Hes-so Valais

Next workshops

Topic : Small data: challenges and opportunities
Date : November 10 2017
Location : Idiap Research Institute, Martigny
Registration : http://www.idiap.ch/workshop/small-data/registration
Idiap website : http://www.idiap.ch/workshop/small-data/

Data-driven approaches are increasingly used in support of scientific, technological and economical activities.
Yet, in many situations the amount of data is drastically limited due to time and cost constraints, feasibility, and intrinsic rarity. This includes notably:

  • Ecological and medical studies, where observations of rare species/pathologies are scarce
  • Catastrophe modelling (related to waste storage, climate evolution, financial crises, etc.) where samples are small or even void due, e.g., to extreme conditions or large time-scales
  • Resource-intensive experiments performed on natural and artificial systems, e.g. in protein synthesis, in robot calibration, or in hi-fidelity simulation-based design

In this second Valais/Wallis Artificial Intelligence (AI) Workshop, we will focus on several facets of such challenging topics, which will be tackled through presentations from several leading Swiss-based researchers having to cope with small data with motivating applications in environment and geosciences, medicine, energy engineering, robotics and beyond. In this context, general and domain-specific topics in visualization and extraction of relevant features from small and big data will also be discussed.

Program :


08h45 David Ginsbourger (Opening talk on small data)

09h10 Antoine Guisan Keynote lecture (

10h00 Spotlight session (short talks by PhD students)

10h30 Coffee break

11h00 Roger Schaer (Web-Based Tools for Exploring the Potential of Quantitative Imaging Biomarkers in Radiology)

11h25 Florian Evequoz (Small Data, Big Picture. Visualizing data to get insights)

11h50  Sylvain Calinon  Robot learning from few demonstrations by exploiting the structure and geometry of data

Human-centered robot applications require the robots to learn new skills by interacting with the end-users. From a machine learning perspective, the challenge is to acquire skills from only few interactions, with strong generalization demands. It requires: 1) the development of intuitive active learning interfaces to acquire meaningful demonstrations; 2) the development of models that can exploit the structure and geometry of the acquired data in an efficient way; 3) the development of adaptive control techniques that can exploit the learned task variations and coordination patterns. The developed models often need to serve several purposes (recognition, prediction, online synthesis), and be compatible with different learning strategies (imitation, emulation, exploration). For the reproduction of skills, these models need to be enriched with force and impedance information to enable human-robot collaboration and to generate safe and natural movements. 

I will present an approach combining model predictive control and statistical learning of movement primitives in multiple coordinate systems. The approach will be illustrated in various applications, with robots either close to us (robot for dressing assistance), part of us (prosthetic hand with EMG and tactile sensing), or far from us (teleoperation of bimanual robot in deep water).

12h15  Mariam Barque (Small data for wind energy prediction of production)

12h40 Closing remarks by the organizers followed by a lunch

Access :