Rebecca Fiebrink, Assistant Professor of Computer Science (also Music), : Interactive Machine Learning in Music Performance and Composition
Room:
CS Room 1240
Speaker Name:
Rebecca Fiebrink, Assistant Professor of Computer Science (also Music),
Speaker Institution:
Princeton University In my work at Princeton, I build, study, teach about, and perform with new human-computer interfaces for real-time digital music. Much of my work has concerned the use of supervised learning as a tool for musicians, artists, and composers to build digital musical instruments and other real-time interactive systems. Through the use of training data, supervised learning algorithms offer users a means to specify the relationship between low-level, human-generated control signals (such as the outputs of gesturally-manipulated sensor interfaces, or audio captured by a microphone) and the desired computer response (such as a change in the parameters that dynamically drive computer-generated audio). The task of creating an interactive system can therefore be formulated not as a task of writing and debugging code, but rather one of designing and revising a set of training examples that implicitly encode a target function, and of choosing and tuning an algorithm to learn that function.
In this talk, I will provide a brief introduction to interactive computer music and the use of supervised learning in this field. I will show a live musical demo of the software that I have created to enable non-computer-scientists to interactively apply standard supervised learning algorithms to music and other real-time problem domains. This software, called the Wekinator, supports human interaction throughout the entire supervised learning process, including the generation of training data by real-time demonstration and the evaluation of trained models through hands-on application to real-time inputs.
Drawing on my work with users applying the Wekinator to real-world problems, I'll discuss how data-driven methods can enable more effective approaches to building interactive systems, through supporting rapid prototyping and an embodied approach to design, and through "training" users to become better machine learning practitioners. I'll also discuss some of the remaining challenges at the intersection of machine learning and human-computer interaction that must be addressed for end users to apply machine learning more efficiently and effectively, especially in interactive contexts.
Event Date: