Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Connectionist Interfaces

No description
by

Ben Smith

on 13 April 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Connectionist Interfaces

Music Technology presents one the largest domains for truly innovative musical expression and creation today. After building a scaffold for understanding musical meaning, Dr. Smith will show how connectionist cognitive science models can be leveraged as powerful tools for creating new musical interfaces, instruments, and works. Applied machine learning and artificial intelligence enable the structuring of musical and artistic data patterns into relational datasets which are opening new realms of possibility for the digital arts both on and off the stage! Brain Music Abstract :: use models & theories of brain construction to make innovative (inspiring!) art! Meaning in Music & Art? Extrinsic
Meaning Intrinsic Meaning Indexical Referential Semantic Semiotic Does it exist? Highly emotional media. Relationships Patterns Contrast & Interruption, foreshadowing How do we learn new concepts? How do we store knowledge in our mind? "non-indexical" How do you explain a new idea to a friend? "like this" or "like that" Comparisons & Relationships? From Cognitive Science, Pragmatism Connectionist Theory Central tenet: Knowledge is built up as a network of connected ideas and patterns. Concepts are defined by their relationships. "Orange" "Red" Machine Learning neural net::class of students interval loudness I.O.I. P.C. pitch noisiness centroid train class to identify fruit. Features? Color Texture Shape Sweetness Weight Size "fruit" How can we use this? What are the extremes?
What is the loudest, softest, fastest, slowest, etc.? Yay! Now we can identify fruit. So what? Win: precision. Express relationships as distances in cognitive space. Abstracting: works for any feature set! any media with patterns! How do we create in an artistic medium? or express things? understanding our medium: how can I move in that "space"? answers should give clues to features! Features We understand relationships along feature axes rhythm tonality tempo timbre instrumentation/orchestration What features should we use? Q: How do you know two melodies are more distinct or more similar? want to move... ? Extremes A "Best Practice" in Creative Production Understand the expressive limits of your medium. Q: Could ___ be even more ___? If yes then try it! Goal: make the best work we can! [this is hard] phrase length melody contour high & low pitch acceleration dynamics Classroom, Again. Each "grad student" identifies an "extreme" pattern in our creative media "vocabulary." [aka Adaptive Neural Network] Uses: Interface or Instrument Player navigates through the map, we use the activity of the students (nodes) to drive generative systems. Analyzer Input from our medium (live input!) is presented to the class (network) and the grad students record the activity. We now have a way of measuring relationships in the input. Sectional changes, key changes, gradual movement through melodic spaces, similarities in rhythmic patterns, dynamics, etc. !! Imagine: improvising musician. Or with fixed media: way to identify transitions, as a tool to understand what is working in the piece. could this be "scientific" music theory?? Computational Creativity Make a machine that autonomously creates! Is this possible? Define: Novelty New stimuli related to known patterns. Define: Creativity "Reward" from processing novel stimuli. "H" Creativity "P" Creativity Historical Personal Abstract Similarity reprise/recollection How do we recall ideas? What makes it? What form does it take? are major chords happy? x is 25% like y and 50% like z and 2% like w... x y z w 1st: How to calculate these features for the computer? Then the network can interpolate the middle spaces! Example: Interactive Self-Organizing Map Example Wundt Curve Intrinsic Motivation: Example Piano Works... comparing oranges and apples... Application Dr. Benjamin Smith
Lead Technical Engineer, Case Western Reserve ITS
Artist In Residence, Cleveland Institute of Art
Adjunct Assistant Professor, CWRU EECS Show Fluid Sim.?
Full transcript