This page provides the definitions and usage of all building blocks available to Eureqa models. To access further information about building block configuration, click the Documentation link provided in any Eureqa model blueprint (found in the upper right corner of any building block under the Advanced Tuning tab).
Squashing functions take a continuous input variable and map it to a constrained output range.
Recommended use: Depending on the shape of the particular squashing function, that function may be useful in identifying transition points in the data, and/or limiting the total impact of a particular term.
Building Block
Definition
Usage
Complementary Error Function
1.0 - erf(x) where erf(x) is the integral of the normal distribution. Returns a value between 2 and 0.
erfc(x)
Error Function
Integral of the normal distribution. Returns a value between -1 and +1.
erf(x)
Gaussian Function
Returns exp(-x^2). This is a bell-shaped squashing function.
gauss(x)
Hyperbolic Tangent
The hyperbolic tangent of x. Hyperbolic tangent is a common squashing function that returns a value between -1 and +1.
tanh(x)
Logistic Function
Returns 1/(1+exp(-x)). This is a common sigmoid (s-shaped) squashing function that returns a value between 0 and 1.
logistic(x)
Step Function
Returns 1 if x is positive, 0 otherwise.
step(x)
Sign
Returns -1 if x is negative, +1 if x is positive, and 0 if x is zero.