BAPP is a series on probabilistic modeling and machine learning.
"Probabilistic" here just refers to the use of probability to quantify uncertainty.
What is this about?
You used to have to have a PhD in probability or statistics to work with these methods. Now, there are tools in every major programming language that abstract the math away and let you focus on modeling your problem.
The goal of this group is to come together and chew on questions like;
• How can I, an engineer/scientist/hacker/bizdev, quickly prototype machine learning models specific to my decision problems?
• How do I model the data generating process as well as the training data?.
• How do I build tools that augment human decision making under uncertainty?
What topics have been covered in previous meetings?
• Tutorials in languages and tools such as Stan, PyMC, Edward, Venture, Anglican, Figaro, and others
• Algorithms such as Hamiltonian MCMC and variational inference.
• Domain applications such as image perception, finance, systems and synthetic biology, and cognitive modeling.
• Cutting edge techniques such as causal discovery and causal machine learning, program induction, Bayesian non-parametrics, Bayesian sequential experiments (A/B testing and bandit algorithms, active learning).
• Probabilistic modeling approaches in deep learning and integration with deep learning tools and hardware.
• Building production quality apps using cloud-based infrastructure.
Who should attend? Anyone interested in applying machine learning and Bayesian modeling to their problem domain, such as software engineering, data science, computation biology/chemistry and other natural sciences, cognitive modeling, social science and economics.
Where? Differing locations around the Bay Area, depending on available venues and demand.