A family of statistical symmetric divergences based on Jensen's inequality
Citations
109 citations
Cites background from "A family of statistical symmetric d..."
...Then the Jensen–Shannon divergence and the Jeffreys divergence can be rewritten [24] as...
[...]
...A family of symmetric distances unifying the Jeffreys divergence with the Jensen–Shannon divergence was proposed in [24]....
[...]
76 citations
Cites background from "A family of statistical symmetric d..."
...The smaller the Jensen–Shannon distance, the more similar the distribution of the two documents [11]....
[...]
64 citations
Cites methods from "A family of statistical symmetric d..."
...Because the partitioned features fV and fI are constrained by identity loss, thus the feature space is compact and we can use JS1(NV ,NI )[16] tomeasure the similarity....
[...]
56 citations
References
45,034 citations
5,933 citations
Additional excerpts
...…+ (1− α)ct) + α∇F (αct + (1− α)pi)), (57) That is, ct+1 = (∇F )−1 ( n∑ i=1 wi((1− α)∇F (αpi + (1− α)ct) + α∇F (αct + (1− α)pi)) ) (58) (Observe that since F is strictly convex, its Hessian ∇2F is positive-definite so that the reciprocal gradient is ∇F−1 is well-defined, see [Rockafeller(1969)].)...
[...]
5,228 citations