sgd

Documentation

Index

• type Config
  • func NewConfig(lr, momentum float64, nesterov bool) Config
• type SGD
  • func New(c Config) *SGD
  • func (o *SGD) Delta(param nn.Param) mat.Matrix
  • func (o *SGD) Label() int
  • func (o *SGD) NewSupport(r, c int) *nn.Payload

Types

type Config

type Config struct {
	gd.MethodConfig
	LR       float64
	Mu       float64
	Nesterov bool
}

Config provides configuration settings for an SGD optimizer.

func NewConfig

func NewConfig(lr, momentum float64, nesterov bool) Config

NewConfig returns a new SGD Config.

type SGD

type SGD struct {
	Config
	Alpha float64
}

SGD implements the SGD gradient descent optimization method.

func New

func New(c Config) *SGD

New returns a new SGD optimizer, initialized according to the given configuration.

func (*SGD) Delta

func (o *SGD) Delta(param nn.Param) mat.Matrix

Delta returns the difference between the current params and where the method wants it to be.

func (*SGD) Label

func (o *SGD) Label() int

Label returns the enumeration-like value which identifies this gradient descent method.

func (*SGD) NewSupport

func (o *SGD) NewSupport(r, c int) *nn.Payload

NewSupport returns a new support structure with the given dimensions.