Documentation
¶
Overview ¶
Package sochastic implements a sochastic connected layer and combiner
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type Sochastic ¶
type Sochastic struct {
// contains filtered or unexported fields
}
func (*Sochastic) Disregard ¶
Disregard tells whether putting value false at position n would not affect any feature output (as opposed to putting value true at position n). This excludes training samples early which would have no impact, because the next layer would see the same thing regardless of what we put.
type SochasticLayer ¶
type SochasticLayer struct {
// contains filtered or unexported fields
}
func MustNew ¶
func MustNew(size int, maxbits byte, seed uint32) *SochasticLayer
MustNew creates a new full layer with size and bits
func New ¶
func New(size int, maxbits byte, seed uint32) (o *SochasticLayer, err error)
New creates a new full layer with size and bits
func (*SochasticLayer) Lay ¶
func (i *SochasticLayer) Lay() layer.Combiner
Lay turns full layer into a combiner
Source Files
Click to show internal directories.
Click to hide internal directories.