pyclarion / mockingbird

14
PyClarion / Mockingbird Michael F. Lynch

Upload: wanda

Post on 06-Jan-2016

24 views

Category:

Documents


0 download

DESCRIPTION

PyClarion / Mockingbird. Michael F. Lynch. What is Clarion?. A hybrid cognitive architecture developed at the Department of Cognitive Science at RPI by Ron Sun and colleagues. Four major subsystems, each supporting an upper, symbolic layer and lower, sub-symbolic layer. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: PyClarion  / Mockingbird

PyClarion / Mockingbird

Michael F. Lynch

Page 2: PyClarion  / Mockingbird

What is Clarion?

• A hybrid cognitive architecture developed at the Department of Cognitive Science at RPI by Ron Sun and colleagues.

• Four major subsystems, each supporting an upper, symbolic layer and lower, sub-symbolic layer.

• Models, among other things, motivation (drives and goals) and metacognition

• Also models (in Wilson’s version) personality.

Page 3: PyClarion  / Mockingbird

Action-CenteredExplicit

Representation

Action-CenteredImplicit

Representation

Non-Action-CenteredExplicit

Representation

Non-Action-CenteredImplicit

Representation

Filtering/Selection/Regulation

Goal Setting

Reinforcement

Action

Perception

Goal Structure

Drives

Clarion

ACS

MS

NACS

MCS

Page 4: PyClarion  / Mockingbird

What is PyClarion?

• Another rewrite, necessitated by limitations in the C# edition, this time in Python.

• Uses the PyBrain library by Schmidhuber et al for its neural net implementations.

• This is turn will form the basis for further work into episodic memory.

• Also the basis for Mockingbird, an interactive musical agent (IMA).

Page 5: PyClarion  / Mockingbird

Mockingbird

Page 6: PyClarion  / Mockingbird

What is Mockingbird?

• … an attempt to build an Intelligent musical agent (IMA) based on Ron Sun’s Clarion cognitive architecture and Doug Van Nort’s FILTER expanded musical instrument (aka “hyperinstrument”).

Page 7: PyClarion  / Mockingbird

Mockingbird – What it Does

• Mockingbird interacts with a live performer in a free improvisation performance.

• Mockingbird analyzes, via FILTER, the human performer’s auditory live audio signal, and makes decisions in real time to create its own musical accompaniments.

Page 8: PyClarion  / Mockingbird

Positioning Mockingbird

• “Tape delay” (and later digital delay) music has a long tradition, dating back to the 1950s.

• Builds on the musique concrète tradition of Pierre Schaeffer.

• Among the early pioneers: Pauline Oliveros, right here at RPI.

Page 9: PyClarion  / Mockingbird

Expanded Instruments

• Oliveros’s EIS• Van Nort’s GREIS (inspired by EIS)• Van Nort’s FILTER• CAIRA (Creative, Artificially-Intuitive and

Reasoning Agent)• Mockingbird

Page 10: PyClarion  / Mockingbird

Mockingbird – Design Goals

• Construct a musical “agent” (a musical accompanist and improviser)– Our term: Intelligent Musical Agent (IMA)

• Identify and develop cognitive models of musical performance needed in such an agent.

• Implement these models in an actual agent, using the Clarion cognitive architecture as the “substrate.”

Page 11: PyClarion  / Mockingbird

Mockingbird – Initial Goals

• Be able to listen to a single live performer• “Learn” the contours of the performance

during the actual performance• From that, accompany the performer in the

manner of free improvisation, one-on-one, playing back the performer’s own materials, time-delayed and transformed.

• That’s why it’s called Mockingbird.

Page 12: PyClarion  / Mockingbird

What is Needed…

Page 13: PyClarion  / Mockingbird

• Python programmers… especially if you are really good with multiprocessing programming techniques.

• People interested in neural nets and working with the PyBrain library.

• People interested in cognitive architectures and working on Clarion.

• People interested in these sorts of issues in experimental music.

Page 14: PyClarion  / Mockingbird

[email protected]