2 Postdoctoral researchers in interactive AI and machine learning
The positions belong to the Aalto career system and the selected persons will be appointed for a two-year fixed term appointment with an option for renewal.
- Probabilistic machine learning with human in the loop (Prof. Samuel Kaski)
One of the core questions in machine learning at the moment is how to interact with humans.
We turn this question into a probabilistic modelling problem, and model both the user and the task to drive the interaction. The solutions need combinations of probabilistic modelling, reinforcement learning and approximate Bayesian computation. We are looking for a postdoc who already masters some of these and offer an opportunity to learn the rest and work with us on this exciting bleeding-edge problem. Two recent sample papers:
 Daee et al., Knowledge elicitation via sequential probabilistic inference for high-dimensional prediction. Machine Learning, 106: 1599–1620 (arXiv: 1612.03328)
 Kangasrääsiö Kaski, Inverse Reinforcement Learning from Summary Data, arXiv:1703.09700
- Modeling and machine learning in HCI (Prof. Antti Oulasvirta)
The position offers an exciting opportunity to learn about and work on applications of machine learning methods and computational models of cognition, perception, and behavior in interactive systems. For relevant previous papers of the team, example papers, see , , and .
 CHI 2017: https://dl.acm.org/ft_gateway.cfm?id=3025576 type=pdf
Should have a PhD in computer science, statistics, cognitive science, or other area relevant to the themes and publication record in top conferences or journals in machine learning, AI, statistics, cognitive science, or HCI. Good command of English is a necessary prerequisite.
Compensation, working hours and place of work
The salary for the position is between 3 498 and 3 673 EUR per month depending on experience and qualifications. In addition to the salary, the contract includes occupational health benefits, and Finland has a comprehensive social security system. The annual total workload of teaching staff at Aalto University is 1 600 hours. The position is located at the Aalto University Otaniemi campus.
Application material and procedure
Please send your application through the electronic recruitment system ‘Saima’ (see link at the bottom of the page) no later than Oct 31, 2017 Finnish time. Include your CV, list of publications, a brief research statement, transcript of the doctoral studies, and names and contact information of two senior academics willing to give more information.
Short-listed candidates may be invited for an interview on the Otaniemi campus of Aalto University in Helsinki or for an interview conducted via Skype. While all applicants who have submitted an application by the deadline will be appropriately considered, Aalto University reserves the right to consider also other candidates for the announced position.
• Professor Samuel Kaski and Professor Antti Oulasvirta, e-mail "firstname.lastname [ chez ] aalto.fi" (research related information)
• HR Coordinator, Mr. Stefan Ehrstedt, e-mail "firstname.lastname [ chez ] aalto.fi" (application process, practical arrangements)
Application material will not be returned.
Espoo, 13 Oct 2017
About Aalto University
Aalto University is a new university created in 2010 from the merger of the Helsinki University of Technology, Helsinki School of Economics and the University of Art and Design Helsinki. The University’s cornerstones are its strengths in education and research, with 20,000 basic degree and graduate students. These positions are in collaborating groups of the Department of Computer Science and Department of Communications and Networking.
The Helsinki Metropolitan area forms a world-class information technology hub, attracting leading scientists and researchers in various fields of ICT and related disciplines. Moreover, as the birthplace of Linux, and the home base of Nokia/Alcatel-Lucent/Bell Labs, F-Secure, Rovio, Supercell, Slush (the biggest annual startup event in Europe) and numerous other technologies and innovations, Helsinki is fast becoming one of the leading technology startup hubs in Europe. See more e.g. at http://www.investinfinland.fi/.
As a living and working environment, Finland consistently ranks high in quality of life, and Helsinki, the capital of Finland, is regularly ranked as one of the most livable cities in the world. See more at https://finland.fi and http://www.helsinkitimes.fi/finland/finland-news/domestic/14966-helsinki-ranked-again-as-world-s-9th-most-liveable-city.html
Brandeis University, Department of Mathematics, invites applications for a tenure-track position in applied mathematics at the rank of assistant professor beginning fall 2018. Candidates must have a Ph.D. in mathematics or a related field, demonstrate potential for excellence in research, and display a commitment to teaching. An ideal candidate will be expected to help to build an applied mathematics program within the department, and to interact with other science faculty at Brandeis. Candidates from all areas of applied mathematics will be considered.
First consideration will be given to applications received by November 15.
Applications should include an AMS cover sheet, a curriculum vitae, research and teaching statements, and four letters of recommendation, one of which addresses teaching effectiveness. Applications should be submitted through MathJobs.org:
Tenure-track or Tenured Professor in Machine Learning
The Department of Computer Science (CS) at Aalto University (Helsinki, Finland) invites applications for a tenure-track or tenured professor in Machine Learning. Aalto is currently investing heavily in fundamental research in Artificial Intelligence and Finland in applied AI; Aalto Department of Computer Science is the right place to pursue the next generation of machine learning and AI. As of last year the CS department has already recruited two professors in machine learning and is now recruiting again two new professors: one in machine learning and another call in AI is to open soon.
With nine professors and ca. 100 PhD students and postdocs working in machine learning, data mining, and probabilistic modelling, Aalto Department of Computer Science is one of Europe’s leading centres of research in the field of the call. The current faculty’s strength is apparent in e.g. the high volume of extremely competitive funding from the Academy of Finland (equivalent to National Science Foundation) and the European Research Council (ERC), as well as the volume of high-achieving international students entering the MSc and PhD programs in Machine Learning and Data Mining. Furthermore, as the foremost CS educator in Finland, the department is home to the majority of the best Finnish students. In total, the department hosts 44 professors with diverse interests, forming a fertile environment for cross-disciplinary collaborations. On a larger scale, as a technology-friendly, yet small country, Finland offers ample opportunities for low-overhead collaboration with ind!
ustrial and government partners as well.
The review of the new position will begin on Oct 29, 2017 and the position will remain open until filled. For further information, please visit http://www.aalto.fi/fi/about/careers/jobs/view/1533/.
This Call was drafted on the campus Jussieu in Paris by a French group comprising researchers and scientific publishing professionals working together in Open Access and Public Scientific Publishing task forces of BSN (Bibliothèque scientifique numérique, or Digital Scientific Library).
