Predicting The Future

Bad Words, Good Ideas

Digital Semiotics

Fuzzy Thinking

A Brief History of Nonlinearity

Some Time Series

Forecasting The Future

Tools For Inference

Sociology of Science

The Need for State

Learning From Experience

Embedding:

Embedding: spaces

Embedding: forecasting

The Need

Problems With Nonlinear Characterization

Bases for Functional Approximation

Searching For A Minimum

Getting Caught In Local Minima

Momentum

Momentum

Simulated Annealing

Simulated Annealing

Genetic Algorithms

Genetic Algorithms

The Curse and Blessing of Dimensionality

Model Complexity

Regularization

Problems With Nonlinear Inference

The Problem With

Cluster Weighted Modeling

Context

Kernel Density Estimation

Options

CWM E-M

A Posteriori Estimation

Max Likelihood Model Parameters

Example Application

Example: Convergence

Example: Predictions

15-Dimensional Example

Discrete Output States

Example Application:

Music Synthesis

Numerical Bowed String

CWM Model

Data

Resynthesis

The Digital Stradivarius