Related Experiment Videos
On encoding and enumerating threshold functions.
1Department of Computer Science, Exeter University, Exeter EX4 4QF, U.K. J.Zunic@ex.ac.uk
IEEE Transactions on Neural Networks
|September 24, 2004
Summary
This study introduces efficient methods for encoding and enumerating threshold functions using discrete moments. It provides improved upper bounds for characterizing these functions, optimizing bit rates.
Area of Science:
- Computer Science
- Discrete Mathematics
- Information Theory
Background:
- Threshold functions are fundamental in computational learning theory and pattern recognition.
- Efficient encoding and enumeration are crucial for analyzing complex functions.
- Existing methods for characterizing threshold functions have limitations in terms of optimality and bit rate.
Purpose of the Study:
- To develop novel methods for encoding and enumerating threshold functions on n-dimensional binary inputs.
- To identify conditions under which discrete moments uniquely characterize these functions.
- To derive improved upper bounds for the number of threshold functions.
Main Methods:
- Utilizing a set of discrete moments for function characterization.
- Analyzing the properties of these moments to establish unique representations.
- Estimating the number of possible values for discrete moments.
- Deriving upper bounds based on these estimations.
Main Results:
- Demonstrated that specific sets of discrete moments can uniquely characterize threshold functions.
- Identified conditions for optimal coding in terms of bit rate.
- Derived new upper bounds for several classes of threshold functions.
- Some derived bounds are shown to be superior to previously known results.
Conclusions:
- Discrete moments offer an effective approach for encoding and enumerating threshold functions.
- The proposed methods provide more efficient characterizations and improved bounds.
- This work contributes to a deeper understanding of threshold function complexity and representation.