This Call is aimed at scientific communities, professional associations and research institutions to promote a scientific publishing open-access model fostering bibliodiversity and innovation without involving the exclusive transfer of journal subscription monies to APC payments.
Jussieu Call for Open science and bibliodiversity
As asserted in the Amsterdam Call released in 2016, Open Access to scientific publishing is at a crossroads. After several years of an exacting struggle aiming at persuading somewhat skeptical stakeholders, Open Access has now won strong support and a rapid shift of the scientific communication system to an Open Access publishing model can be expected. “The time for talking about open access is now past”. “The time for talking about open access is now past”.
The means to achieve the goal of Open Access are yet to be discussed. We believe that the issue of business models has to be refocused in the broader perspective of the editorial processes and methods upon which research and innovation will rely in the future. and that they may only develop for the benefit of a very broad bibliodiversity.
We find it necessary to foster an Open Access model that is not restricted to a single approach based on the transfer of subscriptions towards APCs (publication fees charged to authors to allow free access to their articles). Such an approach would hamper innovation and otherwise would slow if not check the advent of bibliodiversity. Therefore, we adhere to the Joint Statement of UNESCO and the Confederation of Open Access Repositories (COAR) on Open Access which highlights all the difficulties caused by this single model.
Our goal is thus to develop and implement alternative models matching the aims of open science by asserting the need of supporting innovation for a thorough renewal of publishing functions as proclaimed by the Association of European Research Libraries (LIBER) and the International Council for Science (ICSU).
We, stakeholders of Open Access scientific publishing, hereby claim that:
- 1 Open Access must be complemented by support for the diversity of those acting in scientific publishing – what we call bibliodiversity – putting an end to the dominance of a small number among us imposing their terms to scientific communities;
- 2 the development of innovative scientific publishing models must be a budget priority because it represents an investment into services meeting the genuine needs of researchers in our digital age;
- 3 experiments should be encouraged in writing practices (publishing associated data), refereeing (open peer-reviewing), content editorial services (beyond-pdf web publishing) and additional services (text mining);
- 4 the research evaluation system should be thoroughly reformed and adapted to the practices of scientific communication;
- 5 more investment efforts in open source tools upon which these innovative practices are based should be made and coordinated;
- 6 the scientific community needs a secure and stable body of law across different countries to facilitate the availability of text mining services and thus strengthen their use;
- 7 the scientific communities must be able to access national and international infrastructures which guarantee the preservation and circulation of knowledge against any privatization of contents. Business models should be found which preserve their long-term continuity;
- 8 priority should be given to business models that do not involve any payments, neither for authors to have their texts published nor for readers to access them. Many fair funding models exist and only require to be further developed and extended: institutional support, library contributions or subsidies, premium services, participatory funding or creation of open archives, etc. We endorse the clear message to the scientific community at large released by the League of European Research Universities (LERU): Research funding should go to research, not to publishers! This is why current journal subscription spendings should be changed into investments enabling the scientific community to regain control over the publishing system and not merely into new spendings only earmarked to pay the publication fees for researchers to commercial publishers.
We call on creating an international consortium of stakeholders whose primary aim should be to pool local and national initiatives or to build an operational framework to fund open access publishing, innovation and sharing of resulting developments. We call on research organizations and their libraries to secure and earmark as of now a share of their acquisition budgets to support the development of scientific publishing activities, which are genuinely open and innovative, and address the needs of the scientific community.
Serge BAUIN; Céline BARTHONNAT; Christine BERTHAUD; Thierry BOUCHE; Francois CAVALIER; Gregory COLCANAP; Odile CONTAT; Nathalie FARGIER; Thierry FOURNIER; Anne-Solweig GREMILLET; Frédéric HÉLEIN; Odile HOLOGNE; Emmanuelle JANNES-OBER; Jacques LAFAIT; Jean François LUTZ; Sandrine MALOTAUX; Jacques MILLET; Pierre MOUNIER; Jean-Francois NOMINÉ; Christine OKRET-MANVILLE; Christine OLLENDORFF; Sébastien RESPINGUE-PERRIN; Julien ROCHE; Laurent ROMARY; Dominique ROUX; Joachim SCHOPFEL; Bernard TEISSIER; Armelle THOMAS; Céline VAUTRIN
Signing Institutions: (TBA)
The signing process of the Call is currently underway. The list of signing institutions will be added shortly.
The signatories’names, institution title and logo will then be displayed on this page as their approval is received.
- le Conseil scientifique de l'INSMI
- le CA de la SMAI
- le First Consortio Assembly from Ibero-Americana and the Caribbean
How to sign this Call :
Scientific communities, research institutions and scientific and technical information professional associations and organizations are called to support the Jussieu Call.
How to do this : The President/chairperson of the signatory sends an email indicating the agreement of his/her institution together with an official web logo and institutional title
to : JussieuCall at gmail dot com
Two Postdoctoral Research Positions in Machine Learning, University of Cambridge, UK
Two Post-doc positions in Machine Learning for Remote Sensing and Geosciences [ERC project], Universitat de Valencia, Spain
Postdoc Position in Machine Learning and Network Science, The Northeastern University’s Network Science Institute and College of Computer and Information Science in Boston, MA
Research Fellow in Deep Learning for Autonomous Driving, University of Lincoln, UK
Full Professor of Machine Learning for Computer Simulations, University Stuttgart, Germany
- De : Zoubin Ghahramani zoubin [ chez ] eng.cam.ac.uk
Two Postdoctoral Research Positions in Machine Learning
University of Cambridge, UK
We are seeking two highly creative and motivated Research
Assistants/Associates to join the Machine Learning Group at the
University of Cambridge. The positions are for up to 24 months with a
possible extension. Details below!
http://www.jobs.cam.ac.uk/job/14617/ (Closing Date: 7 September 2017)
http://www.jobs.cam.ac.uk/job/14619/ (Closing Date: 7 September 2017)
The two positions are funded by Samsung. The successful candidates
will collaborate with Professor Zoubin Ghahramani, Dr. José Miguel
Hernández Lobato, Dr. Isabel Valera and Professor Carl E. Rasmussen,
including one PhD student funded by the same grant and Samsung data
Interviews are expected to happen in mid-September 2017 at the
Department of Engineering. A skype interview will be possible for
applicants who cannot attend in person.
