Mathematics of Data Science
Source: https://www.utwente.nl/en/education/master/programmes/applied-mathematics/specialisation/mathematics-data-science/ Parent: https://www.utwente.nl/en/education/master/programmes/applied-mathematics/
Develop robust, reliable and explainable mathematical models and machine learning algorithms for analysing data arising in diverse applications.
In our increasingly digital world, data is everywhere. It’s in your social network timeline, in your fitness tracker, in MRI scans or in the transactions of a bank. These huge amounts of data are full of valuable information, but extracting that information so that you can put it to good use, can be extremely complicated. It is for good reason that data science is often referred to as a black box: very often there is a lack of explainability and transparency of the algorithms used. In the specialisation in Mathematics of Data Science, you will aim to open this black box and learn to fundamentally understand, improve and develop mathematical models and machine learning algorithms that are essential for analysing data in a wide variety of fields.
When developing algorithms and methods in data science, many intricate choices are to be made. Should we favor performance, or explainability? How much structure should we assume, and how much should we let the data speak for itself? In this specialisation, you will learn how to navigate these questions guided by mathematical and statistical principles.
José Alberto Iglesias Martínez, Assistant Professor
What is Mathematics of Data Science
If you’re interested in gaining a fundamental, mathematical understanding of data science, this specialisation is right for you. You will learn how to use the underlying theory from, for instance, statistics, functional analysis or graph theory, to optimally employ and further develop the field of data science. This specialisation will enable you to pinpoint complex problems that occur in certain algorithms and fix them. Eventually, your knowledge of data science will continue where that of non-mathematical data scientists ends.
Examples of courses you (can) follow during this specialisation:
- Would you like to master the mathematical bases underlying current and future deep learning architectures by learning about their modeling, optimization and generalizability? In the course Deep Learning: from Theory to Practice you can delve into the foundations of methods reaching astonishing performance in a myriad of applications, ranging from computer vision, robotics, and pattern recognition to generative modelling.
- Assessing air quality, predicting weather patterns, or analyzing the epicentres of earthquakes. The course Spatial Statistics is key to the analysis of spatially correlated data laying the mathematical foundations for these purposes.
- From social media to the neural connections in our brain, from power grids to financial markets: in today's hyper-connected world networks are everywhere. Many of these share common properties like power-laws (a few nodes have many connections, while most nodes have few) or small-worlds (most nodes can be reached by a small number of steps). If you want to understand these characteristics and find out how they impact critical processes such as epidemics, the course Complex Networks is just for you.
With your ability to understand and develop reliable, robust mathematical methods, you will be a great asset to many organisations in a variety of fields. Your knowledge will be particularly relevant in sectors where inference from data should be combined with underlying mathematical structures and models, as is the case in companies employing digital twins technology and in medical imaging applications. Explainability and robustness guarantees are crucial for many sensitive use cases, such as personalized patient diagnoses and other medical applications. Besides, more classical data science applications will also be within your reach. These range from detecting fraudulent transactions in the financial sector to advancing text-based techniques like natural language processing and large language models, to large-scale industrial applications like predicting power consumption for the optimisation of energy distribution.
AI for Health
If you are interested in learning how data science can fundamentally impact healthcare through a hands-on including case studies and a master project close to direct applications, you can choose to focus your specialisation on a specific profile: Artificial Intelligence for Health.
What will you learn?
As a graduate of this Master's and this specialisation, you have acquired specific, scientific knowledge and skills and values, which you can put to good use in your future job.
Knowledge
After completing this Master’s specialisation, you:
- have in-depth knowledge of the mathematical principles and structures behind machine learning methods;
- can comprehend the statistical foundations of data science methods, as well as the underlying complex phenomena and structures present in data;
- can implement algorithms for data analysis using state-of-the-art programming languages for statistical computing and data visualisation.
Skills
After successfully finishing this Master’s specialisation, you:
- are able to design and deploy mathematical models and algorithms that can be used to analyse data, draw conclusions and make decisions from it;
- understand the mathematical reasoning underlying the analytical tools for data analysis and interpret the output of corresponding algorithms on this basis;
- can effectively interact with specialists in different application areas of data science through a careful understanding of requirements and transparent communication of results.
Values
After completing this Master’s specialisation, you:
- Are able to identify and mitigate biases in data, algorithms, and the interpretations your data analyses are based on;
- Ensure transparency in methodologies and make analyses reproducible for peer validation and improvement;
- Display critical thinking towards data-driven decision processes, considering the explainability, robustness, and generalisability of the methods employed.
Other master’s and specialisations
Is this specialisation not exactly what you’re looking for? Maybe one of the other specialisations suits you better. Or find out more about these other related Master’s:
- Master’s in Business Information Technology - specialisation Data Science & Business
- Master’s in Computer Science - specialisation Data Science & Technology
- Master’s in Electrical Engineering - specialisation Computer Vision & Biometrics