dk.bayes.math

gaussian

package gaussian

Visibility
  1. Public
  2. All

Type Members

  1. case class Gaussian(m: Double, v: Double) extends NumericOps[Gaussian] with Product with Serializable

    Univariate Gaussian Distribution.

    Univariate Gaussian Distribution.

    m

    Mean

    v

    Variance

  2. trait GaussianNumericOps extends AnyRef

  3. case class LinearGaussian(a: Double, b: Double, v: Double) extends Product with Serializable

    Linear Gaussian Model: N(m,v), where, m = ax + b

    Linear Gaussian Model: N(m,v), where, m = ax + b

    a

    Mean component: m = ax + b

    b

    Mean component: m = ax + b

    v

    Variance

  4. case class MoG(z: Array[Double], x: Array[Gaussian]) extends Product with Serializable

    Mixture of Gaussians.

    Mixture of Gaussians.

    Math details: http://en.wikipedia.org/wiki/Mixture_model http://en.wikipedia.org/wiki/Normal_distribution#Moments http://stats.stackexchange.com/questions/16608/what-is-the-variance-of-the-weighted-mixture-of-two-gaussians

    z

    Mixture component of probabilities (prior)

    x

    Conditional probabilities p(x|k) for all components of z (likelihood)

  5. case class MultivariateGaussian(m: Matrix, v: Matrix) extends Product with Serializable

    Multivariate Gaussian from the book 'Christopher M.

    Multivariate Gaussian from the book 'Christopher M. Bishop. Pattern Recognition and Machine Learning (Information Science and Statistics), 2009'

Value Members

  1. object Gaussian extends GaussianNumericOps with Serializable

  2. object KalmanFilter

    Kalman Formulas following chapter 15 from the book 'Stuart Russell, Peter Norvig.

    Kalman Formulas following chapter 15 from the book 'Stuart Russell, Peter Norvig. Artificial Intelligence - A Modern Approach, Third Edition, 2010'

  3. object MultivariateGaussian extends Serializable

  4. object Proj

    Projects f(x)*q(x) distribution to a gaussian distribution q_new(x) by matching mean and variance moments.

    Projects f(x)*q(x) distribution to a gaussian distribution q_new(x) by matching mean and variance moments.

    Thomas P Minka. A family of algorithms for approximate Bayesian inference, 2001

  5. package canonical

Ungrouped