e5e63da79fcd2bebbd7cb8bf1c1d0274-Reviews.html

In total, the paper is meticulous in suggesting the framework of chemical reaction networks and mapping belief propagation to it, but the experiments appear a little lacking in real scope. Suggestions: a)more and more convincing experiments with more complicated and/or bigger graphs b)better theoretical explanation of damped BP in relation to this work c) discuss how reaction speeds can be implemented in reality with different kappas. I expect them to be regulated through chemical compounds, which would most likely lead to discrete subsampling of the speed-space. Would this lead to local minima or other problems during inference? Are the assumptions of the'perfect chemical reaction network' based on arbitrary species realistic? Where's the catch if graphs get bigger and have largewr state spaces and hundreds/thousands of chemical species are needed to implement a problem.