Class Layer

Neuralnetwork layer

Index

Constructors

Accessors

dependentLayers

Methods

bind calc grad toObject update fromObject registLayer

Constructors

constructor

new Layer(obj: object): Layer
Parameters
- obj: object
  Config
Returns Layer
- Defined in lib/model/nns/layer/base.d.ts:31

Accessors

dependentLayers

get dependentLayers(): string[]
List of names of other layers dependent on this layer.

Returns string[]
- Defined in lib/model/nns/layer/base.d.ts:36

Methods

bind

bind(
    values: {
        input: Tensor | Matrix | { [key: string]: Tensor | Matrix };
        n: number;
        rest: any;
        supervisor?: Matrix;
    },
): void
Bind pre-condition values.
Parameters
- values: {
      input: Tensor | Matrix | { [key: string]: Tensor | Matrix };
      n: number;
      rest: any;
      supervisor?: Matrix;
  }
  Binding object
  - input: Tensor | Matrix | { [key: string]: Tensor | Matrix }
    Input data for neuralnetwork
  - n: number
    Data count
  - rest: any
    Some other values
  - Optionalsupervisor?: Matrix
    Supervisor data
Returns void
- Defined in lib/model/nns/layer/base.d.ts:45

calc

calc(...x: (Tensor | Matrix)[]): Tensor | Matrix | (Tensor | Matrix)[]
Returns calculated values.
Parameters
- ...x: (Tensor | Matrix)[]
  Input values
Returns Tensor | Matrix | (Tensor | Matrix)[]
Output values
- Defined in lib/model/nns/layer/base.d.ts:58

grad

grad(...bo: (Tensor | Matrix)[]): Tensor | Matrix | (Tensor | Matrix)[]
Returns gradient values.
Parameters
- ...bo: (Tensor | Matrix)[]
  Input value of backpropagation
Returns Tensor | Matrix | (Tensor | Matrix)[]
Output value of backpropagation
- Defined in lib/model/nns/layer/base.d.ts:64

toObject

toObject(): PlainLayerObject
Returns object of this layer.

Returns PlainLayerObject
Object represented this layer
- Defined in lib/model/nns/layer/base.d.ts:77

update

update(
optimizer: { delta: (arg0: string, arg1: Matrix) => Matrix; lr: number },
): void
Update parameters.
Parameters
- optimizer: { delta: (arg0: string, arg1: Matrix) => Matrix; lr: number }
  Optimizer for this layer
Returns void
- Defined in lib/model/nns/layer/base.d.ts:69

`Static`fromObject

fromObject(obj: PlainLayerObject): Layer
Returns layer from JSON.
Parameters
- obj: PlainLayerObject
  Object represented a layer
Returns Layer
Layer
- Defined in lib/model/nns/layer/base.d.ts:21

`Static`registLayer

registLayer(name?: string, cls?: Layer): void
Regist layer class.
Parameters
- Optionalname: string
  Name of the layer
- Optionalcls: Layer
  Layer class
Returns void
- Defined in lib/model/nns/layer/base.d.ts:27