Post 1: This postdoctoral Research Fellow will be working on areas
related to deep learning and Bayesian methods, as well as ideally
having an interest in one-shot learning, semi-supervised learning and
Experience in two or more of the following areas will be necessary:
Bayesian methods, deep learning, one-shot learning, semi-supervised
learning and active learning.
http://www.jobs.cam.ac.uk/job/14617/ (Closing Date: 7 September 2017)
Post 2: This postdoctoral Research Fellow will be working on
automating machine learning methods, addressing problems of automatic
data preprocessing and modelling, including missing data estimation
and outlier identification.
Experience in two or more of the following areas will be necessary:
Bayesian methods, probabilistic programming, MCMC, graphical models,
http://www.jobs.cam.ac.uk/job/14619/ (Closing Date: 7 September 2017)
Successful applicants will have or be near to completing a PhD in
computer science, information engineering, statistics, mathematics, physics,
or a related area, with extensive research experience and a strong publication
record. Ideally candidates will have papers in top machine learning conferences
such as NIPS, UAI, ICML, ICLR or AISTATS. Excellent mathematical and programming
skills are essential.
If you have any questions about these vacancies or the application
process, contact Mrs. Rachel Fogg, email: div-f [ chez ] eng.cam.ac.uk, Tel:
+44 1223 3 32752
- De : Gustau Camps-Valls gustau.camps [ chez ] uv.es
Two Post-doc positions in Machine Learning for Remote Sensing and Geosciences [ERC project]
We are searching for outstanding postdoc candidates with a strong interest in machine learning and geosciences to cover two post-doc positions in the Image and Signal Processing (ISP) group in the Universitat de Valencia, Spain, http://isp.uv.es. The positions are fully funded by an ERC Consolidator Grant 2015-2020 entitled "Statistical Learning for Earth Observation Data Analysis" (SEDAL), http://isp.uv.es/sedal.html, under the direction of Prof. Gustau Camps-Valls.
The project and job description
We aim to develop new statistical inference methods to analyze Earth Observation (EO) data. Machine learning models have helped to monitor land, oceans, and atmosphere through the analysis and estimation of climate and biophysical parameters. Current approaches, however, cannot deal efficiently with the particular characteristics of remote sensing data. We will develop advanced regression (retrieval, model inversion) methods to improve efficiency, prediction accuracy and uncertainties, encode physical knowledge about the problem, attain self-explanatory models, learn graphical causal models to explain the complex interactions between essential climate variables and observations, and discover hidden essential drivers and confounding factors in Climate/Geo Sciences.
Highly motivated researchers with a PhD degree in computer science, statistics, machine learning, electrical engineering, physics, or mathematics are encouraged to apply!
All candidates should have a solid understanding and knowledge of machine learning and statistics, and being particularly interested in remote sensing and/or geoscience problems. The topics are focused on regression, graphical models and causal inference. Good programming skills (Matlab/Python/R/C++), a critical and organized sense for data analysis, as well as commitment, strong communication, presentation and writing skills are a big plus.
- Deadline: Send your application as soon as possible. Positions will be filled as soon as we have the right candidate!
- How? Send me: 2-pages CV, motivation letter, list of papers and one recommendation letter or contact
- Who: PhD in maths, physics, machine learning, or related disciplines. Also, we care about the gender issue!
- When? Preferred starting date: October 2017
- How long? 3 years contract
- How much? Salary according to UV scales including social security, health insurance benefits, and travel money
- Where? Valencia, Spain, Mediterranean city, nice weather, hike and beach. Excellent cost-of-living index = 55
- Before applying: Informal inquiries may be addressed to Prof. Dr. Gustau Camps-Valls, gustau.camps [ chez ] uv.es
- Ready to apply? Send your dossier in one single PDF to gustau.camps [ chez ] uv.es, subject: "SEDAL application"
- De : Tina Eliassi-Rad eliassi [ chez ] cs.wisc.edu
Postdoc Position in Machine Learning and Network Science
A postdoctoral position is available for an outstanding individual to conduct research at the intersection of machine learning and network science. The project involves research on network embedding, representation learning, higher-order structures in networks, and the network completion problem. The mentor for this position is Professor Tina Eliassi-Rad at the Northeastern University’s Network Science Institute and College of Computer and Information Science in Boston, MA.
- A recent Ph.D. in Computer Science, Network Science, Information Science, Statistics, or related fields.
- Expertise in machine learning and data mining.
- Experience working with complex networks.
- Strong programming skills in Python, C, MATLAB, or R.
Start date: January 8, 2018
Salary: Commensurate with experience.
Duration: One year with the possibility of renewal for a second year.
Deadline: The initial review date is November 15, 2017 (but sooner is better). Position will remain open until filled.
Please email a cover letter and your curriculum vitae to t.eliassirad [ chez ] northeastern.edu as one single PDF file; and make sure to include the string "[MLNS Postdoc]" at the beginning of your subject line.
Arrange for two letters of recommendation to be sent directly to t.eliassirad [ chez ] northeastern.edu.
- De : Gerhard Neumann geri [ chez ] robot-learning.de
Research Fellow in Machine Learning - Fixed Term
School of Computer Science
Salary: From £32,548 per annum
Fixed Term for approximately 1 year
Closing Date: Monday 25 September 2017
Interview Date: To be confirmed
We seek to employ a highly motivated post-doctoral researcher that will be working on a joint project with Toyota on the topic "Hierarchical Deep Learning for Autonomous Driving". You should hold a PhD or be near to completion, and should be able to demonstrate a good track record in at least one of the following research fields:
Movement Primitives and Movement Representations
Variational Bayes and Hierarchical Bayesian Models
The contract is for one year. If the project is successful it is likely to be extended by Toyota for several years.
Once in post, you will be working with Professor Gerhard Neumann on the aforementioned research topics. The successful candidate will be a member of the Lincoln Centre for Autonomous Systems (L-CAS). The L-CAS is part of the School of Computer Science at the University of Lincoln and specialises in the integration of perception, learning, decision-making, control and interaction capabilities in autonomous systems and the application of this research in fields such as personal robotics, agri-food, healthcare, security, and intelligent transportation. The L-CAS is one of the fastest growing robotics groups in the UK. We provide a highly-dynamic inter-disciplinary research environment with a broad range of collaboration opportunities and a large variety of robots to work with. In this project, you will have access to sophisticated car simulators and real autonomous cars provided by Toyota Europe.
