Jensen–Shannon Divergence

KL-Divergence 의 symmetric 및 normalized 버전

• $\mathrm{JSD}(P\Vert Q)=\frac{1}{2}D(P\Vert M)+\frac{1}{2}D(Q\Vert M)$
  • $M=\frac{1}{2}(P+Q)$
  • $D(P\Vert Q)$ 는 두 분포 $P,Q$ 의 KL-Divergence
• $0\le \mathrm{JSD}(P\Vert Q)\le 1$

B) Related

C) References

• https://en.wikipedia.org/wiki/Jensen%E2%80%93Shannon_divergence