From 32f2429562ba7ab973b177fdbbfb0a862814c2eb Mon Sep 17 00:00:00 2001 From: MaxenceGiraud Date: Mon, 11 Jan 2021 22:36:08 +0100 Subject: [PATCH] [Report] Add conclusion --- report/main.tex | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/report/main.tex b/report/main.tex index 7dd0400..982fbe3 100644 --- a/report/main.tex +++ b/report/main.tex @@ -279,7 +279,12 @@ \section{Experiments} \section{Conclusion} -Other continuation of the paper can be made : properly express the optimal dimensionalanity (use conditional probabilities)... +In this paper we explored how PCA can be re-formulated in the Bayesian paradigm. The main contribution of this new approach is its ability to approximate the effective number of latent space dimension.\\ +We then saw how can the formulation can be combined with the Kernel PCA and how to construct a Bayesian Kernel PCA.\\ + +We saw some surprising results with Probabilistic Kernel PCA and how even this formulation can be used to approximate the number of effective latent space dimension, this behavior could be investigated in future work.\\ + +Other continuations of the original paper on Bayesian PCA by Bishop can be made, we can think of properly expressing the optimal dimensionality using conditional probabilities which would be very interesting with the goal of the algorithm. \bibliography{bayespca}