The University of Lincoln is a forward-thinking, ambitious institution and you will be working in the heart of a thriving, beautiful, safe and friendly city. The School provides a stimulating environment for academic research, and is located on the picturesque waterfront campus in the historic and vibrant city of Lincoln. The University has just announced a £130M investment programme, a significant part of which is being invested in new, purpose-built facilities for the School of Computer Science. Lincoln itself is a small but fast growing city in the east-midlands. It offers a fantastic life quality given by moderate living costs, a medieval city centre including a famous cathedral and a beautiful ancient canal system that is still in use by many house boats nowadays.
If you would like to know more about this opportunity, please contact Professor Gerhard Neumann (Professor of Computational Learning for Autonomous Systems, gneumann [ chez ] lincoln.ac.uk).
Prof. Gerhard Neumann
Chair of Computational Learning for Autonomous Systems
School of Computer Science
University of Lincoln
Lincoln, LN6 7TS
The University of Lincoln, located in the heart of the city of Lincoln, has established an international reputation based on high student satisfaction, excellent graduate employment and world-class research.
The information in this e-mail and any attachments may be confidential. If you have received this email in error please notify the sender immediately and remove it from your system. Do not disclose the contents to another person or take copies.
Email is not secure and may contain viruses. The University of Lincoln makes every effort to ensure email is sent without viruses, but cannot guarantee this and recommends recipients take appropriate precautions.
The University may monitor email traffic data and content in accordance with its policies and English law. Further information can be found at:http://www.lincoln.ac.uk/legal.
- De : Marc Toussaint marc.toussaint [ chez ] informatik.uni-stuttgart.de
Full Professor (W3)
Machine Learning for Computer Simulations
STUTTGART CENTRE FOR SIMULATION SCIENCES (SC SIMTECH)
AT THE EARLIEST OPPORTUNITY
The new professor is expected to cover the field of machine learning
for computer simulations within the Stuttgart Centre for Simulation
Sciences (SC SimTech). We are looking for a researcher with an
excellent international reputation in innovative methods for machine
learning who will apply, advance and integrate these methods in
the general area of simulation sciences. A combination between
theoretical developments and applications to engineering and/or
natural sciences is desired.
We expect significant contributions to teaching in selected (mostly
German) Bachelor and Master programs. The ideal candidate is
active in acquiring third-party funding and will participate in
collaborative and interdisciplinary research projects at the university.
The announced position is one of ten new professorships that are to
be installed at the Universities of Stuttgart and Tübingen as part of
the Cyber Valley initiative, a research alliance in the field of intelligent systems.
The requirements for employment listed in § 47 and § 50 Baden-
Württemberg university law apply.
Applicants are asked to send their detailed curriculum vitae, copies
of certificates, list of publications, research and teaching statement,
reprints of three to five selected publications, and the completed
application form from
by e-mail to
Prof. Dr.-Ing. Wolfgang Nowak wolfgang.nowak [ chez ] iws.uni-stuttgart.de
Institut für Wasser- und Umweltsystemmodellierung
70569 Stuttgart, Germany
no later than 24th of October 2017.
The University of Stuttgart has established a Dual Career Program
to offer assistance to partners of those moving to Stuttgart.
For more information, please visit the website
If you have any questions, please do not hesitate to contact us informally:
This page and project is under-construction and is just a preliminary draft.
Following the last US elections, safety of US government climate data appeared at risk (Brady Dennis, Scientists are frantically copying U.S. climate data, fearing it might vanish under Trump, Washington Post, 13th December 2016). The new head the Environmental Protection Agency nominated by the government, Scott Pruitt, called himself a "leading advocate against the EPA's activist agenda". The new head of the Department of Energy, Rick Perry, claimed that "we have been experiencing a cooling trend", and said "there are a substantial number of scientists who have manipulated data so that they will have dollars rolling into their projects".
Some US researchers have reported deletion of environmental data archivs (ex: V.Hermann report in the gardian 28th march 2017). An abreviated timeline of the Trump administration’s environmental actions and policy changes, as well as reactions to them, is maintained and periodically updated by the National Geographic (National Geographic- A Running List of How Trump Is Changing the Environment). On the the 1st of July 2017, president Donald Trump announced that U.S. is pulling out of Paris Climate Agreement (a review is available on wikipedia).
In reaction, several initiatives to back up the many climate databases held by US government agencies arose in order to prevent their potential removing by the administration. Jan Galkowski, a statistician working at Akamai Technologies, began downloading climate data on the 11th of December 2016. John Baez, a mathematician who animates the Azymuth project, joined in to coordinate publicity for the project and it gave birth to the Azimuth Climate Data Backup Project. Azimuth initiated a preliminary crowd founding and achieved some backup of some important publicly open climate data basis. Scott Maxwell (Google) set up a 10-terabyte account on Google Drive and started backing up data. Sakari Maaranen (Ubisecure) and Hetzner set up a server with that provides 10 Tb of storage, gigabit bandwidth and 30 Tb of a monthly traffic. A crowd funding campaign was initiated and 40 terabytes of US government databases on climate change and the environment were backed up on servers.
In a joint and parallel effort, the Data Refuge Project of the University of Pennsylvania organised the climate mirror project, an open project to mirror public climate datasets. Soon after some datasets were published, individuals started mirroring them. climate.daknob.net was one of the first to help and currently a lot of datasets are hosted here.
Pierre Collet, Anne Jeannin-Girardon, Pierre Baudot (Complex System-Digital Campus Unitwin Unesco and the University of Strasbourg) set up a server with 40Tb for the Azimuth Climate Data Backup Project. The list of databasis and progress of the backup by the Azimuth Climate Data Backup Project is available in this archiv. The safety of US government environmental databases safely back is ensured by computing hash codes for these datasets to help to prove the backups are authentic. More informations are available on Azimuth Climate Data Backup Project.
All this has been achieved thanks to personal and volunteers initiatives and funding, and it is now time for an international institution, Unesco to get involved in order to allow a long term backing up and to avoid some other future problem of this kind. These local US and recent events points out a general problem that should have been assessed long ago, upon the existence and status of archivs of data that appear crucial in sustainable development at the global scale and on the long term. Such data shall not be vulnerable to some ponctual or local political fluctuations. Since some other data, like health data (ex: epidemiological data), ecological data (ex: biodiversity survey) or cultural (indigenous oral and writing cultural data), in the field of Unesco, are faced to the same problem of potential deletion, the scope of the climate data archivs started by the US recent events shall be broaden and pursued on the long run and it is indeed in the original foundational guidelines of Unesco to enlarge the data archivs to biodiversity and cultural data.
