Current Trends in Artificial Intelligence
2nd DTU Compute Workshop

On 17 November 2017, DTU Compute is organising the second of two workshops on Current Trends in AI. The workshop will take place at the

See the Venue section below for details on how to get there.

Motivation

Specialised artificial intelligence has been making fast progress in recent years. This progress is based upon fast computation and development in machine learning and traditional artificial intelligence algorithms - and, crucially, in making new combinations of these existing methods. A fine example of this is Google Deepmind’s AlphaGo system that recently beat the best human player in the board game Go. Go is challenging because of its large branching factor and difficulty of evaluating the value of a position. AlphaGo uses a combination of supervised and reinforcement learning and Monte Carlo Tree Search (MCTS) to solve the problem.

Purpose

At DTU Compute there is quite a few researchers working on different aspects of artificial intelligence from machine learning, algorithms, logic-based and scientific computing perspectives. DTU Compute is organising two workshops with the purpose of bringing together researchers, industry and academia interested in these areas of AI. The first workshop focused on the Copenhagen area and took place on 23 November 2016. See the home page of the first workshop for details.

A central theme in both workshops is combinations of, and synergies between, machine learning (e.g. neural networks) and symbolic AI (e.g. state-space searches, automated planning and logical reasoning). The workshops focus on recent trends, aiming at finding common ground and interests that point forward to the coming years. It is expected that many of the future successes in AI will lie in the combination of methods from separate subfields, e.g. combinations of symbolic AI and machine learning.

The outcome of the workshops will ideally be to identify where the field as a whole is heading and to pinpoint how we can use the competences at DTU Compute and in the Copenhagen area to exploit the potential synergies and stay at the front in research, teaching and industrial collaboration.

Format

The format will be invited speakers that will be asked both to talk about their own research and their views on where the field as a whole is heading and how their own research area fits into the overall picture. In particular, we are interested in the speakers’ thoughts and reflections on combining methods from distinct paradigms of AI, in particular combinations of symbolic, sub-symbolic and statistical methods.

Schedule

Registration

Registration is simple: Just send an email to Thomas Bolander (tobo@dtu.dk) with your name and affiliation, using the subject header Registration for Trends 2 at latest on Friday 10 November, 2017. Everyone is welcome and registration is free. However, the venue has limited seating, and if too many people register, we reserve the right to reject registrations. People will be registered on a first come, first served basis.

Venue

The workshop takes place in the Christiania district of Copenhagen, right in the heart of the city:

To get to Christiania:

• Metro from Nørreport Station, stop Christianshavn.
• Buses from Copenhagen central station, e.g., 9A.

Organisers

The main organisers of the event are Thomas Bolander and Ole Winther, both from DTU Compute. Assisting organisers are Andrea Dittadi (DTU Compute), Andrés Liberman (DTU Compute) and Alireza Kashani.

Contact

E-mail enquiries should be directed to Thomas Bolander, tobo@dtu.dk.

Sponsors

The event is generously sponsored by: