CWI Scientific Meeting - 24 November 2017

Dear colleagues,

This is the second announcement for the upcoming Scientific Meeting, which will take place this Friday, November 24th, in the Turing room (13:00). Four postdocs from across CWI will present their research:

13-13:15 Tim Baarslag (IAS) - Coordination of Intelligent and Autonomous Systems Through Negotiation
13:15-13:30 Joost Bosman (ST) - A domain specific language for performance modeling
13:30-13:45 Thomas Degueule (SWAT) - Modular Extension of Domain-specific Languages
13:45-14 Alexander Kostenko (CI) - See-Through Museum Project: computational methods in the service of cultural heritage

See below for the abstracts.

Beforehand, there will be lunch. We hope to see you all there!

Best wishes,

Tijs & Peter

-----------------

Tim Baarslag (IAS) - Coordination of Intelligent and Autonomous Systems Through Negotiation

Complex and social systems consist of many interacting nodes that exchange information and can compete and/or collaborate with each other.
For these nodes to coordinate effectively, they need to engage in joint decision making through the process of /negotiation/.

In this talk, I will present my ongoing research on intelligent and autonomous systems that can negotiate in a wide variety of upcoming
domains, such as smart energy trading, the Internet of Things, and digital privacy. One of the key challenges in designing a successful
automated negotiator is that only limited information is available about the other parties. Therefore, we need to investigate various decision
and learning techniques to determine what questions to ask, what offers to make, and when to accept. I will provide some intuition behind some of the solution concepts (e.g. Pandora's Rule, optimal stopping), which have a wide range of potential applications. I will also illustrate the
current challenges in automated negotiation, together with a roadmap for the successful design of intelligent negotiation systems.


Joost Bosman (ST) - A domain specific language for performance modeling

Abstract:
In the business domain (quantitative) performance modeling is used to explain and predict behavior of systems and to identify and solve performance issues. In banking for example it is crucial that electronic payments can be processed in a timely fashion.
There is a variety of tools and software available for performance modeling and analysis. However, these tools are often limited by extensive user interfaces (drag and click) and/or do not provide a sufficient set of queuing and scheduling mechanisms. Another option is to implement a model yourself using a framework. This approach is likely to be time consuming.
Most tools and frameworks are based on discrete event simulation (DES). This aproach focuses on discrete change events in a system while mostly performance modeling is about interactions of processes.
To this end we are developing a domain specific language that includes an extensible set of queueing, scheduling, and coordination primitives. The goal is to represent and model domain specific processes and their interactions in a clear and more aggregated way.


13:30-13:45 Thomas Degueule (SWAT) - Modular Extension of Domain-specific Languages

A domain-specific language (DSL) is a software language focused on a particular application domain. DSLs are everywhere: from avionics to
banking systems, from blockchains to the Internet of Things. By their very nature, they must evolve to keep pace with their ever-changing
application domains.

DSLs are defined by their syntax (the domain concepts, properties, and relations) and their execution semantics (how computers interpret them). It follows that DSLs must be extended along two axes: syntax (introduction of new concepts) and semantics (definition of new  interpretations). The two-decades-old 'Expression Problem' states that it is inherently difficult to extend the structure and behavior of a
datatype in a modular fashion. Similarly, it is difficult to extend the syntax and semantics of a DSL modularly.

In this talk, I will present a new language implementation pattern, named 'Revisitor', that enables independent and modular extensibility of
the syntax and semantics of DSLs. I will introduce some motivating examples, the general intuition underlying the pattern, some  applications, and the future research directions it opens.


Alexander Kostenko (CI) - See-Through Museum Project: computational methods in the service of cultural heritage

Until now, the best way to preserve a record of a past event, whether it was an invention of a new form of fine arts 200 years ago or an extinction of some species 200 million years ago, was to keep a physical evidence of the event. So the modern museums such as Rijksmuseum and Naturalis played an important role of memory banks of our own cultural history and the history of other species on Earth.

Obviously, storing physical records of the past events has drawbacks: need for space, various efforts needed to stop the decay of physical objects, difficulties with displaying and transportation.
So how can we make sure that the heritage and natural history collections become fully digital and virtual in future?

In this talk I will give some examples of typical research questions that are considered in the world of cultural heritage and show how mathematical research and computer hardware developments can lead to improvements in the key 3D imaging methods: X-ray and Neutron Computer Tomography (CT).