- Idea an aims: A status of Unesco World Heritage for Climate, Heath and Cultural critical data
“What steam was to the 18th century, electricity to the 19th, and hydrocarbons to the 20th, data will be to the 21st century. That’s why I call data a new natural resource.” Ginni Rometty, Chairman, President and CEO of IBM. Indeed some data are peculiarly sensible and critical environmental, cultural, societal and scientific ressources that has to preserved from any deletion, falsification and stored safely on the long term. They are the record of the evolution of our ecosystems, of the evolution of Human societies in its environment and shall be kept available for the next generations and preserved against political or societal fluctuations. Sustainable development, predicting and monitoring future outcomes, prevention of societal and environmental risks (...), as a first necessary and crucial condition, rely on the storing and availability of such data: erasing the past makes us blind to the future.
The United Nations Educational, Scientific and Cultural Organization (UNESCO) is a specialized agency of the United Nations (UN). Its declared purpose is to contribute to peace and security by promoting international collaboration through educational, scientific, and cultural reforms in order to increase universal respect for justice, the rule of law, and human rights along with fundamental freedom proclaimed in the United Nations Charter. Projects sponsored by UNESCO include literacy, technical, and teacher-training programmes, international science programmes, the promotion of independent media and freedom of the press, regional and cultural history projects, the promotion of cultural diversity, translations of world literature, international cooperation agreements on secure the world cultural and natural heritage (World Heritage Sites) and to preserve human rights, and attempts to bridge the worldwide digital divide. UNESCO's aim is "to contribute to the building of peace, the eradication of poverty, sustainable development and intercultural dialogue through education, the sciences, culture, communication and information". Other priorities of the organization include attaining quality Education For All and lifelong learning, addressing emerging social and ethical challenges, fostering cultural diversity, a culture of peace and building inclusive knowledge societies through information and communication. (Wikipedia). On the initiative of the United Nations Secretary-General, Global Pulse a flagship innovation on big data was launched and should provide a partner of the project within UN. Its vision is a future in which big data is harnessed safely and responsibly as a public good. Its mission is to accelerate discovery, development and scaled adoption of big data innovation for sustainable development and humanitarian action.
The Security Council of UN, through its 24th march 2017 report, stated that "illegal attacks on sites and buildings devoted to religion, education, art, science or for historical charitable purposes or monuments may constitute , in certain circumstances and in accordance with international law a war crime and that the perpetrators of these attacks must be brought to justice" (Cf. article). Before the Security Council on Friday, 24 March 2017, Irina Bokova, Director-General of Unesco said: "The deliberate destruction of heritage is a war crime, it has become a tactic of war to undermine society in the long term, in a strategy of cultural cleansing, which is why the defense of cultural heritage is much more than a cultural issue, it is a security imperative inseparable from the defense of human lives ". She also recalled that weapons were not enough to defeat violent extremism. "Building peace also depends on culture, which requires education, prevention and the transmission of heritage." (Cf. article). Moreover, Unesco will organize on the 28th Sept 2017 a whole conference 'IPDCtalks' to highlight and elaborate on the importance of Access to Information for all sustainable development efforts around the world.
The World Heritage Convention is an international treaty between Member States of the United Nations. It seeks to identify, protect, conserve, present and transmit to future generations cultural and natural heritage of Outstanding Universal Value. The World Heritage Convention is rooted in the recognition that cultural and natural heritage is among the priceless and irreplaceable assets, not only of each nation, but of humanity as a whole. The loss, through deterioration or disappearance, of any of these most prized properties constitutes an impoverishment of the heritage of all the peoples of the world (Unesco Guide Chap1, p.14. The Operational Guidelines define Outstanding Universal Value as being cultural and/or natural significance which is so exceptional as to transcend national boundaries and to be of common importance for present and future generations of all humanity (Paragraph 49).
- Application to Unesco World Heritage
The guide for submitting a Unesco World patrimony proposal provides all the information for the submission. The project has to fulfil the forms and criteria imposed by Unesco to submit a proposal.
The main difficulty with respect to usual criteria of Unesco World Heritage is the "localization" requirements: a Numerical patrimony is in a weak sens (at least can be in some cases) "de-materialized" or "unlocalized" (in the sense of cloud computation and storage, that appears to be the sustainable future development of computational resources). We will have to stress the peculiarity of an e-patrimony, of a numerical patrimony and to propose an evolution in Unesco criteria. In a first step, we will organize and found 'local server', such that the usual Unesco requirements of "localization" are fulfilled, but we already advance since this early stage, that at maturity and on the long term Unesco will have to consider to weaken its localization criteria in the case of Numerical patrimony.
The environmental database complies with the criteria **mixed properties Heritage ** as those which satisfy part or the whole of the definitions of both cultural and natural heritage (Unesco Guide Chap1, p.24. First, it complies notably with the status of monument: "works of monumental sculpture and painting, elements features, or structures of an archaeological nature, inscriptions, and combinations of which are of Outstanding Universal Value from the point of view of history, art or science." (Unesco Guide Chap1, p.20. Second, it complies with natural Heritage defined by the World Heritage Convention as: natural features consisting of physical and biological formations or groups of such formations, which are of Outstanding Universal Value from the aesthetic or scientific point of view. Or a geological and physiographical formations and precisely delineated areas which constitute the habitat of threatened species of animals and plants of Outstanding Universal Value from the point of view of science or conservation.
Once accepted that environmental data constitute a monument of scientific knowledge and a unique record of our natural ecosystem (of Outstanding Universal Value), the environmental database, satisfy obviously not just one, but the criterion i, ii, iii, iv, vi, viii, and ix (Unesco Guide Chap1, p.34).
The boundaries of the environmental data backup are given in the three or four domains:
- Climate data backup: the backup realized by azimuth, the climate mirror, and other databases specialized in climate recordings to be defined.
- Biodiversity data backup: backup of biodiversity databases to be defined.
- Cultural data backup: backup of indigenous oral and written traditions databases to be defined.
