DSSC Newsletter

For full display please view this email in your browser.

UPDATES
Winners of the DSSC call for proposals summarize their projects
DSSC research profiles
DSSC Master specializations
DSSC addresses the NWO "Complexity" programme   
GCSCS expression of interest for the NWO theme "Improving resilience of society: understanding, modelling and managing social/economic complexity"
 
ACHIEVEMENTS & NEWS
~2million euro NWO-G grant for the data-intensive Euclid space mission
Image recognition software for the Spanish police
A real time web viewer for bike rentals
Geoservice (RUG) receives NWO subsidy for web application in archeology project

CALLS FOR APPLICATIONS
Horizon 2020
eScience Centre
NWO

UPCOMING EVENTS

• ICT.Open (March 22-23, Amersfoort)
• Dagstuhl Seminar: Multidisciplinary Approaches to Multivalued Data: Modeling, Visualization, Analysis (April 3-8, Wadern)
• DSSC thematic seminar (May 3): Prof. Simon Portegies Zwart & Dr. Michael Wilkinson
• Weaving the Internet of Data (April 6, Amsterdam)
• European HPC summit (May 9-13, Prague)
• EuroVis (June 6-10, Groningen)
• Digital Humanities Benelux (June 9-10, Belval)
• Dagstuhl Seminar: Integration of Expert Knowledge for Interpretable Models in Biomedical Data Analysis (June 26-July 1, Wadern)
• European Data Forum (June 29-30, Eindhoven)
• Conference on Complex Systems (September 19-22, Amsterdam)
• International Conference on Computing in High Energy and Nuclear Physics (October 10-14, San Francisco)
• IAU Symposium Astroinformatics (October 20-24, Sorrento)

UPDATES
Winners of the DSSC call for proposals summarize their projects
The 4 research teams that received DSSC funding are set to begin recruitment for 4 PhD positions in data and complexity science. In this issue the winning teams briefly describe their projects:

• COmplex DYnamical networks: From data to ConnEctivity Structure

Researchers: Dr. Pietro Tesi (Engineering and Technology Institute Groningen - ENTEG); Prof. Dr. Kanat Camlibel (Johan Bernoulli Institute - JBI)

Dynamical networks are pervasive in today’s world, ranging from social and economic networks to biological systems and man-made infrastructures. The connection structure plays a crucial role in determining the overall behavior of these networks. For instance, the topology of social networks affects the spread of information and disease, and the topology of the power grid affects the robustness and stability of power transmission. Knowing the connection structure is fundamental in order to predict how these networks might evolve and to anticipate/counteract critical transitions. Unfortunately, the connection structure is often hard to observe directly. Moreover, observing/measuring the behavior of each network component is infeasible due to the complex and large-scale nature of such systems.

In this research project, we aim at developing methods and algorithms to infer the connectivity structure of complex dynamical networks from sparsely collected data, i.e., data collected from a fraction of network components. The project is expected to take major first steps toward understanding some of the core issues in this research area:

i) what are the conditions under which identification is conceptually feasible;

ii) how many measurements are needed and at which locations in the network they should be taken; and

iii) which topologies and dynamics can render identification easier/harder. Answering these questions may further our understanding of the behavior of complex dynamical networks, and, in the long run, lead to the development of numerically efficient identification algorithms.

Applications will be considered in the area of energy systems, in particular power networks monitoring for detection of transmission line outages.

• [VF]ast Data

Researchers: Dr. Michael Wilkinson (JBI); Prof. Dr. Alexander Lazovik (JBI); Dr. Johan Messchendorp (Centre for Advanced Radiation Technology - KVI); Prof. Dr. Leon Koopmans (Kapteyn Astronomical Institute - Kapteyn); Prof. Dr. Scott Trager (Kapteyn)

In many fields of science, data sets are not just big, the data rate is also huge, and the sizes of individual data items or cliques of data that need to processed as a whole become very large indeed. Coping with big data sets is quite a challenge on its own, but the problem is compounded if the individual data items are too large to process on a single node, or if the sensor data rate is so high you cannot possibly store all of it. Our proposal aims to develop fast, multi-scale algorithms for graph-based data processing suitable for efficient distributed-memory parallel processing.

Applications include the vast image data cubes from the Square Kilometer Array (SKA) radio telescope and the fast data streams from the PANDA (Anti-Proton Annihilation) Collaboration.











