diff --git a/mdgp_based_wetchicken.pdf b/mdgp_based_wetchicken.pdf index 9cba5ae..b720c45 100644 Binary files a/mdgp_based_wetchicken.pdf and b/mdgp_based_wetchicken.pdf differ diff --git a/mdgp_based_wetchicken.tex b/mdgp_based_wetchicken.tex index 44baadf..5a86ede 100644 --- a/mdgp_based_wetchicken.tex +++ b/mdgp_based_wetchicken.tex @@ -154,7 +154,7 @@ What kind of turbulences exist? When does the canoeist fall down? How do the actions influence the system? -We formulate a graphical model in \cref{fig:graphical_model} based on the Multi-Modal Deep GP model~\cite{kaiser_data_2018}, which allows us to handle the multi-modality introduced by falling down the waterfall. +We formulate a graphical model in \cref{fig:graphical_model} using the data association with GPs (DAGP) model~\cite{kaiser_data_2018}, which allows us to handle the multi-modality introduced by falling down the waterfall. We specify this separation via the marginal likelihood \begin{align} \begin{split} @@ -248,7 +248,7 @@ Importantly, we only need a small number of Monte Carlo samples at every iterati \setlength{\tabcolsep}{1pt} \begin{tabular}{cSSS} \toprule - {N} & {NFQ} & {GP} & {MDGP} \\ + {N} & {NFQ} & {GP} & {DAGP} \\ \midrule 100 & 0.66 \pm 0.16 & \bfseries 1.41 \pm 0.01 & 1.18 \pm 0.09 \\ 250 & 1.71 \pm 0.07 & 1.54 \pm 0.01 & \bfseries 2.33 \pm 0.01 \\