Class ALMA

Approximate Large Margin algorithm

See

A New Approximate Maximal Margin Classification Algorithm. (2001)

Index

Constructors

Properties

_alpha _b _c _k _p _w _w0

Methods

fit predict update

Constructors

constructor

new ALMA(p?: number, alpha?: number, b?: number, c?: number): ALMA
Parameters
- Optionalp: number
  Power parameter for norm
- Optionalalpha: number
  Degree of approximation to the optimal margin hyperplane
- Optionalb: number
  Tuning parameter
- Optionalc: number
  Tuning parameter
Returns ALMA
- Defined in lib/model/alma.d.ts:12

Properties

_alpha

_alpha: number

_b

_b: number

_c

_c: number

_k

_k: number

_p

_p: number

_w

_w: any

_w0

_w0: number

Methods

fit

fit(x: number[][], y: (-1 | 1)[]): void
Fit model parameters.
Parameters
- x: number[][]
  Training data
- y: (-1 | 1)[]
  Target values
Returns void
- Defined in lib/model/alma.d.ts:31

predict

predict(data: number[][]): (-1 | 1)[]
Returns predicted datas.
Parameters
- data: number[][]
  Sample data
Returns (-1 | 1)[]
Predicted values
- Defined in lib/model/alma.d.ts:37

update

update(x: number[], y: -1 | 1): void
Update model parameters with one data.
Parameters
- x: number[]
  Training data
- y: -1 | 1
  Target value
Returns void
- Defined in lib/model/alma.d.ts:25