Categorical estimation task: Steady-state performance summary. Difference between the mean squared error of each algorithm and the optimal solution (Exact Bayes), that is, the average of over time, for all combinations of and together. For each algorithm, we plot the 42 values (7 times 6 values) of Figure 6 with respect to randomly jittered values in the -axis. The color coding is the same as in Figure 5. The error bars mark the standard error of the mean across 10 random task instances. The difference between the worst case of SOR20 and pf20 is significant (, two-sample -test, 10 random seeds for each algorithm). Abbreviations: See the Figure 5 caption. Note that MP1 and SMiLe are out of bound with a maximum at 0.53.
This site uses cookies. By continuing to use our website, you are agreeing to our privacy policy. No content on this site may be used to train artificial intelligence systems without permission in writing from the MIT Press.