Bonsai: Reactive Tools for Data Science

I will present Bonsai, a general purpose visual programming language (GPVPL) for rapid prototyping of reactive data acquisition and control systems. We will start with a live demo of how the system works and I will then briefly describe how we leveraged the .NET Reactive Extensions (Rx) combinators as well as runtime code generation to develop flexible, compact and efficient graphical specifications of multi-modal reactive systems for neuroscience and beyond. Finally, I will discuss two recurrent key concerns in GPVPLs that have proven hard to get right in visual programming languages: flexible specification of asynchronous computations; and reactive control structures for discrete event systems (e.g. state machines; behavior trees). I will present how these two problems were addressed in Bonsai by leveraging LINQ and Rx combinators and hopefully we will finish with a brainstorm about future directions (virtual reality, embedded systems, and more).

Speaker Details

Gonçalo is fascinated by intelligent systems and started out his research in 2006 at the Centre for Artificial Intelligence (CENTRIA) to study computational models of cognition. However, he quickly became attracted to the challenges of actually developing interactive real-time systems, so shortly after that he joined the YLabs, the R&D division of YDreams. There he became one of the lead architects behind the YVision software framework and conducted applied research in physics based augmented reality interfaces, computer vision, parallel processing and autonomous agents. In 2010, still frustrated by how little we understand the mechanisms of intelligence, he joined the Champalimaud Neuroscience PhD Programme and his now working with Adam Kampff and Joe Paton to understand how brains build models of the world.

Date:
Speakers:
Goncalo Lopes
Affiliation:
The Sainsbury Wellcome Centre @ UCL
    • Portrait of Jeff Running

      Jeff Running