diff --git a/statistics/readme.md b/statistics/readme.md
new file mode 100644
index 0000000..41a49da
--- /dev/null
+++ b/statistics/readme.md
@@ -0,0 +1,23 @@
+
+# Overview 
+
+This is for Statistics related Papers including topics like 
+
+- Bayesian Inference 
+- MCMC
+- Variational Bayes 
+
+# Bayesian Inference 
+
+## Variational Bayes Monte Carlo 
+
+- Original Paper [Variational Bayesian Monte Carlo](https://arxiv.org/abs/1810.05558?fbclid=IwAR2Irobgi5jW_RVJL4iRyv0QS_WfmgzKk-KmyoeXSZf3X26eFkz8gaTRxk4) published at [NIPS 2018](https://papers.nips.cc/paper/8043-variational-bayesian-monte-carlo)
+
+- [Summary on GitHub](vbmc_20190505_1832_1/) ([MathJax Plugin for GitHub](https://chrome.google.com/webstore/detail/mathjax-plugin-for-github/ioemnmodlmafdkllaclgeombjnmnbima) suggested to render Latex browser side)
+
+- [Summary on Kaggle](https://www.kaggle.com/nicolabernini/papersummary-variational-bayes-monte-carlo)
+ 
+
+
+
+Work in progress 
diff --git a/statistics/vbmc_20190505_1832_1/VBMC_Algo1.PNG b/statistics/vbmc_20190505_1832_1/VBMC_Algo1.PNG
new file mode 100644
index 0000000..b14b848
Binary files /dev/null and b/statistics/vbmc_20190505_1832_1/VBMC_Algo1.PNG differ
diff --git a/statistics/vbmc_20190505_1832_1/VBMC_Performance1.PNG b/statistics/vbmc_20190505_1832_1/VBMC_Performance1.PNG
new file mode 100644
index 0000000..549746f
Binary files /dev/null and b/statistics/vbmc_20190505_1832_1/VBMC_Performance1.PNG differ
diff --git a/statistics/vbmc_20190505_1832_1/readme.md b/statistics/vbmc_20190505_1832_1/readme.md
new file mode 100644
index 0000000..34f860d
--- /dev/null
+++ b/statistics/vbmc_20190505_1832_1/readme.md
@@ -0,0 +1,61 @@
+# Paper Summary - Variational Bayesian Monte Carlo
+
+Original Paper 
+
+[Variational Bayesian Monte Carlo](https://arxiv.org/abs/1810.05558?fbclid=IwAR2Irobgi5jW_RVJL4iRyv0QS_WfmgzKk-KmyoeXSZf3X26eFkz8gaTRxk4)
+- Published in [NIPS 2018](https://papers.nips.cc/paper/8043-variational-bayesian-monte-carlo)
+
+The **goal** is about performing Bayesian Inference hence computing the Model Posterior given a Dataset 
+
+Sticking to paper notation we have 
+
+- $\mathcal{D}$ : Dataset 
+- $x \in \mathcal{X}$ : Model Parametrization 
+
+the goal is to compute 
+
+1. Model Posterior 
+
+$$ P(x | \mathcal{D}) = \frac{P(\mathcal{D} | x) P(X)}{P(\mathcal{D})} $$
+
+
+2. Marginal Likelihood or Model Evidence 
+
+$$ P(\mathcal{D}) = \int_{\mathcal{X}} P(\mathcal{D} | x) P(x) dx $$
+
+
+Of course in practical cases the closed form computation is impossible because of intractable integral hence **approximation methods** need to be used 
+
+## Approximation Methods 
+
+The cost related to the approximation methods regards the **kind** and **amount of knowledge** that is needed: 
+
+- **gradient** is typically a valuable information but not always accessible, like in the case of Black Box Functions 
+- **samples** are accessible also in the case of black box functions, however as they have an associated **evaluation cost** the **samples efficiency** of the estimation algorithm is an important aspect 
+
+The Variational Inference Framework is based on the of approximating the True Posterior $P(x | \mathcal{D})$ with a simpler Parametric Function $q_{\theta}(x)$ and to fit its params $\theta$ so to make it as simple as possible to the original one 
+
+This params fitting problem is defined as a Similarity Distance Minimization problem between the 2 PDFs hence using an appropriate PDF Similarity Function (e.g. KL Divergence)
+
+
+
+## Goal 
+
+This paper proposes a method working with **Black Box Model Likelihood** (no gradient needed) to perform **Model Posterior Approximation** via **Variational Inference** using 
+- Active Sampling strategy, to make it a **data efficient** method 
+- Bayesian Quadrature Framework relying on Gaussian Process as a prior to approximate the Likelihood 
+
+
+# Algorithm 
+
+![Algo1](https://raw.githubusercontent.com/NicolaBernini/PapersAnalysis/statistics_20190505_1832_1/statistics/vbmc_20190505_1832_1/VBMC_Algo1.PNG)
+
+
+# Performance 
+
+![Performance1](https://raw.githubusercontent.com/NicolaBernini/PapersAnalysis/statistics_20190505_1832_1/statistics/vbmc_20190505_1832_1/VBMC_Performance1.PNG)
+
+
+
+Work in progress 
+