Mathematics to understand data

Mathematics to understand data

The Center for Mathematical Modeling has created a new strategic area: CMM Data. The unit will be focused on developing advanced technology to extract information from data, with the aim of solving key problems of society, industry, public policy, and other sciences.

The world is in the midst of a data revolution. Thanks to the wide availability of sensors and to advances in high-performance computing and mathematics, it is possible today to explain phenomena which we would have deemed impossible to understand before. This is what we call data science.

Since its foundation in 2000, the Center for Mathematical Modeling at Universidad de Chile (CMM) has carried out projects that demand the obtention and structuring of databases, the formulation of methods to understand that data, and the interpretation of the results.

In a new step towards addressing the challenges that data science brings to the society, academy and industry, in 2018 CMM adds a new strategic area from 2018: CMM Data.

The creation of this new unit at CMM is in line with the challenges in the research and development of advanced mathematical solutions in the other CMM strategic areas: Bioinformatics and Health, Mathematical Education, Resource Management, and Mining.

The creation of CMM Data is a milestone in the history of the center in data science. Other key developments have been:
– The seminars in the discipline carried out throughout Chile since 2008.
– The Astroinformatics Lab established in 2009
– The National Laboratory of High-Performance Computing installed in 2011.
– The acquisition of the most powerful supercomputer in Chile in 2014.
– The marketing analytics solutions developed with the Chilean startup NoiseGrasp since 2014, and
– The Games (LINK) group created in 2015.

Through CMM Data’s multidisciplinary teams and the most advanced computational infrastructure in Chile, the group will deal with data-driven projects involving data acquisition, transmission and storage, machine learning, high-performance computing, and visualization.

These methods are transversal to the other CMM strategic applied areas. The aim is to use advanced technology and science to extract the value hidden in data.

“In general, understanding these tools to unlock their full potential is challenging,” says Felipe Tobar, director of the area. “Our approach to data analysis is a scientific one: in addition to implementing state-of-the-art methods, we focus on researching new methodologies. The design of solutions from a scientifically-rigorous and formal standpoint allows us to address data science problems in a confident and interpretable manner.”

According to Daniel Remenik, the other director of the unit, “CMM is characterized by its modeling capability, too. We have a lot of techniques and types of models to solve a problem. But one has to understand that problem and really model, which is a previous or parallel stage to data science. CMM has a unique experience in modeling for problems which are relevant for the country.”

Many world-class institutions work with the center: University of Harvard, Centre National de Recherche Scientifique, INRIA, École Polytechnique, Collège de France, the Massachusetts Institute of Technology, and Amazon Web Services. Being part of the Faculty of Physical and Mathematical Sciences at Universidad de Chile, its researchers have the chance of interacting and working together with different professionals from different areas on relevant issues for Chile.


“There are many key problems for the country where widely known tools are not enough, be it because relevant solutions have not been developed yet or because the problem is too complex. This is where CMM has a lot to offer,” said Remenik.

Today, CMM is carrying out projects in several fields: marketing analytics, occupational safety, astroinformatics, financial risk, logistics, retail, financial fraud, sustainability, and public security.

A key challenge is the implementation of a new extension for Leftraru, the most powerful supercomputer in Chile.

The approach taken by CMM in different applications is tailored specifically to each case. Sometimes, the center develops new tools to solve problems involving large volumes of data or when data is poorly structured. In other cases, the center helps institutions find challenges they can solve by making use of their data. Solutions developed by CMM in the past can sometimes be adapted to help with new problems. And CMM’s technologies can also be useful for emerging data analysis consulting firms.

“We address substantial problems which demand innovative solutions. Our motivation is to tackle problems whose solutions require original inputs from Mathematics,” explained Remenik.

Professional training is a relevant challenge for CMM Data, too. Members of the group have taught courses in companies, international organizations, and public institutions. They are also involved in the opening of a master program in data science and undergraduate courses in the subject.

“There is certainly valuable information hidden in the data. I firmly believe that this information can be extracted in a sound and scalable manner through the development of principled scientific and computational methods. This is our endeavour at CMM,” concludes Tobar.

Posted on Jun 27, 2018 in Frontpage, News