Programme in detail

Smart Conversational Assistants: While voice assistants have become commonplace in many consumer devices like mobile phones or smart speakers, their functionality and capabilities are typically still limited to simple commands. Interacting with such assistants is still far from the natural and human-like interaction originally envisioned. Therefore, current research at TU Delft focuses on more sophisticated techniques for natural language processing, semantic query processing, and human-computer interaction.

Recent advances in AI allow a novel and deeper understanding of established traffic networks like road, rails or flight networks. This can cover predicting traffic, or modelling behavior of the network under changing conditions. Using AI, current research at TU Delft uses this knowledge to optimize and control such networks, even allowing the network’s operators to react to dynamic changes in real-time.

Medical imaging techniques like for example MRI or CT are at the core of many modern medical diagnosis and treatment processes. However, current approaches are still hampered by their lacking analytical capabilities and slow processing speeds. Research at TU Delft focuses on developing new medical imaging techniques which can be interactive and real-time, while at the same time providing stronger analytics.

Many current AI technologies result in systems whose decisions are hard to understand and explain. This threatens the trust stakeholders and users have into those systems, and it is hard to prove that those systems indeed behave as intended. Furthermore, many AI-driven decisions like in legal or financial applications touch on delicate areas of peoples’ personal lives, and thus ethical issues with respect to fairness, privacy and bias need to be considered which still poses a challenge of current systems.

Is about the development of all types of medical and/or health support systems such as diagnostic support systems, long-term buddies to help people in establishing and maintaining healthy behaviours, eHealth Systems, Electronic Patient Record system etc. Common factors in these systems are e.g., security, privacy, distributed systems, and various role players . In the Master Track Software Technology you will learn to design software systems that allow access to data on a need to know basis only, that are secure, that can help trace the patients history, that reliably and distributedly store data, that provide the users with information that suits their role, their way of information processing, their personality, and their agenda. These systems are inherently distributed, need to be properly embedded, might have to work as web-based systems, and have to understand human motives, affective states and information processing capabilities and need to continuously improve their behaviour. You will learn about cognitive and software designing, distributed architectures and algorithms, agent technology, cybersecurity, cloud computing techniques, data visualization, machine learning, web data management, affective computing, and information processing techniques.

Is about creating coupled systems that optimally control the power systems in relation to physical infrastructures , environment and the interests of the consumers and producers of energy, in real-time. Smart Grids are systems of intelligent systems in which smart meters, solar cells, wind mills, electric cars, intelligent houses, power plants, and so on form essential elements. They refer to the whole infrastructure connecting all systems that consume and produce energy. The challenge is that these systems have to meet overall robustness criteria and that they have to be able to reconfigure themselves if necessary. Efficient use has to be made of all appliances, e.g., electric cars in terms of when to consume energy, when to use them to store energy, and when to release energy into the grid. Optimiszation has to be balanced for the individual, but also system-wide. As all components are distributed over land and have their own controllers, you will have to study distributed artificial intelligence, distributed architectures, software engineering, agent technology, cybersecurity, distributed data management, and optimisation and collaboration techniques. Furthermore, all appliances have to be programmable in an easy way and such a way that the appliances can adapt and learn. This implies that also machine learning, adaptation, and programming languages are important study goals.

Is about the design and deployment of computer and information systems that support and control the various infrastructures. Infrastructures that are vital for today’s society and economy like transportation infrastructures, infrastructures for utilities, and infrastructures that are inherently computer systems in themselves (telecommunications systems, the internet). Computer systems control these infrastructures at global level by air traffic control and routing internet traffic, but also on individual components such as cars and smart phones. As a consequence, these infrastructures have global control points, but may also allow the individual components to communicate among each other and show collective behavior, such as cars on highways trying to avoid congestion. The challenge of designing these systems is to make the individual components behave correctly and predictably, and to guarantee the optimal and secure operation of the global infrastructure. To meet this challenge you will learn to design distributed and cloud-based systems, algorithms for logistic planning, agent technologies, distributed and embedded software and data engineering methods, and security-enhancing technologies, e.g., for preventing a malicious user from causing traffic jams or hacking the brakes of cars.

Is about the design and deployment of computer and information systems that support and control human interaction with their living environment in a sustainable way. To ensure sustainability, software and information systems need to register and observe relevant properties of the living environment and of the way humans interact with it. This includes for example gathering information about the way people live, work and move in given geographic areas, about rainfall and temperature measurement across specific locations, or about the water levels in urban infrastructures. These Systems will employ physical sensors embedded in devices and infrastructures or techniques for social sensing through direct interaction with humans and techniques of data fusion and enrichment to make information actionable. Through interfaces with infrastructural systems the Environment-related systems contribute to the control of the global environment, and by supporting people to exhibit a behaviour that promotes sustainability the systems also contributes to a personal environment. In the master track of software technology, you will be challenged to design software systems that produce meaningful and actionable knowledge about the environment and that exhibit properties that ensure the system operates at scale, in real-time, and securely. To meet this challenge, you will learn to design distributed and cloud-based systems, agent technology, distributed and embedded software and data designing methods, and security-enhancing technologies.

Discovering important data patterns in huge medical data streams plays a vital role in the healthcare system. Examples of these data streams include large amounts of molecular data based on next generation sequencing, medical imaging like MRI, X-rays, CT or PET scans, and sensory data measuring health activity like motion or sleeping patterns. To unlock this data, you will learn e.g., to develop pattern recognition algorithms for associating DNA sequence data with diseases, multimedia search and retrieval techniques to gather relevant data, and to visualise medical data sets in a comprehensible way.

Unlocking Energy-relevant Data concerns the gathering and processing of relevant data from smart meters, solar cells, windmills, electric cars, intelligent houses and power plants, etc. to optimise our usage of energy. You will learn to develop algorithms for recognition of energy usage patterns, user modelling techniques for personalisation and preference elicitation, (multi-modal) data fusion algorithms to integrate data, and visualisation techniques to inform human operators.

Unlocking Infrastructure Operation Data concerns collecting, storing, and analysing data on the operation of various infrastructures ranging from transportation infrastructures, infrastructures for utilities, telecommunications, and the Internet. You will learn to design algorithms for data analysis, for analysing networks to optimise them or to make them more robust, for detecting the behaviours and preferences of infrastructure users, and for recognizing features from surveillance videos.

Unlocking Environment-relevant Data concerns the gathering, storing, enriching, analysing, and applying data on the interaction that humans have with their living environment. Examples of the societal challenges that we are facing include climate-adaptive buildings and cities, water management in urban environments, and improved safety in our delta regions. To work as a data scientist on these challenges, you will learn techniques and algorithms for the collection, integration and enrichment of data (from sensors, humans and infrastructures), and for modelling, simulation and analysis. In addition, you will learn how to design and implement software and information architectures for performing this type of analysis.