- Health data backup: (?) backup of epidemiological, genetic and epigenetic databases to be defined.
Proposed deadline for the submission of the project (two years of preparation is advised): december 2018
- Draft of the project
E-team to hold the project:
Preparing a World Heritage nomination usually requires a team approach because of the complexity of the task, the range of key stakeholders, and the range of expertise required. (Chap 2.2 ). For the moment, the kernel team of voluntaries is:
- John Baez - Mathematician with interests in Physic, Biology and Ecology - Founder of Azymuth project.
- Émilie Barrucand - Anthropologist - Founder of Wayanga Association
- Pierre Baudot - Biologist with interest in mathematic
- Paul Bourgine - Economist - Founder of Complex System Institute Paris and of CS-DC Uniwin UNESCO
- Pierre Collet - Bio-Informatician - Co-ordinator of CS-DC Uniwin UNESCO
- David Tanzer Software developer - Co-ordinator of Azymuth project.
Unesco suggests some bigger team to be involved, and we ask for expert or institutional volunteers to contact us.
Institutional partnership: Unesco suggests that official institutions like universities, and research autorities to be involved in the project. Each member of the e-team may ask for his own research institution to be an official partner of the project or propose other pertinent one. The Complex System-Digital Campus Unitwin Unesco is by default one partner that already involves more than a hundred of scientific institutions around the world. (TBA: Global Pulse ? Strasbourg University?, Aix-Marseille University?, IRD? CNRS, Inserm? ... )
State party partnership: Unesco suggests that state party to be involved in the project (Chap 2.2 ), and we will ask for the support of French government who already took position on the topic (invitation from the President Macron to researchers, Official talk of Mr. Macron after US left Paris agreement "France will not give up the fight" "Make Our Planet Great Again" 01/06/2017), and notably the French Ministry of environment who following the request of the Minister Ségoléne Royal is already studying the support.
Scientific expert committee for archivs
The e-team designate one expert committee per domain of archives inside the e-team (listed above). Backuping databases: The usual peer-reviewing scientific process appears as the most efficient criteria for scientific illegibility of a backup. In consequence, the data archives are managed by a scientific committee that verifies the eligibility to the back-up on the only base of the existing peer-reviewed process without substituting to it. The committees only ensures that the data basis that are submitted for a backup are issued from a peer-reviewed scientific process, that its topic fit to the archives domain and that the storing capacity allows such a backup. Researchers that submit databases is responsible for the verification that its submission complies the open source requirement. In further development, the scientific legibility review of the committee could get involved in case of submission of non peer-reviewed databases from private companies such as energy or telecommunication companies.
Accessing to the databases: by default, all the archives are available and open freely to consultation and downloading. On special request, the database can be restricted to a specific set of users.
Opening to modelisation and organizing Challenges and Benchmarks: The databases will be made available to data challenges (example of challenge : Predicting next year local or global majors environmental events). The project will develop partnership with major open data contest, non-exhaustively Data Challenges (Gilles Wainrib and Stephane Mallat), CS-DC olympiades, and United Global Pulse Data for Climate Action.
Archives of the databases - Computing resources
In a first implementation of the project, the backup will be organized on localized and dedicated servers, and the possibility of distributed (cloud) storing will be proposed to Unesco on the long term. The CS-DC, via Pierre Collet, the Icube lab and Strasboug University and thanks to private donation to the Strasbourg University Fondation UNISTRA, already provided 40 Terabytes in RAID 5 with FTP access. We will propose to settle the first backups on this site using this infrastructure and process. Interface for the depository: a web interface for the depository will be developed, this may be achieved in partnership and using the efforts already provided by the climate mirror project (To be done).
The main costs of the project concerns data servers, their maintenance with a high rate transfer availability, and the development of an interface for the depository:
- Private donation campaign : the donation of individuals and private companies are welcome. They can be easily (few minutes) made via the Strasbourg University Fondation UNISTRA, by filling this DONATION FORM with specifyed "CS-DC Unesco UniTwin" as the recipient of the donation.
- Partners funding Unesco and other institutional partners will be asked to participate on the financial aspects (TBA). French government proposed also to support this project, and is currently studying the proposition (to ask again cf. the request of the Minister Ségoléne Royal).
Invited & Keynote Speakers
Guest Honorary speaker
- Jean-Michel Bismut (professeur à l’Université Paris-Sud (Orsay), member of Académie des Sciences)
Jean-Michel Bismut was born in 1948 in Lisbon (Portugal). He studied at Ecole Polytechnique in 1967-1969, and he received his Doctorat d’Etat from Université Paris VI in 1973. He became a professor of Mathematics in Orsay in 1981. He was a plenary speaker at ICM-Berlin 1998, and a vice-president of International Mathematical Union from 2002 to 2006. His research has been devoted to stochastic control, to the Malliavin calculus, to index theory, and its connections with spectral theory and number theory.
The hypoelliptic Laplacian
If X is a Riemannian manifold, the hypoelliptic Laplacian is a family of hypoelliptic operators acting on X , the total space of the tangent bundle of X , that interpolates between the ordinary Laplacian and the geodesioc ﬂow. The probabilistic counterpart is an interpolation between Brownian motion and geodesics.
In the talk, I will explain the construction of the hypoelliptic Laplacian, and describe some of its properties.
J.-M. Bismut. The hypoelliptic Laplacian on the cotangent bundle. J. Amer. Math. Soc., 18(2):379-476 (electronic), 2005.
J.-M. Bismut and G. Lebeau. The hypoelliptic Laplacian and Ray-Singer metrics, volume 167 of Annals of Mathematics Studies. Princeton University Press, Princeton, NJ, 2008.
J.-M. Bismut. Loop spaces and the hypoelliptic Laplacian. Comm. Pure Appl. Math., 61(4):559-593, 2008.
J.-M. Bismut. Hypoelliptic Laplacian and orbital integrals, volume 177 of Annals of Mathematics Studies. Princeton University Press, Princeton, NJ, 2011.
