Machine learning for natural language processing & Data analytics and machine learning for smart grids
Inschrijven/RegisterAdditional information on demand for selected topics (see below)
prof. Chris Develder & Thomas Demeester
iGent, 9th floor, Office of Thomas & our PhD students
max 15 persons per session
We will be available to provide more background and information on the following subjects:
Electromagnetics research group
info moment
Prof. Ann Franchois, Prof. Sam Lemey, Prof. Hendrik Rogier, Prof. Dries Vande Ginste
Please schedule an appointment with one of the promoters on:
iGent, 11th floor
/
Interactive presentation of the master’s theses about:
Data-efficient Machine Learning
info moment
prof. Tom Dhaene, prof. Dirk Deschrijver
the promotors are available in their office on April 23 & April 26 in the afternoon
iGent, 9th floor, office 190.017 & 190.016
/
We offer a wide range machine learning topics, including automated machine learning, data-efficient machine learning, time series prediction, bayesian optimization, anomaly detection, plant growth modeling, drone path planning, health applications, ... Most topics can lead to a scientific publication and the research can be continued within a PhD thesis.
Design of Network Algorithms
Inschrijven/RegisterInfo moment
prof. Mario Pickavet, prof. Didier Colle, Pieter Audenaert
iGent, 10th floor, office 200.022 (office prof. Mario Pickavet)
/
Complex networks and graphs appear in a variety of applications: modelling of road traffic, social networks, biological (genetic) networks, the Internet, ... To tackle their challenging problems, clever network algorithms are needed. In these thesis topics, you design your own creative algorithm to tackle a real-life problem.
High-Performance & Programmable Networks
Inschrijven/RegisterInfo moment
prof. Mario Pickavet, prof. Didier Colle, prof. Wouter Tavernier
iGent, 10th floor, office 200.022 (office prof. Mario Pickavet)
/
IDLab is highly involved in European research projects on the future Internet: programmable networks with virtualized network functions, routing in extreme conditions, clean-slate network architectures, ... In these thesis topics, you are involved in such a European project while tackling a specific research challenge.
Neural networks, robotics, (machine) learning and brain-inspired computation
Inschrijven/Registerguided tour, info moment, group conversation
prof. Joni Dambre, prof. Francis wyffels, prof. Tony Belpaeme
iGent, 8th floor, hall & robotica lab
20 students per slot
On floor 8, we cover a range of topics related to machine learning, (deep) neural networks, brain-inspired computing, robotics, human-robot interaction and human learning. We offer a spectrum ranging from very applied to highly theoretical subjects and from very well-defined to speculative and exploratory research. In all thesis subjects, our often ambitious goals and methods are set in interaction with the student, taking his/her interests and skills into account. All topics are related to ongoing research in our group and good results often lead to a publication.
Bioinformatics
Inschrijven/RegisterInfo moment
prof. Jan Fostier, prof. Kathleen Marchal
iGent, 9th floor, 190.019 (office of prof. Fostier)
10
Information moment in which the content of the thesis topics related to bioinformatics can be discussed in more depth.
Design of high-speed and high-frequency electronics for communications and measurement systems
Inschrijven/RegisterDemos and info moment
prof. Johan Bauwelinck, prof. Guy Torfs, prof. Xin Yin, Peter Ossieur
iGent, 12th floor, demos at Design Lab, promotors and supervisors are also available in offices 220.015, 220.016 & 220.017 at April 23 & 26 in the afternoon
per demo: 15
Demonstration of recent design activities: chip design, PCB design, physical layer system design and signal processing on FPGA are well represented, e.g. chip design for multichannel 100Gb/s transmitter and receiver chips for optical links, transceiver design and opto-electronic co-design for RF-over-fiber, and fast real-time digital signal processing on FPGA for broadband wireless communications.
