A central location for all of my lessons and courses on machine learning. Much of this material was developed for folks who are not traditional computer science researchers (often physics or other quantiative science researchers) and my hope is that the content is accessible to a wide audience.
Please feel free to use these materials for your own learning or teaching (with appropriate credit attribution!). I am also very happy to consider guest lectures or similar endeavors, so please don't hesitate to reach out. If you have any suggestions please feel free to open a pull request or contact me at st3565 at columbia.edu
10 hours of material divided into 5 lessons
Description
This course is primarily intended for non-computer science students who want to understand the foundations of building and testing an ML pipeline, different model types, important considerations in data and model design, and the role ML plays in research and society. To get the most out of the course you should have some experience coding (ideally in Python) and remember some general concepts from your calculus class, though you can certainly get a lot out of the material even without these pre-reqs!
Each lesson has a lecture and a corresponding notebook that implements the topics discussed and provides some short exercises (student notebook) and solutions (answer notebook). All coding exercises use python, and most models are built in Keras.
This course was originally designed as a Princeton University Winterssion Intensive and was offered in January 2021 and 2022.
Topics Covered
- Conceptual foundations of ML
- Supervised and unsupervised learning
- Training a model (optimization, data splitting, loss functions, model evaluation)
- Python data science ecosystem
- Neural Networks
- Convolutional Neural Networks
- Recurrent Neural Networks
- Unsupervised Learning
- Clustering
- Dimensionality reduction
- Autoencoders
- Gnerative Models
- GANs
- Gaussian mixture models
- Scientific approaches to model building
- ML in scientific research
- Primer on EI ethics concerns
- Data and model bias
- Algorithmic auditing
- Regulation
Materials
- Recorded lectures
- Slides
- Coding Exercises
A two hour introduction to machine learning. This lecture provides historical and scientific context of the current state of ML, starting from the development of artificial intelligence of a field of research through the deep learning revolution. I then explain how we develop and evaluation different types of ML models and discuss two popular application areas: computer vision and natural language process. Finally, I explore some current open questions in ML and where the field is headed next.
This lecture was part of the NYU ML School 2022.