In this project we will develop new approaches to deal with both vast and fast data problems, focusing on three themes:

Theme 1: Massive Multi-Scale Analysis
Theme 2: Keeping up with the Data Rates
Theme 3: Faint Signals or Needles in a Haystack.

• Clinical Big Data for multifactorial diseases: from molecular profiles to precision medicine

Multifactorial diseases with complex traits such as various types of cancer and chronic obstructive pulmonary disease (COPD) are among the leading causes of death in the Western society and form primary challenges of current health system. Poor understanding of the complex molecular mechanisms of such diseases and currently available diagnostic options are often inadequate to find efficient treatment for a large proportion of patients. Personalized treatment of patients or identification of subgroups of patients where treatment is efficient using precision medicine approaches are pivotal to improve health and patient care.

The goal of the project is to develop a Machine Learning approach based on Bayesian networks for large multi-omics and clinical data sets where parameters vary due to individual sampling in different patients. Bayesian networks are a very powerful and flexible Machine Learning tools to infer the complex regulatory relationships between the molecular profiles and the clinical or experimental metadata. The computational expensive Bayesian network inference is efficient when coupled with Markov Chain Monte Carlo algorithms, which systematically search the space of network structures for those that are most supported by the data. From those complex networks the key components can be identified and then be used as biomarkers to identify patient sub-phenotypes, perform accurate disease staging or new drug targets. Alternatively, the molecular profiles can be employed to predict the outcomes of treatment.

The project will provide a generic Bayesian network analysis tools that can be parametrized to address different clinical questions in projects studying complex diseases such as cancer and COPD. The outlined project will contribute to the understanding of the molecular mechanism of COPD, ovarian, cervical cancer and head and neck cancers and will be embedded in highly interdisciplinary team combining cancer and COPD clinical research, analytical chemistry, bioinformatics and statistics.



Figure 1. Project outline and collaboration network within RUG (FWN, ERIBA, UMCG). Molecular profiles are collected in the laboratories of Analytical Biochemistry and ERIBA, while patient materials with clinical metadata are obtained from groups in UMCG and through collaborations. Drug usage statistics and potential side effects information is obtained from a population based drug prescription database. The DSSC project will build molecular network and identify biomarkers, new drug target candidates or perform prediction of treatment efficacy using Bayesian network and Machine Learning approaches (text in red).

• Uncovering the information processing underlying the interactions between brain area

Researchers: Marieke van Vugt (Artificial Intelligence and Cognitive Engineering - ALICE); Ming Cao (ENTEG); Niels Taatgen (ALICE); Jelmer Borst (ALICE); Mircea Lungu (JBI)

While considerable progress has been made on understanding how the brain works, most of this is focused on the functions of individual brain areas in isolation. The next frontier is to understand how these brain areas work together in the service of cognition. The main aim of the project is therefore to understand how brain areas communicate in the service of information processing.

To answer this question, we will combine cognitive models of cognition with a dynamical systems analysis of EEG. We have previously shown that cognitive models can be used to help interpret patterns of interaction between brain oscillations in a specific attention task. This project seeks to extend this approach to other tasks, and to link more generally the stable states and state transitions manifesting in EEG to information processing steps predicted by the model. Eventually we hope to develop a general-purpose “brain-activity parser” that can interpret what the interactions between different brain areas may mean.


DSSC research profiles
The DSSC reunites a group of innovative researchers whose interests will be featured in every issue of the DSSC Newsletter. 2 researchers at Kapteyn and JBI introduce their projects.

• Dr. Geijs Verdoes Kleijn (Kapteyn)
• Dr. Konstantinos Efstathiou (JBI)

Dr. Geijs Verdoes Kleijn (Kapteyn)

My astronomical interest is to understand the symbiotic relationship between galaxies and the supermassive black holes at their centers. Already in the very young Universe the growth of stellar mass and the growth of black holes appears closely linked, leading to a tight relationship between black hole mass and dynamical galaxy mass in today's galaxies. At first sight this appears surprising as the black hole mass and its gravitational sphere of influence are puny compared to its host. It is unknown how supermassive black holes form. It is unknown what physical mechanism maintains the balance between growth in stellar mass and growth of the supermassive black hole. A physical understanding of this symbiotic relationship might advance significantly our understanding of both galaxy evolution and also supermassive black hole formation.

Besides this I am involved in various projects exploiting astronomical survey data. These include galaxy cluster studies, searches for gravitational lenses and searches for near-earth objects and other small solar system bodies.

