Other Tutorials and Resources
Natural Gradients
What is a natural gradient was a question that came up during the mathematics of machine learning tutorial. Here is a blog post, and the original paper by Amari, for an introduction to the topic.
What is a natural gradient was a question that came up during the mathematics of machine learning tutorial. Here is a blog post, and the original paper by Amari, for an introduction to the topic.
Gaussian Processes
What is a Gaussian process and its connections to deep networks came up during the tutorials on Probabilistic Reasoning and Machine Learning for Healthcare. Here is a tutorial on the topic, and tensorflow code.
What is a Gaussian process and its connections to deep networks came up during the tutorials on Probabilistic Reasoning and Machine Learning for Healthcare. Here is a tutorial on the topic, and tensorflow code.
- Gaussian processes tutorial
- GPflow is a package for building Gaussian process models in python, using TensorFlow
Bayesian Optimisation
Ways of automating the search over hyper-parameters came up in the practicals, and other ways of doing policy learning, came up. One way is to use Bayesian optimisation; this is built on knowledge of Gaussian processes. Here are some resources to get started:
Ways of automating the search over hyper-parameters came up in the practicals, and other ways of doing policy learning, came up. One way is to use Bayesian optimisation; this is built on knowledge of Gaussian processes. Here are some resources to get started:
Instrumental Variables and Causality
Nando encouraged us all to have an understanding of instrumental variables and basic causal reasoning. This topic also came up during the tutorials on RL and healthcare. The wikipedia page is a good resource to give an understanding of this as a starting point.
Nando encouraged us all to have an understanding of instrumental variables and basic causal reasoning. This topic also came up during the tutorials on RL and healthcare. The wikipedia page is a good resource to give an understanding of this as a starting point.