DISA

Centre for Data Intensive Sciences and Applications

Meet Speaker Anna-Lena Axelsson from The Forest Data Lab

2020-11-27

During the Big Data Conference on December 3-4, 2020 we will have several interesting invited speakers, one of them is Anna-Lena Axelsson from The Forest Data Lab .

Anna-Lena Axelsson will talk about The National Forest Data Lab, an open platform that promotes co-creation and data-driven innovation within the forest sector. The platform builds upon existing data, infrastructure and collaboration between two strong players within management, analysis and curation of forest related data; the Swedish Forest Agency and the Swedish University of Agricultural Sciences (SLU). The main users and collaborators are companies and public authorities but also academia and research institutes. The presentation will focus on a number of use cases that demonstrate the value of open data and services for data-driven innovation in the forest sector. The Lab also arrange seminars, networking and training events. Currently The Forest Data Lab participate in the new version of Hack for Sweden 365, which is an innovation competition related to public open data.

Anna-Lena Axelsson. Foto: Linnea Lutto

Foto: Linnea Lutto

Anna-Lena Axelsson works with development of external collaboration and coordinates the forest environmental monitoring and assessment program at the Swedish University of Agricultural Sciences. She is one of the initiators of the National Forest Data Lab.

Don’t miss out on the opportunity to listen to him and take part of the conference by signing up here by December 1st.

Meet Speaker – Johan Thor from Södra at the Big Data Conference

2020-11-26

During the Big Data Conference on December 3-4, 2020 we will have several interesting invited speakers, one of them is Johan Thor, Head of data science at Södra.

Johan Thor will talk about how the small European bark beetle, less than half a centimeter long, gets visible from space via satellite images, or rather; its effects. The bark beetle infests and kills large amounts of spruces, not only in Sweden, to large economic amounts. Today, we have limited possible actions to take in order to prevent further infestation and he will describe how Södra teamed up with a Dutch startup in order to explore a new way to fight back!

Johan Thor, Head of data science at Södra

More information about Johan Thor is currently working as Head of data science at Södra. He has been for Södra for over 14 years where he have had several different focuses and positions. He has a master of science in applied physics from The Institute of Technology at Linköping University. At Södra Johan is involved in collaborations with academia, consultants and other partners and is a great inspiration for the students he meets.

Don’t miss out on the opportunity to listen to him and take part of the conference by signing up here by December 1st.

Meet Keynote: Erik Willén – Big Data applications in Forestry

2020-11-20

During the Big Data Conference on December 3-4, 2020 we will have several interesting Keynote speakers, one of them is Erik Willén, who is Process Manager at Skogforsk, in Uppsala, Sweden

During the conference he will speak about how Swedish forestry is using and producing vast amount of data for planning, during operations and transportation to the industry. The presentation focus on the enablers for digitalization in Swedish forestry and their current status. The most important data collection and processing as well as several applications in operational use and in applied R&D will be presented.

Erik Willén, SKogforsk

Don’t miss out on the opportunity to listen to him and take part of the conference by signing up here by December 1st.

Meet Keynote: Håkan Olsson – Remote Sensing provides Big Data for assessment of our forest resources

2020-11-19

During the Big Data Conference on December 3-4, 2020 we will have several interesting Keynote speakers, one of them is Håkan Olsson is professor in forest remote sensing at the Swedish University of Agricultural sciences. He leads the data acquisition work package in the research programme Mistra Digital Forest. He is also a member in the steering group for the ongoing national laser scanning for forest resource assessment, and a member of the national council for geodata (Geodatarådet).

Håkan Olsson

During his talk he focus on that there is an increasing stream of remote sensing data that together with digital techniques can be used for assessment of forest resources. Satellites provides frequent images that can be compared and used for forest damage assessment. Airborne laser scanners provide 3D point clouds that in combination with field surveyed reference data are used operationally for producing nationwide and accurate maps with data about the forest on raster cell level. The sensors develops rapidly and provides data with higher resolution, making assessments of single trees realistic. Among the current research frontiers are: combining single tree data from airborne sensors with stem shape data from sensors carried by man or vehicles; automated classification of tree species; assessment of forest growth from repeated sensor data acquisitions. The ultimate goal is to assimilate all new data into a continuously updated model of the forest resources.

Don’t miss out on the opportunity to listen to him and take part of the conference by signing up here by December 1st.

Meet Keynote Virginia Dignum: Responsible AI: from principles to action

2020-11-09

During the Big Data Conference on December 3-4, 2020 we will have several interesting Keynote speakers, one of them is Virginia Dignum, who is a Professor of Responsible Artificial Intelligence at Umeå University, Sweden and the director of WASP-HS.