What is wrong with data and the Web (and how you can help fix it)
Inschrijven/Registera nice discussion
prof. Ruben Verborgh
Anywhere on 26 April (preferable early on, so people can go see demos after)
iGent, 10th floor, meeting room 10.2
20
No invention of the past 30 years has shaped the world more than the Web. Yet the Web is facing unprecedented challenges, not in the least the extreme centralization of data and services, and the fact that we have lost control over our personal data. Our research team is working on new approaches to tackle these issues. Instead of Big Data, we are doing Linked Data: instead of one big dataset, it is a very large number of small datasets. This brings unique challenges, and we need you to tackle them. Join this discussion and find out what exciting things you could be doing next year.
Knowledge Management
info moment
Inschrijven/Register
prof. Filip De Turck, Femke De Backere, Femke Ongenae, Maria Torres Vega
iGent, 9th floor, IDLab 9
15
The analysis of data combined with expert and domain knowledge allows to derive and to predict interesting events. To provide real-time decision-making, this analysis needs to be scalable and easy to interpret. By means of techniques such as semantic reasoning and machine learning, we research novel algorithms able to perform data analysis in an accurate and effective manner. Within this research field, 2 research goals can be distinguished. First, how can we incorporate domain knowledge into the data analytics process so the result is a more correct, personalized and context-aware decision. Second, how can we ensure that the decision made by the data analytics process are interpretable for domain experts. These techniques and this research can be applied to a number of application areas, such as IoT, smart factories or eHealth, where specific attention is given to behavioral change. Technologies that can come into play are for example, gamification, Virtual Reality (VR) and Augmented Reality (AR).
Holographic and Virtual Reality Video
Inschrijven/RegisterInfo moment and demo
Filip De Turck, Tim Wauters, Maria Torres Vega
iGent, 10th floor, meeting room 10.1
10 per session
Optimizing media transport over the network is crucial to enhance the experience of users of Virtual Reality and Holographic Type Communications. At IDLab, enhancements are studied on three levels. First, protocol enhancements are researched to optimize the transport of the video packets. Examples are HTTP/2 Push for reducing the start-up delay or WebRTC to allow for interactivity. A second way is to improve the network transport of the video packets through intelligent management components, e.g. to prioritize packets for faster delivery or to provide fairness in quality for multiple users. A last topic focuses on client-side machine learning techniques to help the video player in offering the user a high-quality experience: requesting the right segments at the right quality or predicting where the user is looking at exactly to minimize bandwidth waste in Virtual Reality.
Intelligent software systems for drones
Inschrijven/Registerinfo moment
prof. Bruno Volckaert, prof. Filip De Turck
April 23 in the afternoon until 16:30
iGent, 10th floor, office 200.012 (office of Jerico Moeyersons)
1 on 1 sessions are preferred, but a talk with multiple students at the same time is possible
A large variety of drones / UAVs is being introduced in our everyday lives, ranging from low-cost commodity drones to highly specialised ones. Developers are looking to come up with innovative applications utilising these drones, but are facing typical ‘emerging technology’ issues such as lacklustre APIs, making it harder than needed to develop applications for drones. At IDLab we therefore investigate software solutions that allow developers to unlock the true potential of drones.
Scalable microservice based distributed software systems
Inschrijven/RegisterInfo moment
prof. Bruno Volckaert, prof. Filip De Turck
iGent, 10th floor, office 200.012 (office of Tom Goethals, Merlijn Sebrechts & Sander Borny – depending on who the contact person is of the topic)
1 on 1 sessions are preferred, but a talk with multiple students at the same time is possible
Current (distributed) applications like those built on top of the City-of-Things smart city project require more and more support for application scalability, to cater to growing user bases and input / output data. Therefore, developers need to pay special attention to throughput and so-called light-weightness when creating the application developer frameworks / APIs that form the core of these innovations. At IDLab, we do research on the creation of such high-throughput software enablers.