Invited Honorary speaker
- Daniel Bennequin (Université Paris 7 - Institut Mathématique de Jussieu)
Born 3 January 1952. Graduate from Ecole Normale Supérieure. PHD in 1982 with Alain Chenciner at Paris VII. Then Professor at Strasbourg University. Today Professor at Paris-Diderot University, and member of the IMJ. During the 1980’s he was initiator of contact topology with Y.Eliashberg. During the 1990’s, he worked on integrable systems and geometry of Mathematical Physics. Since 2000 he has been working in Neurosciences (mainly with A.Berthoz, C-d-F, and T.Flash, Weizmann Institute); he made contributions to the study of human movements duration, vestibular informatin flow and gaze functions during locomotion. His most recent publications are on information topology (with P.Baudot), psychic pain (with M.Bompard-Porte) and labyrinths (with R.David et al.).
Geometry and Vestibular Information
Every complex living entities, as plants, insects or vertebrates, possess visuo-vestibular systems which sense their own motion in space and are crucial for controling volontary movements and for understanding space. We will show how the Galilée group guides the visuo-vestibular information flows. Differential Geometry permits to understand the particular forms of the end vestibular organs, that are situated in the inner ear of mammals and birds, from a principle of energy minimization and information maximization. These forms correspond to the surfaces of divisors of real (resp. imaginary) twisted curves, for the epithelia which sense linear accelerations (resp. rotations) of the head. The Hodge-DeRham theory, applied to the labyrinths volume of vertebrates, permits to explain how a complex fluid movement is transformed in six solutions of ordinary second order differential equations, for registering the head rotations in space. Combined with an original and delicate method of analysis of the membranous tissues, invented by Romain David, this allows for the first time, to describe the precise relation between the structure and the function of the labyrinth.
R.David, A.Stoessel, A.Berthoz, F.Spoor, D.Bennequin, “Assessing morphology and function of the semicircular duct system: introducing new in situ visualization and software toolbox ”, Scientific Reports, 2016.
P.Marianelli, A.Berthoz, D.Bennequin, “Crista egregia: a geometrical model of the crista ampullaris, a sensory surface that detects head rotations”, Biological Cybernetics, 2015.
M.Dimiccoli, B.Girad, A.Berthoz, D.Bennequin, “Striola Magica. A functional explanation of the otolith geometry”, J Comput Neurosciences, 2013.
D.Bennequin, A.Berthoz, “Non-linear Galilean receptive fields”, IEEE Med Biol Soc, Boston, 2011
- Alain Trouvé (ENS Paris-Saclay, CMLA Department)
Alain Trouvé, bachelor’s degree from Ecole Normale Supérieure Ulm, a doctor of the University of Orsay, began his career as “agrégé préparateur” at the ENS Ulm before becoming a professor at the University of Paris13 (1996) and then at ENS Cachan (2003). Alain Trouvé is currently Professor at the Center of Mathematics and Their Application (CMLA) at ENS Paris-Saclay. He did his Ph.D. in Stochastic Optimization and Bayesian Image Analysis under the supervision of Robert Azencott. His main research interests are computational vision and shape analysis with a particular emphasis on the use of Riemannian geometry and infinite dimensional group actions driven by applications in computational anatomy and medical imaging.
Hamiltonian modeling for shape evolution and Statistical modeling of shapes variability
In his book "Growth and Forms", first published in 1917, d’Arcy Thompson, a Scottish naturalist and mathematician, develops his theory of transformations, whose central idea is the morphological comparison of anatomies through groups of transformations of Space that act on it. This idea, a century later, remains at the heart of contemporary geometric approaches of quantitative comparison of forms but in a very different mathematical and technological context. In this talk, we present the ideas and techniques that underlie the "diffeomorphometric" approach developed in the context of computational anatomy, its links with infinite dimensional Riemannian geometry, the theory of control And Hamiltonian systems, but also the dimension reduction tools that underlie the algorithms used in the analysis of sub-varieties and make them effective. We will also present new prospects for extension on the geometric-functional objects that combine geometric and functional information and pose new and numerous challenges.
Devilliers, L., Allassonnière, S., Trouvé, A., & Pennec, X. (2017). Inconsistency of Template Estimation by Minimizing of the Variance/Pre-Variance in the Quotient Space. Entropy, 19(6), 288.
Lee, S., Charon, N., Charlier, B., Popuri, K., Lebed, E., Sarunic, M. V., ... & Beg, M. F. (2017). Atlas-based shape analysis and classification of retinal optical coherence tomography images using the functional shape (fshape) framework. Medical image analysis, 35, 570-581.
D. Tward, M. Miller, A. Trouvé and L. Younes, "Parametric Surface Diffeomorphometry for Low Dimensional Embeddings of Dense Segmentations and Imagery," in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 6, pp. 1195-1208, June 1 2017.
Charlier, B., Charon, N., & Trouvé, A. (2017). The fshape framework for the variability analysis of functional shapes. Foundations of Computational Mathematics, 17(2), 287-357.
M. Miller, L. Younes and A. Trouve. "Hamiltonian Systems in Computational Anatomy : 100 Years since D'Arcy Thompson". Annual Review of Biomedical Engineering 17 , 2015
S. Arguillere, E. Trélat, A. Trouvé, L. Younes. "Shape deformation analysis from the optimal control point of view". Journal de Mathématiques Pures et Appliquées 104 (1): 139-178, 2015
- Mark Girolami (Imperial College London - Department of Mathematics)
Mark Girolami holds a Chair in Statistics in the Department of Mathematics of Imperial College London. He is an EPSRC Established Career Research Fellow (2012 - 2017) and previously an EPSRC Advanced Research Fellow (2007 - 2012). He is the Director of the Alan Turing Institute-Lloyds Register Foundation Programme on Data Centric Engineering and in 2011 was elected to the Fellowship of the Royal Society of Edinburgh when he was also awarded a Royal Society Wolfson Research Merit Award. He was one of the founding Executive Directors of the Alan Turing Institute for Data Science from 2015 to 2016. He has been nominated by the IMS to deliver a Medallion Lecture at JSM 2017 and has been invited to give a Forum Lecture at the European Meeting of Statisticians 2017. His paper on Riemann manifold Langevin and Hamiltonian Monte Carlo Methods was publicly read before the Royal Statistical Society and received the largest number of contributed discussions for any paper in the entire history of the society, discussants included Sir D.R. Cox and C.R. Rao.
