Starter’s Guide for students at cuer

After https://biaslab.github.io/research/starter-guide, written by Bert de Vries.

This document is a starter’s guide for students who do an internship, semester project, or thesis at the cuer lab. For the duration of your project you will become part of our team of graduate students. I discuss here our research mission, our way of working, the tools we use and provide some links to key scientific papers that relate to our research.

This guide is always a work in progress and each project will have its own goals and dynamics. Still, generally speaking, I would appreciate if you have a look at the links and think about the issues that I raise here. Talk to me when you have questions or ideas to improve this document.

Mission

We are interested in modern AI methods that support automated design of digital signal processing (DSP) systems, in particular in the context of audio processing algorithms.

We aim to develop such a support tool, and our methods are inspired by theories about learning and adaptation in the fields of (Bayesian) machine learning, deep learning, edge intelligence, and models of computational neuroscience. In particular, we derive inspiration from recent work in neuroscience that describes how brains perceive, learn and design their algorithms for speech and object recognition, navigation, etc. This framework is called Active Inference (AIF), and it is the backbone of the Spatial Web.

At cuer, we build efficient tools using AIF, test them in-silica or with real users in physical / virtual worlds, and deploy them on the Spatial Web. Our methods are just as easily applicable to digital twins, XR AVATARS, or other cyber/physical systems.

Methodology

I may ask you to work on problems that I have not solved myself and you will not be able to find a packaged solution on the internet. Also, I may not be able to provide solutions to your questions. Actually, there may not even be a solution. Let’s accept it, you’re doing research now. Find a simpler question that will put you on the path to an answer. Doing research is a continuous process of adaptation in both posing the questions and proposing solutions. In other words, if you can’t find the answer, change the question! Start with the simplest questions and simplest solution proposals and work your way towards the more relevant issues.

Expectations and Deliverables

Your project will likely be embedded as a building block in our research plan.

I expect a professional attitude which includes that you conduct yourself as a team player. You are at the lab, because you are contributing to the research vision of the lab: let the quality of your report and (software) code show this dedication and team play.

I like the idea of working together asynchronously. This means we use tools like Teams, github, Notion to share ideas and code without a strict need to be in the same physical space at the same time.