DAY 1 - Invited Speakers per track/session

Artificial Intelligence

Sander Bohté

Dr Sander M. Bohté is a senior researcher and PI in the CWI Machine Learning group. He received his BS and MSc in experimental physics from the University of Amsterdam in 1997, and his Ph.D in computer science from the University of Leiden in 2003. He currently works on bridging the field of neuroscience with applications thereof as advanced neural networks. He is a pioneer in the development of advanced and efficient spiking neural networks, including efforts at efficiently implementing such networks on GPUs. Additionally, he has worked extensively on biologically plausible deep neural networks that can learn from trial-and- error, in particular when relevant state information has to
be learned from the environment. His current interests include dynamical computation in spiking neural networks, neural encoding/decoding, (deep) neural reinforcement learning, neuroprosthetics, and neural control in robotics.



Title: Efficient Artificial Spiking Neural Networks

Artificial Neural Networks (ANNs) are bio-inspired models of neural computation that have proven highly effective. Still, ANNs lack a natural notion of time, and neural units in ANNs exchange analog values in a frame-based manner, a computationally and energetically inefficient form of communication. This contrasts sharply with biological neurons that communicate sparingly and efficiently using binary spikes. While artificial Spiking Neural Networks (SNNs) can be constructed by replacing the units of an ANN with spiking neurons, the performance of current SNNs does not match from that of deep ANNs on hard benchmarks and these SNNs use much higher firing rates compared to their biological counterparts, limiting their efficiency.  Here we show how spiking neurons that employ an efficient form of neural coding can be used to construct SNNs that match high-performance ANNs and exceed state-of-the-art in SNNs on important benchmarks, while requiring much lower average firing rates. For this, we use spike-time coding based on the firing rate limiting adaptation phenomenon observed in biological spiking neurons. This phenomenon can be captured in fast adapting spiking neuron models, for which we derive the effective transfer function. Neural units in ANNs trained with this transfer function can be substituted directly with adaptive spiking neurons, and the resulting Adaptive SNNs (AdSNNs) can carry out classification in deep neural networks using up to an order of magnitude fewer spikes compared to previous AdSNNs. Adaptive spike-time coding additionally allows for the dynamic control of neural coding precision: we show how a simple model of arousal in AdSNNsp further halves the average required firing rate and this notion naturally extends to other forms of attention. AdSNNs thus hold promise as a novel and efficient model for neural computation that naturally fits to temporally continuous and asynchronous applications.


Autonomous Systems

Mohsen Alirezaei

Mohsen Alirezaei  received his PhD in Mechanical Engineering / Robotics and Control and was a postdoc researcher at faculty of Mechanical Engineering of Delft University of Technology in 2012. Since 2015, he is technical lead and senior scientist at the Integrated Vehicle Safety Department of TNO mobility, and senior scientific staff at Delft University of Technology. His research interests are application of the nonlinear control in advance driver assistance system, vehicle dynamics and autonomous driving.



Title: Control of Autonomous Vehicle within Limits of Handling and beyond Stable Limits of Handling (Drifting)

Tires operated at or close to their friction limits show a highly nonlinear motion to force response. This state is called limit handling. The objective of this research is to minimize lateral path tracking error whilst the tires operate in limit handling. The State Dependent Riccati Equation (SDRE) technique is employed to develop a feedback-feedforward steering controller. It gives a systematic approach to take into account model nonlinearities such as combined slip tire characteristics. Furthermore the controller is implemented in a test vehicle and tested on a low friction skid pad. The controller shows robust path tracking performance even when the rear wheels are operated beyond their friction limits, and large body sideslip prevails.


Gijs Dubbelman

Gijs Dubbelman (male) PhD, is an assistant professor with the Eindhoven University of Technology and heading the Mobile Perception Research cluster (, which aims to improve the real-time perception capabilities of mobile sensor platforms through research on Semantic Simultaneous Localization and Mapping. Key focus areas are: 3-D computer vision, data fusion, and machine learning. Formerly, Gijs Dubbelman was a member of the internationally renowned Field Robotics Center of Carnegie Mellon's Robotics Institute, where he performed research on 3-D computer vision for autonomous robots and vehicles. He contributed the COP-SLAM algorithm for real-time embedded Visual-SLAM to the scientific open-source project



Title: Towards Fully-automated Driving: Challenges and Potential Solutions

The current state-of-the-art in highly automated driving is to incorporate all static objects, such as traffic lights, traffic signs, pedestrian crossings, sidewalks, etc. of the environment in Highly Automated Driving (HAD) maps. The automated vehicle accurately positions itself in these maps, using sensing technology, and thereby is able to navigate through the “static” environment. In addition, the vehicle uses real-time sensing technology to detect, classify, and track dynamic objects, such as other vehicles and vulnerable road users, like pedestrians and cyclists. On basis of this static and dynamic information, the vehicle performs continuous path planning, in order to prevent dangerous situations and collisions.

