Fernanda Mora-Alba

Presented on the International Conference on Operations Research for Development, ICORD 2016. The paper aims to present local layered algorithmic model for the topological design of rural telecommunications networks, which can be used to estimate the infrastructure requirements and associated costs that are necessary to deliver broadband coverage to unconnected rural communities in Mexico. Check the slides here.

Worked with Profr. Eric Nyberg as a Visiting Research Scholar in Machine Learning models for Q&A in the Language Technologies Institute at Carnegie Mellon University. The project proposed a modular methodology to solve a reading comprehension problem in the SQuAD dataset presented by [Rajpurkar et al. 2016]. Check the weekly deliverables (eight) here and the summary slides here.

My bachelor's degree thesis had two main objectives. First, to give a general overview of Deep Learning: what is it, how does it differentiate from general Machine Learning, on what assumptions it relies and notable architectures and applications. Second, to apply a Deep Learning model (a Long Short Term Memory network) to a pressing problem in Mexico, namely, electric energy consumption. The slides can be checked here. The code of the Deep Learning model can be checked here.

Diabetes Mellitus type II represents a serious health problem in Mexico with 6.4 million diagnosed patients. The economic cost of the disease is hard to calculate since it generates two types of costs: direct costs including medical treatment and drugs, and indirect costs including costs due to early mortality and disability. In this study an estimation of both costs is proposed.

Progress in automatic reading comprehension has been slow. In response, recently, machine learning approaches have been taken. There has been two major efforts towards the advance of machine reading comprehension: creation of datasets and development of models; we discuss both in this survey: the most relevant and the state-of-the art approaches. Check summary slides here.

In this survey we explore the most novel and relevant approaches for malware detection in Android Operative System using Machine Learning techniques. Specifically, we reviewed the most cited published works from 2012 to 2016. We analyze Machine Learning relevant elements: the features, the dataset, the models and the performance results. We conclude the work with some opportunity areas together with interesting venues of work regarding a general approach to detect Android malware.