Schedule of 220B: Modeling Creativity - Winter Quarter, 1999
Instructor : Dan Gang
Teaching Assistant: Charles Nichols
Takes place in: CCRMA, Ballroom
Time: Monday 1:15-3:05 PM.
Dan's office hours: Monday 4:00-5:00,
CCARH Lab (in the trailer)
Tel. (72)3-4971
email address: dang@ccrma.stanford.edu
Charles' office hours: Wednesday 3:00-5:00, by appointment please
Classroom Annex
Tel. (72)3-4971 ext. 354
email address: cnichols@ccrma.stanford.edu
Week 1 - Jan. 11
Introduction and Aims
- Course structure and Introduction
- The aim of the course
- Cognitive musical activities - Listening
- Listening as a sequential stream
- Artificial Intelligence and Music
More details on Week 1
Exercise 1
The foot tapper
Submition date: Jan. 25 (week 3)
Week 2 - Jan. 18
Martin Luther King, Jr. Day/ Civil Rights Day
Week 3 - Jan. 25
Models of Music Cognition
- Cognitive models
- The computational metaphor
- "Human" and "engineering" models
- Artificial Intelligence (AI) methodologies and music(symboloic, sub-symbolic, hybrid)
- Introduction to Artificial Neural Networks (ANN)
- Introduction to Neural Networks and Music
More details of Week 3
Reading assignment 1
Learning Internal Representation by Error Propagation(Rumelhart at al., PDP Vol. 1, 1989)
Submition date: Feb. 8 (week 5)
Exercise 2
Run the Tutorial of Planet.
Submition date: - (no submission)
Week 4 - Feb. 1
Planet Simulation
- Building, running and visualizing a net with Planet simulation.
- The syntax of: net file, pattern file and screen file
- Jordan Sequential net - musical example
- Jordan Sequential net - implementation details with Planet
- Planet interface -the interactive mode and CSH scripts
More details of Week 4
Exercise 3
Neural Network simulation and modeling a foot tapper
Submition date: - Feb. 22(Week 7)
Week 5 - Feb. 8
Modeling Musical Expectations with Neural Networks
- Musical expectations with Jordan sequential net
- Modeling Degree of Realized Expectations(DRE)
- Meter recognition
- Learning mutual influences of meter and harmony
- Biased metric expectations
Projects
Work on proposal for the final project
Submition date: proposal - Feb. 22; complete project and presentation - March 15.
Week 6 - Feb. 15
A Connectionist Approach to Algorithmic Composition - P. Todd
- Designing and building a neural network for learning melodic sequences.
- Representing issues of a melody: pitch, duration, time.
- Training the net with the set of learning examples - monophonic melodies.
- Strategies of generating new monophonic melodies.
- Examinition and evaluation of the results.
Week 7 - Feb. 22
Building and Tuning a Neural Network Model for Real-time Harmonization
- What is a real time harmonization?
- What is tuning of neural network model and why tuning?
- Corpus
- Tuning by using Distance functions and statistics
- (refined) Architecture and representation
- Setting learning parameters (decay parameters)
- Results - musical analysis
- Comparing net's results and regularities with human musicians hamonization
- The effect of meter representation on the net's performance
- The effect of the size of the corpus on the net's performance
- Cognitive implications oif the model
- Implications of the model for Interactive Computer Music
Week 8 - March 1
AI Methodologies and Cognitive Models for Music - Students' presentations
Choose from the list of reading assignment 2
Week 9 - March 8
Hybrid Systems
- Knowledge representation with hybrid systems
- Classification of Hybrid systems
- Distributed AI (the notion of an agent)
- NetNeg
- Summery of the course
Week 10 - March 18***
8:30-11:30.
Be aware ito the change in day and time !!
Project Presentations