There are three key drawbacks of current sensing, mapping, and localization technologies that prevent global deployment of highly automated driving. Firstly, the creation of the required HAD maps is extremely complex and requires careful and time consuming human verification, making this socio-economically non-viable at global scales. Secondly, no effective technologies are available to regularly, automatically, and cost effectively update HAD maps, although road networks and other dedicated transport infrastructures are subject to daily changes, allowing life-threatening situations. Thirdly, the HAD map is only useful when the vehicle is able to accurately position itself in this map. This requires specific sensing technologies, which are currently not robust enough to be deployed safely in all types of environments and in all types of adverse weather conditions.

The challenge to unlock the societal benefits of Highly Automated Driving is therefore to: 1) improve the scalability and robustness of HAD mapping and localization technology, and 2) make the vehicle less dependent on these HAD maps by advancing Artificial Intelligence. In this presentation, possible solutions to solve these challenges are put forward and current results are presented.


Joined track by ASCI, IPA, SIKS

Alex Telea

Alexandru Telea received his PhD (2000) in Computer Science from the Eindhoven University of Technology, the Netherlands, in the area of visualization system design. Until 2007, he was assistant professor in visualization and computer graphics at the same university. Since 2007, he is professor of multiscale visual analytics at the Faculty of Science and Engineering, University of Groningen, the Netherlands. His interests include 3D multiscale shape processing, information visualization (with a focus on relational and multidimensional data), software visualization, and visual analytics. He has published over 200 papers in the above fields. He is the author of the textbook “Data Visualization - Principles and Practice” (CRC Press, 2014).



Software Engineering

Joost Bosman

Joost Bosman (Applied Mathematics & Informatics) is a senior IT manager and expert in software/knowledge engineering at ING. He is currently involved in the transformation of the bank towards a predictive enterprise examining the possibilities of formal methods and gaming with a view to improving quality, software production and maintenance.


Title: Towards Financial Eco Systems:  Science, Engineering, Economics“

Financial institutions are in the midst of a huge transformation from classical bank to a preferred brand in a (financial-) ECO system.

With 98,5% digital customer interaction, fintech’s entering the scene, increased regulation and globalisation we need to find our way into the digital future. Science and more precise Innovation, Research & Development will play a key role in our journey to a new and challenging future.


Lionel Briand
Lionel C. Briand is professor in software verification and validation at the SnT centre for Security, Reliability, and Trust, University of Luxembourg, where he is also the vice-director of the centre. He is currently running multiple collaborative research projects with companies in the automotive, satellite, financial, and legal domains. Lionel was elevated to the grade of IEEE Fellow in 2010 for his work on the software testing. He was granted the IEEE Computer Society Harlan Mills award and the IEEE Reliability Society engineer-of-the-year award for his work on model-based verification and testing, respectively in 2012 and 2013. He received an ERC Advanced grant in 2016 — on the topic of modelling and testing cyber-physical systems — which is the most prestigious individual research grant in the European Union. 



Title: Automated Vulnerability Testing Using Machine Learning and Search Metaheuristics

Modern enterprise systems can be composed of many web services (e.g., SOAP and RESTful) that interact to each other via HTTP messages. To properly protect such systems from potential attacks, developers have to set up multiple protection layers, such as code level defenses, web application firewalls (WAFs) and gateways. In this talk, I will explain how to combine various techniques, including machine learning and sophisticated search techniques, to effectively detect uncovered security vulnerabilities affecting the different protection layers. In particular, I will show how to instantiate and customize testing techniques for each protection layer (e.g., WAFs). Finally, I will discuss the performance of these techniques when used to uncover code injection vulnerabilities, which are the most frequent yet critical security threats according to the Open Web Application Security Project (OWASP).


Eimar Fandino

Eimar Fandino is a Software Engineer specialised in the Java eco-system, currently working on cloud-native micro-services. On his spare time he likes to play with his kids and enjoy soccer matches.


