Current students


Section: Computer Science and Engineering

Major Research topic:
Extracting Value from Scientific Data

In the last years, there has been an escalation in the creation and development of predictive models in various fields, which are increasingly pervasive and more and more essential. Being data-driven applications, predictive models depend heavily on the data used to design them so that the data management systems define what can be extracted from the data. Moreover, scientific data pose further challenges: they are a precious source of information in terms of time and cost, affected by experimental uncertainty, and coming from different sources and eras that make it difficult to use them in practice to extract knowledge. These characteristics make the applications dependent on many manual procedures, significantly limiting the development capabilities of predictive models, particularly on large numbers of data. In this context, there is the need to define a system capable of collecting scientific data while maintaining a certain level of quality of the repository. In a second moment, a series of methodologies are applied to automate, as far as possible, the predictive model development process while learning different aspects of the data and the previous versions of the predictive model. These techniques tackle different data science and data mining arguments such as machine learning, pattern recognition, and statistical and uncertainty analyses.