She will talk about that every day we see news about advances and the societal impact of AI. AI is changing the way we work, live and solve challenges but concerns about fairness, transparency or privacy are also growing. Ensuring an ethically aligned purpose is more than designing systems whose result can be trusted. It is about the way we design them, why we design them, and who is involved in designing them. If we are to produce responsible trustworthy AI, we need to work towards technical and socio-legal initiatives and solutions which provide concretise instructions, tools, and other means of dictating, helping, and educating AI practitioners at aligning their systems with our societies’ principles and values.

Don’t miss out on the opportunity to listen to her and take part of the conference by signing up here by November 25th.

Virginia Dignum (photo: Mattias Pettersson(photo: Mattias Pettersson)

More information about Virginia Dignum, Professor of Responsible Artificial Intelligence at Umeå University, Sweden and associated with the TU Delft in the Netherlands. She is the director of WASP-HS, the Wallenberg Program on Humanities and Society for AI, Autonomous Systems and Software. She is a Fellow of the European Artificial Intelligence Association (EURAI), a member of the European Commission High Level Expert Group on Artificial Intelligence, of the working group on Responsible AI of the Global Partnership on AI (GPAI), of the World Economic Forum’s Global Artificial Intelligence Council, of the Executive Committee of the IEEE Initiative on Ethically Alligned Design, and a founding member of ALLAI-NL, the Dutch AI Alliance. Her book “Responsible Artificial Intelligence: developing and using AI in a responsible way” was published by Springer-Nature in 2019.

Webinar DISA-DSM November 10th

2020-11-06

The DISA-DSM group welcomes you to a webinar with Yaozhong Hu   (University Alberta, Edmonton, Canada) 10 November at 13:00 Stockholm time

Title: Numerical methods for stochastic Volterra integral equations with weakly singular kernels

Abstract: In this talk, we  will introduce  stochastic Volterra integral equations with weakly singular kernels and study the existence, uniqueness and Holder continuity of the solution. Then, we propose a  theta-Euler-Maruyama  scheme  and a Milstein scheme to solve  the equations numerically  and  we obtain the strong  rates of convergence for both schemes in L^{p} norm for any p\geq 1. For the theta-Euler-Maruyama  scheme the rate is min {1-alpha, 1\2-beta} and for the Milstein scheme the rate is min{1-alpha,1-2beta} when alpha\neq \frac 12, where 0<alpha<1, 0< beta<1\2. These  results on the rates of convergence are significantly different from that of the similar schemes for the stochastic Volterra integral equations with regular  kernels. The difficulty to obtain our results is the lack of Ito formula for the equations. To get around of this difficulty we use instead the Taylor formula and then carry a sophisticated  analysis on the equation the solution satisfies.

For more information and registration contact Nacira Agram – nacira.agram@lnu.se

Upcoming seminars:

  • 17 November  at 13:00 – Paolo Di Tella (University of Technology Dresden, Germany) Title: On enlarged filtrations of point processes
  • 24 November  at 12:00 – David SISKA (University of Edinburg, UK)
  • 24 November  at 13:00 WEINAN E  (Princeton University, USA) Title: An Overview of Deep Learning Based Algorithms for high dimensional PDEs Abstract: I will give an overview of deep learning-based algorithms for PDEs. Topics to be covered includes: (1) The Deep BSDE method. (2) Applications to control theory. (3) Theoretical advances.
  • 30 November at 15:00 Jiequn Han (Princeton University, USA)
  •  8 December at 13:00 Stockholm time Mathieu Lauriere (Princeton University, USA)

Keynote: Anders Arpteg: How will the recent AI-revolution change our society?

2020-11-04

During the Big Data Conference on December 3-4, 2020 we will have a several interesting Keynote speakers, one of them is Anders Arpteg, who is the Head of research at Peltarion

He will talk about that something extraordinary has happened within AI in recent years. Companies are starting to talk about moving into an AI-first future, but what does that mean? Not only are we seeing significant scientific advances in AI, but we are also seeing companies and politicians starting to invest heavily in AI. In order to stay one step ahead, we must be prepared for what is coming next. What has really happened in recent years, and what are the next steps and trends in machine learning? What should companies know to be prepared for the rapid development that is happening with ML and AI? This talk will give a glimpse into the future of AI, what possibilities it holds, and describe concrete real-world examples of how companies such as Spotify, Peltarion, and more are using the latest AI techniques.

Don’t miss out on the opportunity to listen to him and take part of the conference by signing up here by November 25th.