Cybersecurity: autonomous attack detection and mitigation
Inschrijven/Registerinfo moment
prof. Filip De Turck, prof. Bruno Volckaert, Tim Wauters
iGent, 9th floor, office 190.031 (office of Laurens D’Hooge)
1 on 1 sessions are preferred, but a talk with multiple students at the same time is possible
Worldwide critical IT infrastructures are under daily attack from so-called script kiddies to state-sponsored hacker collectives. At IDLab we investigate how we can monitor, control and mitigate these large-scale cyber-attacks (distributed denial of service attacks, crypto-lock attacks from inside the network, etc.) by means of attacker profiling, anomaly detection and dynamic injection of virtual network elements (network filters, traffic load balancers, firewalls, etc.).
Mobile and wireless networks
Info moment to discuss thesis topics in more detail
prof. Ingrid Moerman, prof. Jeroen Hoebeke, prof. Eli De Poorter
Please schedule an appointment with the specific promotor
iGent, 11th floor, office of the promotor (210.017: prof. Ingrid Moerman, 210.016: prof. Jeroen Hoebeke & prof. Eli De Poorter)
/
Explanations about master’s theses concerning:
Querying the Web of Open Data for planning your route home
Inschrijven/RegisterInteractive demo with discussion
Pieter Colpaert, prof. Ruben Verborgh
April 26, entire afternoon, just drop by
AA Tower (next to the iGent tower), 7th floor (ask for Pieter and Anastasia’s office)
50
An infinite amount of data can be used to advise how to get home by public transit (time tables and real-time updates), on foot (where are the side walks, criminality rates of neighborhoods), by car (roadblocks or traffic events), using space elevators and so forth. Today, this kind of datasets are already published in a non-queryable way on open data portals. In the future, what would be the a better way to publish this data, so that it can be queried to also solve route planning queries? In this demo session you will find out how we intend to tackle this question, and why we genuinely wake up with a feeling every morning that our research is making the world a better place. The demos we will show in our web-browser however do not put forward concrete solutions. Instead, each of our demos is the start of a whole range of new questions seeking for an answer. Help us out by joining our research team as a master thesis student!
Knowledge graph generation and where you can make the difference
Inschrijven/RegisterDemo
Anastasia Dimou
AA Tower (next to the iGent tower), 7th floor (ask for Pieter and Anastasia’s office)
50
Knowledge graphs are used by a lot of companies, such as Google and Airbnb, and technologies, such as machine learning, decentralized personal storage, and artificial intelligence. However, not all data is available as a knowledge graph. Thus, the generation of knowledge graphs is of uttermost importance but it can be complex and time-consuming. Our team is working on new approaches to ease this generation. On the one hand, we have tools that generate knowledge graphs based on existing data sources, such as databases, files, Web APIs, and streams. On the other hand, we have tools that make it easier for users to define, via rules, how the knowledge graphs are generated. These tools either use a graph-based visualization of these rules or a human-friendly text-based representation of these rules. Join our demo session to find out how and where these tools are used and where you can make a difference.
Building a (data) oil refinery
Inschrijven/RegisterDemo
Anastasia Dimou
AA Tower (next to the iGent tower), 7th floor (ask for Pieter and Anastasia’s office)
50
Data is the new oil, and more data (i.e., more oil) means more money. The top 5 biggest companies in the world include Amazon, Google, and Facebook, that’s no accident. However, to go from crude oil to jet fuel, you need a refinery. Same so for data. Our team is working on data enrichment and validation systems, whilst generating interoperable unambiguous datasets. The resulting frameworks can generate high-quality datasets for machine learning, healthcare, and many more. Ka-ching. Our research focus is broad going from (A) a user-friendly editor to precisely define what you expect from the generated dataset, over (B) privacy-preserving data processes, to (C) complex integration of multiple data transformations. We need help on all of our A-B-C: front-end, back-end, algorithms, there’s something for everyone. There’s a lot to be discovered, and a lot to build from. Join our demo session for a practical introduction in our current tools and approaches, and discover in what way we can work together for further shaping data processes and data processing companies. Ka-ching.