StackGenVis: Alignment of Data, Algorithms, and Models for Stacking Ensemble Learning Using Performance Metrics
https://doi.org/10.1109/TVCG.2020.3030352
You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
34 lines
929 B
34 lines
929 B
4 years ago
|
'use strict'
|
||
|
|
|
||
|
|
var tape = require('tape')
|
||
|
|
var pack = require('ndarray-pack')
|
||
|
|
var unpack = require('ndarray-unpack')
|
||
|
|
var zeros = require('ndarray-scratch').zeros
|
||
|
|
var grad = require('../fdg')
|
||
|
|
|
||
|
|
function g(X, bc) {
|
||
|
|
var Y = pack(X)
|
||
|
|
var dX = grad(zeros(Y.shape.slice().concat([Y.dimension])), Y, bc)
|
||
|
|
var perm = [Y.dimension]
|
||
|
|
for(var i=0; i<Y.dimension; ++i) {
|
||
|
|
perm[i+1] = i
|
||
|
|
}
|
||
|
|
return unpack(dX.transpose.apply(dX, perm))
|
||
|
|
}
|
||
|
|
|
||
|
|
tape('ndarray-gradient', function(t) {
|
||
|
|
|
||
|
|
t.same(g([1,2,3], 'wrap'), [[0.5, -1, 0.5]])
|
||
|
|
t.same(g([1,2,3], 'clamp'), [[-0.5,-1, -0.5]])
|
||
|
|
t.same(g([1,2,3], 'mirror'), [[0, -1, 0]])
|
||
|
|
|
||
|
|
t.same(g([1,2], 'wrap'), [[0,0]])
|
||
|
|
t.same(g([1,2], 'clamp'), [[-0.5, -0.5]])
|
||
|
|
t.same(g([1,2], 'mirror'), [[0,0]])
|
||
|
|
|
||
|
|
t.same(g([[1,2,3]], 'wrap'), [[[0,0,0]],[[0.5, -1, 0.5]]])
|
||
|
|
t.same(g([[1,2,3]], 'clamp'), [[[0,0,0]],[[-0.5,-1, -0.5]]])
|
||
|
|
t.same(g([[1,2,3]], 'mirror'), [[[0,0,0]],[[0, -1, 0]]])
|
||
|
|
|
||
|
|
t.end()
|
||
|
|
})
|