Being a survey astronomer, I have a strong interest in the technical challenge to disentangle the astrophysical from instrumental information in large sets of survey data. I am involved in the following data-intensive missions / instruments:

• Euclid Space Mission, covering 15000 square degrees of extragalactic sky from space. To be launched by the European Space Agency in 2020. Primary science driver is understanding the accelerated expansion of the Universe. I am co-lead of the Organizational Unit External Data, responsible for Euclidizing all external data to be combined with the Euclid mission data.
• Kilo-Degree Survey, imaging 1500 square degrees with exquisite image quality from the European Southern Observatory (ESO) at Paranal in Chile. In operations since October 2011. Primary science driver is dark matter tomography. I am project manager, leading the survey data production team.
• MICADO, the First-Light near-infrared imager and spectrograph for ESO's European Extremely Large Telescope. I am the dataflow lead for the consortium.

The instrumentalist in me says: “Great to have all this data: it will tell me everything about the physical state of my Observatory”. The astronomer in me says: “Great to have all this data: it will tell me everything about the physics of my Universe!” Our team of astronomers and experts in Astronomical Information Technology aims to meet this challenge by building information systems with extreme “backward chaining”. They link all information contained in final products back to the “raw” initial data. Own-developed technology (“WISE technology”) aims at allowing any level in complexity of the route when traversing the network of linked information through arbitrarily large sets of pixel data and associated meta-data. This effort is being pursued within OmegaCEN, the expertise and datacenter for wide-field imaging, for which I am the deputy lead.

Having advanced astro-informatics tools (e.g., neural networks, machine learning techniques, Bayesian statistical optimization, data-driven visualization) take advantage of such information systems appears to me the way forward to achieve an optimal disentangling of astrophysical and instrumental information from large data. I use the word astro-data-scientist for persons with the expertise to bridge the domains of astro-informatics and astronomical information technology.

Dr. Konstantinos Efstathiou (JBI)

Dynamical systems is a research topic with many connections, both to other, more pure, mathematical fields and to applications. My research is related to dynamical systems in the more general sense. Something that I particularly enjoy in research is to combine very different approaches: purely theoretical approaches on the one hand but also numerical simulations and visualizations on the other. Furthermore, I like to understand a subject from different points of view, often returning back to the same problem to examine it from a different perspective.

One large part of my research revolves around Hamiltonian systems and, specifically, the geometry and bifurcations of integrable Hamiltonian systems which are in a sense some of the least complex systems one could consider. In this direction I use techniques from algebraic topology, differential geometry, and group theory to understand the geometry defined by the conserved quantities in such systems. Variations of such techniques, for example, persistent homology, have also been developed for qualitatively understanding the geometry of big data.

A second direction is the study of the dynamics on networks which are some of the most complex dynamical systems one could consider. The study of networks is a very popular research field that poses interesting mathematical problems and at the same time has many applications, for example, in biology, synchronization, and power distribution networks.

In this direction I have been mostly working on pulse-coupled oscillator networks with delay. These are networks where individual nodes are modeled by periodic oscillators which interact by sending pulses. When, after some delay, other oscillators receive the pulse they discontinuously jump to a new state. This model offers a simplified description of the interactions between neurons. The delay and the discontinuity pose complicate the mathematical analysis of such systems. Here I am interested in describing the dynamics of such systems, proving the existence of synchronized states, and studying the bifurcations between different dynamical regimes. A variation to this theme has been my work on circadian rhythms, where through appropriate assumptions, one can reduce the study of synchronization to the daily light-dark cycle to the study of a circle map. Such maps are well understood in dynamical systems and therefore help us understand how organisms entrain to a 24-hour period.

In the following years, and within the DSSC framework, I plan to further focus on the study of dynamics on networks. I will be looking forward to work with the other DSSC pioneers in setting up joint research projects in this direction and pushing forward our understanding of these complex systems and their applications.

DSSC Master specializations
The Master specialization "Data Science in Astronomy" will be set to start within the Programme at Kapteyn in fall 2016. The course offer in "Data Science in Astronomy" will partially overlap with the course selection in the Master track "Data Science and Systems Complexity" within the Programme Computing Science at the JBI that will also begin in fall 2016.

