Impact Breakfast

JANUARY 29TH 2020:
Impact Breakfast - Artificial Intelligence

 

Program:

08:00-08:30: Registration and breakfast

 

08:30-08:35:
Opening,
by Tobias Dahl, SINTEF /UiO

 

08:35-08:50:
Kyrre Glette, UiO,
Evolutionary algorithms for intelligent robots

08:50-09:05:
Lilja Øvrelid, UiO,
Opinion mining from text

 

09:05-09:20:
Signe Riemer-Sørensen, Sintef,
Hybrid analytics - combining physics and machine learning

 

09:20-09:35:
Arnoldo Frigessi, UiO,
Forecasting viral sales in networks

 

10:00: Finish

 

 

MORE ABOUT THE SCIENTISTS AND THE TALKS:

 

Kyrre Glette, UiO

Kyrre Glette
Assosiate Professor, RITMO Centre for Interdisciplinary Studies in Rhythm, Time and Motion, Department of Informatics, UiO
Evolutionary algorithms for intelligent robots
Future robots need to be robust and adaptable, and new design approaches are needed for new production methods. I will talk about my research in using evolutionary algorithms and biologically inspired methods with the aim of having more intelligent, robust and adaptive behavior in robots. I will give a short introduction to some of the algorithms and show how we apply them in our robotic platforms for exploring automatic design and adaptation.

 

Lilja Øvrelid, UiO

Lilja Øvrelid
UiO Department of Informatics, The Faculty of Mathematics and Natural Sciences Associate Professor - Research Group for Language Technology


Opinion mining from text
Natural Language Processing (NLP) is a sub-field of AI concerned with enabling machines to `make sense' of human language. An increasingly central application of NLP is so-called opinion mining or sentiment analysis. This is the task of automatically identifying the opinions, attitudes or emotions that are expressed by subjective information in text. In the SANT project we develop resources for fine-grained sentiment analysis for Norwegian, through the use of neural models / Deep Learning.

 

 

Signe Riemer-Sørensen

Signe Riemer-Sørensen
SINTEF Digital, Mathematics and Cybernetics


Hybrid analytics - combining physics and machine learning
Machine learning algorithms are flexible and powerful, but the data requirements are high and rarely met by the available data. Real world data is often medium sized (relative to problem side), noisy and full of missing values. Additionally, in order to deploy machine learning in industrial settings, the models should be robust and explainable. Both requirements can in some cases be fulfilled by combining traditional physical models with machine learning in a hybrid analytic approach. I will sketch some methods and examples of applications from the real world.

 

 

Arnoldo Frigessi

Arnoldo Frigessi
UiO Institute of Basic Medical Sciences, Faculty of Medicine, Professor - Department of Biostatistics

 

Forecasting viral sales on networks
The behaviour of customers or users of a service is influenced not only by marketing campaigns, but crucially also by the behaviour of other individuals, with whom they are in social relations. The way a behaviour (purchase, adoption, churn) spreads on a social network is similar to the spread of an infectious disease in a community. Inspired by this similarity, we developed a new method, which detects and quantifies the viral strength of a product or service. Our method relies on machine learning of a probabilistic model. Like a ‘crystal ball’, given a relatively small slice of behaviour history, we are able to predict the future behaviour of a connected group of individuals over a long future period.

 

 

Invitation only

The meetings will be held in Forskningsparken – Oslo Science Park and are by invitation only. Please request for an invite here – Impact Network, or contact astrid@forskningsparken.no