Otávio Fernandes

Otávio Fernandes is a software developer that lost his way out into Operations world, and nowadays trying to build bridges that we would call "DevOps". Passionate about open-source, cloud-native and data-streaming systems, spend a lot of his time building backend and infrastructure at Schuberg Philis.



Title: Streaming for Realtime Banking

Nowadays, modern banks rely on applications that react to events in near realtime, using streaming backends like Apache Kafka and a runtime based on cloud-native approach, powered by Kubernetes.

In this session, we look behind the scenes and show how many backend services are involved in providing an enjoyable customer experience on a mobile banking application. Also, we share more how streaming services are applied to fraud-detection and other security systems in near realtime. 


Systems Engineering

Imre Horváth

Imre Horváth (1954) received M.Sc. in mechanical engineering (1978) and in engineering education (1980. He earned dr.univ., Ph.D. and C.D.Sc. titles. Since 1997 he is a full professor at the Faculty of Industrial Design Engineering of the Delft University of Technology. He is heading the Cyber-physical System Design research group. He initiated the TMCE Symposia. He is emeritus editor-in-chief of the Journal CAD, and associate editor of Journal of Engineering Design. He co-authored more than 380 papers and articles. He served as chair for the Executive Committee of the CIE Division and is a fellow of the ASME. His research interest is in cognitive engineering of CPSs, systematic design research, and personalized/socialized system development.



Title: Shifting paradigm of cyber-physical systems

We are living in an accelerated age, in which rapid shifts of paradigms can be observed. This is also true for the paradigm of cyber-physical systems (CPSs). This notion was introduced hardly more than a decade ago, and since that the original definitions have been reformulated many times with various objectives and in different contexts. This talk casts light on the enablers of this rapid shift of paradigms. It elaborates on the original definitions and examples of CPSs, and proposes a definition, which puts a particular interpretation of cyber-physical computing into the center. According to this interpretation, CPSs can (i) deeply penetrate into real life physical, computational, social, cognitive, emotional, etc. processes, (ii) collect data and derive information run time, (iii) generate alternative operation strategies based on the acquired data, and (iv) operationalize the best matching strategy through functional and architectural adaptation. Having these capabilities, CPSs are able to perform smartly, that is, to build awareness of the working environments, the stakeholders and the internal operational states, and to adapt themselves to varying environmental or working conditions, or even to more beneficial performance targets. There has been a novel reasoning model introduced, which identifies four plus one generation of CPSs. The generations are differentiated based on the level of self-intelligence and the level of self-organization. The 0G-CPSs are look-alike engineered systems or partial implementations of CPSs. 1G-CPSs include (system of) systems with self-regulation and self-tuning capabilities, while the 2G-CPSs are capable to operationalize self-awareness, self-learning and self-adaptation towards smart behavior. The 3G-CPSs are supposed to be equipped with the capabilities of self-cognizance and self-evolution. Only the fourth generation of CPSs is supposed to achieve dominant self-consciousness and self-reproduction in adapted forms based on available system resources. There will be examples presented for these manifestations. In addition, the talk presents the prototype of a 2G-CPS node for detecting and enhancing short term engagement of stroke patients in rehabilitation exercises, and an information processing platform and mechanisms that is being developed to facilitate reasoning with dynamic spatial, attributive and temporal information in the case of a fire evacuation managing 2G-CPS. The talk proposes to concentrate more research on cognitive engineering of CPSs.

Wouter Leibbrandt – ESI

Wouter Leibbrandt is Science and Operational Director of ESI, an industry and academia sponsored research center hosted by TNO. ESI focusses on the development of new methods and techniques for design and engineering of increasingly complex high-tech (embedded) systems. It does so in strong partnership and close collaboration with leading high-tech companies such as ASML, Philips, Thales, NXP, Océ, Thermo-Fisher and DAF as well as with leading academic groups in the Netherlands and across Europe.

Until early 2016 Wouter was with NXP Semiconductors for 10 years, where he managed the Advanced Applications Lab, investigating new application concepts around future advanced silicon products, driving secure connections for a smarter world. The recurring theme here is that everything is getting connected with everything (IoT).

Before joining NXP, he was with Philips Research labs for 14 years, managing a variety of projects and departments. From 2004 to 2006 he lived and worked in China, founding and managing part of the Philips Research labs in Shanghai.

Wouter holds a PhD in physics from Utrecht University.