2
votes
2answers
136 views

Is alpha divergence a convex divergence measure?

Alpha divergence is defined as following : $$ D_\alpha(p||q) = \frac{1}{\alpha (1-\alpha)} \left( 1- \int _x p(x)^{\alpha} q(x)^{(1-\alpha)} dx \right) $$ if the distributions are restricted to ...
1
vote
0answers
64 views

Maximum of the expectation of a concave function

Let's have a function $f(x, \theta)$, and some probability distribution on $x$. Let's say I have found $\theta^* = \operatorname{argmax}(f(E[x], \theta) $, and $f$ is concave in $x$. I would like to ...
1
vote
2answers
167 views

Entropy expression optimization with Langrange multipliers

I have recently encountered variants of the following expression: \begin{equation} S = H(a,b,c,d)-H(a+b,c+d) \end{equation} where $H$ is the Shannon entropy function, that is $H(X)=\sum_{x\in X}-x\log ...
2
votes
2answers
879 views

Using KKT conditions to maximize function

The goal is to maximize the following function: \begin{align} K_p(q) = q\log \frac{q}{p} + (1-q)\log \frac{1-q}{1-p} \end{align} where \begin{align} 0 \leq q \leq 1 \end{align} and $p \in (0,0.5)$ and ...