Loading...
Please wait, while we are loading the content...
Similar Documents
Rejection of Incorrect Answers from a Neural Net Classifier (1993)
| Content Provider | CiteSeerX |
|---|---|
| Author | Smieja, Frank |
| Abstract | . The notion of approximator rejection is described, and applied to a neural network. For a real world classification problem the residual error is shown to decrease with the inverse exponential of the fraction of patterns rejected. The trade-off of "good" patterns rejected and "bad" patterns rejected is shown to increase approximately linearly with rejection rate. A compromise is therefore necessary between tradeoff /rejection rate and residual error. A metalevel solution is proposed for removal of the residual error, through use of a modular system of parallel approximators. 1 Introduction The problem of function approximation can be summarized as: y = f(x) (1) where x is a pattern selected from the input space and y is the approximation (the output space). f represents the approximation function used. All function approximators are more or less involved with estimating their own function f . Many have additional properties, but the necessary qualification is equation (1). Neural ... |
| File Format | |
| Publisher Date | 1993-01-01 |
| Access Restriction | Open |
| Subject Keyword | Approximator Rejection Approximation Function Tradeoff Rejection Rate Input Space Parallel Approximators Neural Net Classifier Metalevel Solution Bad Pattern Output Space Rejection Rate Function Approximation Modular System Incorrect Answer Necessary Qualification Residual Error Neural Network Function Approximators Good Pattern Real World Classification Problem Additional Property Inverse Exponential |
| Content Type | Text |