DSSC addresses the NWO "Complexity" programme

In the upcoming years, the NWO and TNO will launch several calls for proposals within the Complexity programme. Having selected Logistics and Agrifood to be two of the underlying themes for future calls, the NWO invited expressions of interest for other potential themes. The DSSC consulted its researchers and submitted an expression of interest that emphasizes main lines of Complexity research within the Centre:

• High Tech Complexity: Programmable Self-Organisation
• Dealing with Unpredictability
• Imaging and/or Modeling of Neurodegenerative Processes
• Complexity in Neural Computation

Geoservice received a 200.000 euro subsidy from the NWO to develop an interactive web application in the project Portable Antiquities of the Netherlands (PAN) developped by Prof. Nico Roymans at the Vrije Universiteit. The interface feeds the numerous archeological finds from the Netherlands into the system, making them accessible to the scientists and the public. The NWO contribution to the PAN project amounts to 1.862.000 euro.

       
CALLS FOR APPLICATIONS
Horizon 2020
For more information, please click on the calls:

LCE-02-2016: Demonstration of smart grid, storage and system integration technologies with increasing share of renewables: distribution system
Deadline: 5 April

ICT-01-2016: Smart Cyber-Physical Systems
Deadline: 12 April

ICT-03-2016: SSI - Smart System Integration
Deadline: 12 April

ICT-10-2016: Software Technologies
Deadline: 12 April

ICT-14-2016-2017: Big Data PPP: cross-sectorial and cross-lingual data integration and experimentation
Deadline: 12 April

ICT-15-2016-2017: Big Data PPP: Large Scale Pilot actions in sectors best benefitting from data-driven innovation
Deadline: 12 April

ICT-18-2016: Big data PPP: privacy-preserving big data technologies
Deadline: 12 April

FETPROACT-01-2016: emerging themes and communities
Deadline: 12 April

FET OPEN-2016-2017: Novel ideas for radically new technologies
Deadline: 11 May

eScience Centre
Accelerating Scientific Discovery
The purpose of this Call for Proposals is to enable domain scientists, working in application fields of Environment & Sustainability, Humanities & Social Sciences, Life Sciences & eHealth, or Physics & Beyond, to address compute-intensive and/or data-driven problems within their research.

Successful proposals will pursue new scientific domain challenges and enhance and accelerate the process of scientific discovery within the field of application. To achieve this, the proposals will be characterized by the usage, adaptation, and integration of eScience methods and tools including software for optimized data handling, Big Data analytics, and efficient computing.
Deadline for pre-proposals: 17 May
Deadline for proposals: 30 August 
For more details, please visit the website.

Joint eScience and Data Science across Top Sectors
This call for proposals aims at research and development of disruptive solutions associated with big data handling, big data analytics and related computational methods in order to be able to address novel cross-sectoral scientific challenges described in the COMMIT2DATA White Paper.

A successful project should address (a) domain challenge(s) from one or more sectors using novel digital technologies and data science methods that are studied and developed within the project.
Deadline for pre-proposals: 17 May
Deadline for proposals: 30 August 
For more details, please visit the website.

Path-finding project call

Path-Finding Projects are intended to provide the opportunity to address novel data or computationally intensive scientific challenges. In this way they can serve as a pilot for future research projects or investigate the potential to initiate full projects. Path-Finding Projects are supported to the value of € 50K, primarily via the 100% in kind provision of  eScience Research Engineers employed by NL. The proposals should incorporate at least one of the key technological competencies defined by the eScience Centre

• Optimized Data Handling: incl. combining structured and unstructured data, real-time data analysis, database optimization, data-interoperability;
• Big Data Analytics: incl. data interpretation, data exploration, analysis, data mining, machine learning, text analysis, and visualization;
• Efficient Computing:incl. high-performance and distributed computing, heterogeneous computing, efficient algorithm design, accelerated hardware (e.g. Graphics Processing Units).

Deadline: a proposal can be submitted at any time, but will be reviewed only after the earliest fixed deadline in the year: June 1st or December 1st. For more information, please visit the website.

NWO
Data-driven science - computational and data science - for smart and sustainable energy research
Research teams composed of Dutch and Indian researchers can apply for joint bilateral projects, jointly funded by DST and NWO.
Joint projects should:

• contain highly innovative research with strong components in the fields of computational science and/or data science;
• have an interdisciplinary character; it will be desirable that the projects have elements from both computational science and data science;
• contain high quality scientific research that must be executed by researchers and institutes with excellent track records;
• be a cooperation between excellent institutes in both countries (India and the Netherlands); have an application in, or contribute to, smart and sustainable energy research

UPCOMING EVENTS