/**
* Copyright 2010 Neuroph Project http://neuroph.sourceforge.net
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.neuroph.core.transfer;
import java.io.Serializable;
import org.neuroph.util.Properties;
/**
* <pre>
* Tanh neuron transfer function.
*
* output = ( e^(2*input)-1) / ( e^(2*input)+1 )
* </pre>
* @author Zoran Sevarac <sevarac@gmail.com>
*/
public class Tanh extends TransferFunction implements Serializable {
/**
* The class fingerprint that is set to indicate serialization
* compatibility with a previous version of the class.
*/
private static final long serialVersionUID = 2L;
/**
* The slope parametetar of the Tanh function
*/
private double slope = 2d;
/**
* Creates an instance of Tanh neuron transfer function with default
* slope=1.
*/
public Tanh() {
}
/**
* Creates an instance of Tanh neuron transfer function with specified
* value for slope parametar.
* @param slope the slope parametar for the Tanh function
*/
public Tanh(double slope) {
this.slope = slope;
}
/**
* Creates an instance of Tanh neuron transfer function with the
* specified properties.
* @param properties properties of the Tanh function
*/
public Tanh(Properties properties) {
try {
this.slope = (Double)properties.getProperty("transferFunction.slope");
} catch (NullPointerException e) {
// if properties are not set just leave default values
} catch (NumberFormatException e) {
System.err.println("Invalid transfer function properties! Using default values.");
}
}
@Override
final public double getOutput(double net) {
// conditional logic helps to avoid NaN
if (net > 100) {
return 1.0;
}else if (net < -100) {
return -1.0;
}
double E_x = Math.exp(this.slope * net);
return (E_x - 1d) / (E_x + 1d);
}
@Override
final public double getDerivative(double net) {
double out = getOutput(net);
return (1d - out * out);
}
/**
* Returns the slope parametar of this function
* @return slope parametar of this function
*/
public double getSlope() {
return this.slope;
}
/**
* Sets the slope parametar for this function
* @param slope value for the slope parametar
*/
public void setSlope(double slope) {
this.slope = slope;
}
}