func (m *MoM) CCM() {
for d := 0; d < m.M; d++ {
m.unsetZ(d)
z := util.Argmax(m.conditional(d))
m.setZ(d, z)
}
}
func (m *DPMoM) CCM() {
for d := 0; d < m.M; d++ {
m.unsetZ(d)
lcounts, mapping := m.conditional(d)
i := util.Argmax(lcounts)
m.setZ(d, mapping[i])
}
}
func (i *ITM) CCM() {
for _, d := range rand.Perm(i.M) {
for n := 0; n < i.N[d]; n++ {
i.unsetZ(d, n)
z := util.Argmax(i.conditional(d, n))
i.setZ(d, n, z)
}
}
}
func (l *LDA) CCM() {
for _, d := range rand.Perm(l.M) {
for n := 0; n < l.N[d]; n++ {
l.unsetZ(d, n)
z := util.Argmax(l.conditional(d, n))
l.setZ(d, n, z)
}
}
}
func (m *DPMoM) SUGS() {
for d := 0; d < m.M; d++ {
m.setZ(d, unassignCluster)
}
for d := 0; d < m.M; d++ {
lcounts, mapping := m.conditional(d)
i := util.Argmax(lcounts)
m.setZ(d, mapping[i])
}
}
func (b *NaiveBayes) Classify(data []int) interface{} {
lposts := make([]float64, b.LabelIndex.Len())
for label := range lposts {
lpost := b.LabelCounts[label]
for _, word := range data {
lpost += b.WordCounts[label][word]
}
lposts[label] = lpost
}
return b.LabelIndex.Item(util.Argmax(lposts))
}
func (i *ITM) Coinflip(pMax float64) {
for _, d := range rand.Perm(i.M) {
for n := 0; n < i.N[d]; n++ {
i.unsetZ(d, n)
var z int
if rand.Float64() < pMax {
z = util.Argmax(i.conditional(d, n))
} else {
z = util.SampleCounts(i.conditional(d, n))
}
i.setZ(d, n, z)
}
}
}
