Who are we?

The Idiap Research Institute is an independent, nonprofit research foundation based in Martigny (VS), in the Swiss Alps. Its activities encompass fundamental research, training (graduate and post-graduate level), and technology transfer activities in the area of artificial intelligence both theoretical and applied.

Institute's research fields include, among others, voice and visual recognition, machine learning, human-machine interaction, robotics, language analysis, social computing, genomics, or bio-imaging. Idiap's activities are divided into eleven complementary research groups and eleven application areas. The Institute is affiliated with the Ecole Polytechnique Fédérale de Lausanne (EPFL) and is one of the most active independent research institutions in information technology.

Since 2008, Idiap has reinforced its connection with Ecole polytechnique fédérale de Lausanne through a joint development plan and a strategic alliance with the "ETH Domain". Several of Idiap's scientists hold an EPFL academic title and are teaching graduate and undergraduate courses. They also supervise students who pursue their doctoral thesis research at Idiap.

The director of the Institute is Prof. Hervé Bourlard, also professor at EPFL and director of the LIDIAP Laboratory.

Idiap's budget, which amounts to more than 12 million Swiss francs, is 54% financed by research projects awarded following competitive processes, and 46% by public funds. There are a little over 100 staff members in residence at Idiap, including senior researchers, researchers, research associates, post-doctoral fellows, PhD students, and visiting researchers. Administrative staff provides unique computational and research support services. Since 2007, all the personnel work at Centre du Parc in Martigny, where it currently occupies 2,500 m2 of premises over four floors.

 

What are Idiap's values and missions?

Idiap was founded in 1991 by the Town of Martigny, the State of Valais, EPFL (Ecole polytechnique fédérale de Lausanne), the University of Geneva, and Swisscom, with the support of the Dalle Molle Foundation. Inspired by Angelo Dalle Molle visionary approach, the institute’s vision is to promote quality of life through scientific progress made in relation to human and media computing. Therefore the motto: "Artificial Intelligence for Society".

In line with these core values, Idiap wishes to further contribute to the advancement of women in science and engineering, specifically in the area of Artificial Intelligence in the widest sense, by offering a Fellowship Program for Female Researchers.

The Institute is recognized by the Swiss Confederation as part of the strategic domain of the Federal Institutes of Technology and is supported under the Federal Law on the Promotion of Research and Innovation (LERI), art. Al. 3b. Idiap's activities are oriented in accordance with three complementary missions:

  • Research

Conducting fundamental research projects at the highest level in its preferred areas, thus taking its place among the best on a national, European and global scale. Idiap benefits from a wide national and international network of partners and works actively with large universities, public and private research centers, etc. This collaboration is always implemented through formal research projects, including SNSF, EU (and sometime US) research programs.

  • Technology Transfer

Idiap is currently amongst the most important research center in Valais also putting special emphasis on research software maintenance, to facilitate technology take-up, technology transfer, creation of spin-offs, or attracting startups. As discussed later there is an entire team /currently around 12 people) devoted to these important aspects, and working in close collaboration with CTI/KTI, as well as the IdeArk incubator.

  • Academic and Training Activities

Through our academic anchoring with EPFL (and the EDEE and EDIC Doctoral Programs), our teaching activities at EPFL, as well as our connections with numerous other academic institutions, we also fund and supervise a large number of PhD students (35 per year on average), while also hosting international master and intern students. Numerous Idiap internal activities towards improving personal research and communication skills are also provided.

 

Research Programme 2017-2020

Planning document for the State Secretariat for Education, Research and Innovation (SEFRI)

Research Programme 2017-2020

 

Application Domains

1
Biometric Security & Privacy


Development of new image processing and pattern recognition algorithms for face recognition (2D, 3D, and near infrared); speaker recognition; attack detection (antispoofing); the study of emerging biometrics modalities (electrophysiology and veins ); open science through the use of our own library.

2
Biosignal Processing


Analysis of e-health data; detection of human biological signals for medical and scientific applications; use of machine learning techniques; data acquisition and analysis; open science.

3
Energy Informatics


Creation of smarter energy systems; optimization of the energy efficiency of a system beyond its individual; optimization of the energy resources planning to protect the environment; simulation of energy transition scenarios including intelligent control and adjustment mechanisms for buildings; production and storage of renewable energies in the context of climate change.

4
Computational Bio Imaging


Development of algorithms for deconvolution and super-resolution in optical microscopy; three-dimensional tomographic reconstruction, more generally, combination of unusual detection methods and devices with computational software to produce images ideally suited to the observation and quantification of complex, living biological systems.

5
Machine Learning


Computational properties of statistical learning; automatic image analysis, particularly the automatic extraction of significance; object detection; tracking of people and biological structures.

6
Natural Language Understanding


Neural machine translation; summarization; information retrieval and extraction; text classification; attention-based deep learning models of language understanding; representation learning for modeling abstraction and natural language inference.

7
Perception & Activity Understanding


Development of algorithms based on computer vision and learning and data fusion methods in order to detect and track objects and people, to represent and characterize their state, and to model sequential data and interpret them in the form of gestures, behaviors, and social relations.

8
Robot Learning & Interaction


Develop statistical approaches for encoding movements and behaviors in robots evolving in unconstrained environments; models with multiple roles (recognition, prediction, reproduction); learning strategies (imitation, emulation, incremental correction, or exploration).

9
Social Computing


Behavioral analysis of face-to-face interactions; crowdsourcing and the large-scale treatment of urban data; smartphones and social networks.

10
Speech & Audio Processing


Statistical automatic speech recognition; voice synthesis and the generic processing of audio information (source localization; microphone networks; speaker segmentation; information indexing; very-low-bit-rate speech coding; background noise analysis).

11
Uncertainty Quantification & Optimal Design


Gaussian process methods; the planning of numerical experiments for optimization, inversion, and other related problems. Areas of application include energy and geosciences, with collaborations ranging from safety engineering to hydrology and climate sciences.

 

 

Key figures

Budgetary Resources (2019)

  • Total budget: 12.26 million
  • Public funding: 42%
  • Financing from projects: 54%
  • Provisions: 4%

Human Resources (December 2019)

  • 5 professors
  • 1 senior lecturer MER
  • 9 permanent senior researchers
  • 7 research associates
  • 28 postdocs
  • 48 research assistants
  • 14 development engineers
  • 6 system engineers
  • 27 internships (average/year)
  • 19 administration staff
  • 31 nationalities

Scientific activities (2019)

  • participation in 67 research projects
  • project management in more than 48 consortia
  • 112 peer reviewed scientific publications
  • 6 doctoral theses

Infrastructures

  • 2500 m2 of office space
  • 3 meeting rooms
  • 1 conference room
  • 1 server room
  • 1 library
  • 3 laboratories

IT

  • 140 workstations
  • 60 physical servers
  • 500 virtual computers
  • 450 TB of storage
  • 1000 CPU dedicated for scientific computing
  • 130 GPU dedicated for scientific computing
  • 13 TB RAM dedicated for scientific computing

Founding partners

  • City of Martigny

  • State of Valais

  • EPFL

  • University of Geneva

  • Swisscom