader field of ML and the main themes that shape it. Like with most of my recommendations, the selection is more targeted towards beginners and not PhD-level researchers.
Regression, classification, tree-based methods, Support vector machines, Deep Learning… do I need to say more? A good introduction of the most important concepts, one version for R and one for Python.
The title holds true: This actually is a hands-on, implementation- and project-focussed publication for frameworks that are still relevant. Includes CNNs, GANs, RNNs and much more.
One of the most and maybe the most holistic publication on Machine Learning Systems from the design phase to deployment and monitoring. Great chapter on the human side of Machine Learning at the end. Still one of the books on ML that are relevant for anybody.
Probability theory, classical ML methods, Deep Learning approaches. Here it all comes together in a cohesive and comprehensive fashion. The website even has the book as a PDF, code examples, and a ton of further ressources. In case you want to go further, try the successor „Probabilistic Machine Learning: Advanced Topics“.
Fundamental open-source Python package. Provides the array data structure and fast numerical operations that underpin for scientific computing.
Builds on top of NumPy to offer labeled, table-like DataFrames optimized for real-world data science tasks like cleaning, filtering, joining, and analyzing tabular or time-series data. Like NumPy, pandas is also open-source.
This is your definitive Python library for classical Machine Learning. Classification, regression, clustering etc. works super smooth. The architecture does not support deep learning and reinforcement learning, keep that in mind. Open-source.
Powerful open-source deep learning framework built on Python architecture with extensive libraries. Defining and training deep neural networks? Here is your go-to choice.
Open source web application that is great for running data science tasks. Jupyter Notebook is maybe the best known module. It allows for individual notebook files consisting of code, equations, visualizations, and text. Super intuitive.
The free course „Practical Deep Learning for Coders“ (consisting of two parts) is just a great, well-done course for hands-on applying deep learning and machine learning in real-world projects. And you get to use PyTorch, which is also cool. Enough said. 😉