Anders Arpteg

More information about Anders Arpteg
Anders Arpteg (Ph.D., Head of Research at Peltarion) has been working with AI for 20 years in both academia and industry, with a Ph.D. in AI from Linköping University. Worked at Spotify for many years making use of big data and machine learning techniques to optimize the user experience. Now heading up a research team at Peltarion, operationalizing the latest and greatest AI techniques. At Peltarion, we have the ambitious goal of making deep learning and the latest AI techniques available for all companies, not just the large technology organizations. Also a member of AI Innovation of Sweden’s steering committee, AI adviser for the Swedish government, member of the Swedish AI Agenda, member of the European AI Alliance, founder of the Machine Learning Stockholm meetup group, and member of several advisory boards.

Seminar November 3rd Solving mean-field PDE with symmetric neural networks

2020-10-29

Welcome to the next talk related to Webinar DISA-DSM: stochastic analysis, statistics and machine learning, which is held by Huyên Pham (University Paris 7 Diderot, France)

When? 3 November at 13:00 Stockholm time.
Online via zoom: Contact Nacira Agram, nacira.agram@lnu.se, to get the link

Title: Solving mean-field PDE with symmetric neural networks
Abstract: We propose numerical methods for solving non-linear partial differential equations (PDEs) in the Wasserstein space of probability measures, which arise notably in the optimal control of McKean-Vlasov dynamics.
The method relies first on the approximation of the PDE in infinite dimension by a backward stochastic differential equation (BSDE) with a forward system of N interacting particles. We provide the rate of convergence of this finite-dimensional approximation for the solution to the PDE and its Lions-derivative. Next, by exploiting the symmetry of the particles system, we design a machine learning algorithm based on certain types of neural networks, named PointNet and DeepSet, for computing simultaneously the pair solution to the BSDE by backward induction through sequential minimization of loss functions. We illustrate the efficiency of the PointNet/DeepSet networks compared to classical feedforward ones, and provide some numerical results of our algorithm for the examples of a mean-field systemic risk and a mean-variance problem.
Based on joint work with M. Germain (LPSM, EDF) and X Warin (EDF).

We will book a room at LNU for those who wants to attend physically the seminar. Because of space restrictions due to Covid-19, please let me know if you want to do that.

A warm welcome!

Webinar DISA-DSM: stochastic analysis, statistics and machine learning

2020-10-26

Our new DISA-group Deterministic and Stochastic Modelling (DSM) invites you to a seminar on Tuesday 27th at 13.00, this seminar is a part of a seminar series so keep an eye out for more information.

Title: Rare events simulation: least-squares Monte Carlo method vs deep learning based shooting method

Speaker: Omar Kebiri (University B-TU Cottbus-Senftenberg, Germany)

Abstract: When computing small probabilities associated with rare events by Monte Carlo it so happens that the variance of the estimator is of the same order as the quantity of interest. Importance sampling is a means to reduce the variance of the Monte Carlo estimator by sampling from an alternative probability distribution under which the rare event is no longer rare. Determine the optimal (i.e. zero variance) changes of measure leads to a stochastic optimal control problem. The control problem can be solved by a stochastic approximation algorithm, using the Feynman-Kac representation of the associated dynamic programming equations which leads to an FBSDE, and we discuss numerical aspects for high-dimensional problems along with simple toy examples using two methods: least-squares Monte Carlo method and deep learning based shooting method.
Joint work with Carsten Hartmann, Lara Neureither, and Lorenz Richter.

Practical information: We will book a room at LNU for those who wants to attend physically the seminar. Because of space restrictions due to Covid-19, please let me know if you want to do that, otherwise a link will be provided. Contact Nacira Agram – nacira.agram@lnu.se for more information.

Call for presentations, Big Data Conference 2020

2020-10-23

A fast-forward (FF) + virtual poster (VP) sessions will be organized as part of the Big Data Conference 2020. In the FF presentations, each participant gets to show a 3-minute video to briefly summarize her/his research. Directly after the FF,  participants will be redirected to breakout rooms where it will be possible to present their VP and interact with the interested public.

The FF+VP presentations can focus on either ongoing research or new ideas:

1) Ongoing research will focus on research recently published or at an advanced stage of elaboration. The main goal here is to present research results of general interest for the public of the conference and eventually receive feedback on ongoing work.

2) New ideas will focus on future research, plans, or simply new ideas. The goal here is to share with the public their own plans, receive feedback, find partners and possibly find synergies to develop future research together.

 Submission

To submit to the FF+VP session, participants should submit a 500-word abstract briefly presenting the research by November 9th, 2020 to Diana Unander, diana.unander@lnu.se  Each participant can submit at most two abstracts.

Acceptance information will be sent out by November 13th, 2020.

If accepted, a video (in videos in 720p and mp4 format) of maximum 3 minutes should be sent by November 24th, 2020.

For more information or questions, please contact: