## Mauro Camara Escudero

### University of Bristol

I'm a first year Ph.D. student in Computational Statistics and Data Science (COMPASS) at the University of Bristol. My main research interests lie at the intersection of scalable sampling and variational methods with modern neural density estimation techniques, especially in the context of likelihood-free inference. I am also interested in the application of such methods to population genetics data. I co-run the Bristol Neural Networks reading group with Pierre-Aurélien Gilliot. I also run the COMPASS Seminar Series at the University of Bristol. Some of my R portfolios are available here

### Interests

• Machine Learning
• MCMC and SMC Samplers
• Variational Inference and VAEs
• Neural Density Estimation
• Density-Ratio Estimation and GANs
• Genomics & Population Genetics

### Education

• PhD in Computational Statistics and Data Science, 2020-2024

University of Bristol

• MRes in Computational Statistics and Data Science, 2020

University of Bristol

• BSc in Mathematics with Year in Employment, 2019

University of Southampton

# Experience

#### Uniper

Jun 2017 – Aug 2018 Nottingham
Responsibilities include:

• Wind Power Forecasting
• Windfarm layout optimization
• Detecting gas turbines failures using CV
• SQL, Cassandra, Python

#### Dr. Richardson Giles, University of Southampton

May 2016 – Jul 2016 Southampton
Modelled and simulated particle hopping in solar panels with Python.

# Ramblings

Tutorials, courses and various other rambling.

# Recent Posts

### Assessing a Variational Autoencoder on MNIST using Pytorch

Learn how to visualize the latent space and generate data using a VAE in Pytorch.

### Minimalist Variational Autoencoder in Pytorch with CUDA GPU

Variational Autoencoders in Pytorch with CUDA GPU

### Variational Auto-Encoders and the Expectation-Maximization Algorithm

Relationship between Variational Autoencoders (VAE) and the Expectation Maximization Algorithm. Simple Explanation

### Towards SMC: Using the Dirac-delta function in Sampling and Sequential Monte Carlo

We derive the Dirac-delta function, explain how to use to approximate an Empirical PDF for a sample.

### Towards SMC: Sequential Importance Sampling

Sequential Importance Sampling intuition simply explained for SMC

# Contact

• Woodland Road, Bristol, BS8 1TH
• Turn right at the entrance to office GA.14