Activation function

Activation function

In computational networks, the activation function of a node defines the output of that node given an input or set of inputs. A standard computer chip circuit can be seen as a digital network of activation functions that can be "ON" (1) or "OFF" (0), depending on input. This is similar to the behavior of the linear perceptron in neural networks. However, it is the "nonlinear" activation function that allows such networks to compute nontrivial problems using only a small number of nodes.

Functions

In biologically-inspired neural networks, the activation function is usually an abstraction representing the rate of action potential firing in the cell. In its simplest form, this function is binary--that is, either the neuron is firing or not. The function looks like phi(v_i)=U(v_i), where U is the Heaviside step function. In this case a large number of neurons must be used in computation beyond linear separation of categories.

A line of positive slope may also be used to reflect the increase in firing rate that occurs as input current increases. The function would then be of the form phi(v_i)=mu v_i, where mu is the slope. This activation function is linear, and therefore has the same problems as the binary function. In addition, networks constructed using this model have unstable convergence due to the fact that neuron inputs along favored paths tend to increase without bound, as this function is not normalizable.

All problems mentioned above can be handled by using a normalizable sigmoid activation function. One realistic model stays at zero until input current is received, at which point the firing frequency increases quickly at first, but gradually approaches an asymptote at 100% firing rate. Mathematically, this looks like phi(v_i)=U(v_i) anh(v_i), where the hyperbolic tangent function can also be any sigmoid. This behavior is realistically reflected in the neuron, as neurons cannot physically fire faster than a certain rate. This model runs into problems, however, in computational networks as it is not differentiable, a requirement in order to calculate backpropagation.

The final model, then, that is used in multilayer perceptrons is a sigmoidal activation function in the form of a hyperbolic tangent. Two forms of this function are commonly used: phi(v_i)= anh(v_i) whose range is normalized from -1 to 1, and phi(v_i) = (1+exp(-v_i))^{-1} is vertically translated to normalize from 0 to 1. The latter model is often considered more biologically realistic, but it runs into theoretical and experimental difficulties with certain types of computational problems.

Alternative structures

A special class of activation functions known as radial basis functions (RBFs) are used in RBF networks, which are extremely efficient as universal function approximators. These activation functions can take any form, but they are usually found as one of three functions:
*Gaussian: ,phi(v_i)=expleft(-frac{|v_i-c_i|^2}{2sigma^2} ight)
*Multiquadratics: ,phi(v_i)=sqrt{|v_i-c_i|^2 + a^2}
*Inverse multiquadratics: ,phi(v_i)=(|v_i-c_i|^2 + a^2)^{-1/2}where c_i is the vector representing the function "center" and a and sigma are parameters affecting the spread of the radius.

Support vector machines (SVMs) can effectively utilize a class of activation functions that includes both sigmoids and RBFs. In this case, the input is transformed to reflect a decision boundary hyperplane based on a few training inputs called "support vectors" x. The activation function for the hidden layer of these machines is referred to as the "inner product kernel", K(v_i,x)=phi(v_i). The support vectors are represented as the centers in RBFs with the kernel equal to the activation function, but they take a unique form in the perceptron as :,phi(v_i)= anhleft(eta_1 + eta_0 sum_j v_{i,j} x_j ight),where eta_0 and eta_1 must satisfy certain conditions for convergence. These machines can also accept arbitrary-order polynomial activation functions where:,phi(v_i)=left(1+sum_j v_{i,j} x_j ight)^p. [cite book |last=Haykin |first=Simon |authorlink=Simon Haykin |title=Neural Networks: A Comprehensive Foundation |edition=2 |year=1998 |publisher=Prentice Hall |location= |isbn=0132733501 ]

References

External links

* [http://dynamicnotions.blogspot.com/2008/09/sigmoid-function-in-c.html Implementation of the Sigmoid Activation Function in C#]


Wikimedia Foundation. 2010.

Игры ⚽ Поможем сделать НИР

Look at other dictionaries:

  • Softmax activation function — The softmax activation function is a neural transfer function. In neural networks, transfer functions calculate a layer s output from its net input. It is represented as:p i = frac{exp(q i)}{Sigma {j=1}^nexp(q j)}Where p is the value of an output …   Wikipedia

  • Radial basis function network — A radial basis function network is an artificial neural network that uses radial basis functions as activation functions. They are used in function approximation, time series prediction, and control.Network architectureRadial basis function (RBF) …   Wikipedia

  • Sigmoid function — The logistic curve Plot of the error function …   Wikipedia

  • Nested function — definitions can be compared to how a Matryoshka doll nests within larger versions of itself, in several levels. In computer programming, a nested function (or nested procedure/subroutine) is a function which is lexically (textually) encapsulated… …   Wikipedia

  • Macrophage-activation syndrome — (MAS) is a severe, potentially life threatening, complication of several chronic rheumatic diseases of childhood. It occurs most commonly with systemic onset juvenile idiopathic arthritis (SoJIA), which is also known as Still s disease. In… …   Wikipedia

  • Radial basis function — A radial basis function (RBF) is a real valued function whose value depends only on the distance from the origin, so that phi(mathbf{x}) = phi(||mathbf{x}||); or alternatively on the distance from some other point c , called a center , so that… …   Wikipedia

  • Manual fire alarm activation — Single action pull station. A manual call point in the UK …   Wikipedia

  • Androgen receptor — Structure of the ligand binding domain of the androgen receptor (rainbow cartoon) complexed with testosterone (white sticks).[1] …   Wikipedia

  • CREB-binding protein — See also: p300 CBP coactivator family CREB binding protein PDB rendering based on 1f81 …   Wikipedia

  • Estrogen receptor alpha — Estrogen receptor 1 PDB rendering based on 1a52 …   Wikipedia

Share the article and excerpts

Direct link
Do a right-click on the link above
and select “Copy Link”