Riemann Manifold Langevin and Hamiltonian Monte Carlo Methods
The talk considers Metropolis adjusted Langevin and Hamiltonian Monte Carlo sampling methods defined on the Riemann manifold to resolve the shortcomings of existing Monte Carlo algorithms when sampling from target densities that may be high dimensional and exhibit strong correlations. The methods provide fully automated adaptation mechanisms that circumvent the costly pilot runs that are required to tune proposal densities for Metropolis–Hastings or indeed Hamiltonian Monte Carlo and Metropolis adjusted Langevin algorithms. This allows for highly efficient sampling even in very high dimensions where different scalings may be required for the transient and stationary phases of the Markov chain. The methodology proposed exploits the Riemann geometry of the parameter space of statistical models and thus automatically adapts to the local structure when simulating paths across this manifold, providing highly efficient convergence and exploration of the target density. The performance of these Riemann manifold Monte Carlo methods is rigorously assessed by performing inference on logistic regression models, log-Gaussian Cox point processes, stochastic volatility models and Bayesian estimation of dynamic systems described by non-linear differential equations. Substantial improvements in the time-normalized effective sample size are reported when compared with alternative sampling approaches.
Mike Betancourt, Simon Byrne, Sam Livingstone, and Mark Girolami (2016) "The Geometric Foundations of Hamiltonian Monte Carlo" to appear Bernoulli
Oates, C., Girolami, M. and Chopin, N. Control Functionals for Monte Carlo Integration. To appear Journal of Royal Statistical Society - Series B, 2017.
T.House, A.Ford, S.Lan, S. Bilson, E. Buckingham-Jeffery, M.A.Girolami. (August 2016) Bayesian Uncertainty Quantification for Transmissability of Influenza, Norovirus, and Ebola using Information Geometry. Journal of the Royal Society Interface, DOI: 10.1098/rsif.2016.0279
Shiwei Lan, Tan Bui-Thanh, Mike Christie, Mark Girolami (2016). Emulation of Higher-Order Tensors in Manifold Monte Carlo Methods for Bayesian Inverse Problems, Journal of Computational Physics Vol. 308, 81 - 101.
Bui, T. and Girolami, M. "Solving Large-Scale PDE-constrained Bayesian Inverse problems with Riemann Manifold Hamiltonian Monte Carlo", Inverse Problems, 30, 114014, doi:10.1088/0266-5611/30/11/114014.
T Xifara, C Sherlock, S Livingstone, S Byrne, M Girolami. Langevin diffusions and the Metropolis-adjusted Langevin algorithm. Statistics & Probability Letters 91, 14-19, 2014.
S Byrne, M Girolami. Geodesic Monte Carlo on Embedded Manifolds. Scandanavian Journal of Statistics, (with discussion) 40, 825 – 845, 2013.
Girolami, M., Calderhead, B., Riemann Manifold Langevin and Hamiltonian Monte Carlo Methods (with discussion), Journal of the Royal Statistical Society – Series B, 73(2), 123 - 214, 2011.
- Barbara Tumpach (Lille University/ Painlevé Laboratory)
Alice Barbara Tumpach is an Associate Professor in Mathematics (University Lille 1, France) and member of the Laboratoire Painlevé (Lille 1/CNRS UMR 8524), since 2007. She received a Ph.D degree in Mathematics in 2005 at the Ecole Polytechnique, Palaiseau, France. She spent two years at the Ecole Polytechnique Fédérale de Lausanne as a Post-Doc, and two years at the Pauli Institut in Vienna, Austria, as an invited researcher. Her research interests lie in the area of infinite-dimensional Geometry, Lie Groups and Functional Analysis. She gives Master courses on Lie groups and organizes conferences on infinite-dimensional geometry for the Federation of Mathematical Research of Nord-Pas-Calais, France. She also acts in videos for Exo7, available on youtube, where she explains basic notions of Linear Algebra.
Riemannian metrics on shape spaces of curves and surfaces
The aim of the talk is to give an overview of geometric tools used in Shape Analysis. We will see that we can interpret the Shape space of (unparameterized) curves (or surfaces) either as a quotient space or as a section of the Preshape space of parameterized curves (or surfaces). Starting from a diffeomorphism-invariant Riemannian metric on Preshape space, these two different interpretations lead to different Riemannian metrics on Shape space. Another possibility is to start with a degenerate Riemannian metric on Preshape space, with degeneracy along the orbits of the diffeomorphism group. This leads to a framework where the length of a path of curves (or surfaces) does not depend on the parameterizations of the curves (or surfaces) along the path. Of course the choice of the metrics has to be motivated either from the applications or from their mathematical behaviour. We will compare some natural metrics used in the litterature.
A.B.Tumpach and S. Preston, Quotient elastic metrics on the manifold of arc-length parameterized plane curves, to appear in Journal of Geometric Mechanics.
A.B.Tumpach, Gauge invariance of degenerate Riemannian metrics, Notices of AMS, April 2016.
A.B.Tumpach, H. Drira, M. Daoudi, A. Srivastava, Gauge invariant Framework for shape analysis of surfaces, IEEE TPAMI, vol 37, 2015.
A.B.Tumpach, Infinite-dimensional hyperkähler manifolds associated with Hermitian-symmetric affine coadjoint orbits, Annales de l'Institut Fourier, Tome 59, 2009.
A.B.Tumpach, Classification of infinite-dimensional Hermitian-symmetric affine coadjoint orbits, Forum Mathematicum 21:3, 2009.
D. Beltita, T. Ratiu, A.B. Tumpach, The restricted Grassmannian, Banach Lie-Poisson spaces, and coadjoint orbits, Journal of Functional Analysis 247, 2007.
A.B.Tumpach, Hyperkähler structures and infinite-dimensional Grassmannians, Journal of Functional Analysis 243, 2007.
- Jean-Michel Bismut (professeur à l’Université Paris-Sud (Orsay), member of Académie des Sciences)
Preliminary program - Accepted papers
OFFICIAL PROGRAM WEBPAGE
Session “Statistics on non-linear data” (X. Pennec/S. Sommer)
63 Line Kühnel and Stefan Sommer. Stochastic Development Regression using Method of Moments
49 Benjamin Eltzner and Stephan Huckemann. Bootstrapping Descriptors for Non-Euclidean Data
62 Xavier Pennec. Sample-limited Lp Barycentric Subspace Analysis on Constant Curvature Spaces
84 Maxime Louis, Alexandre Bône, Benjamin Charlier and Stanley Durrleman. Parallel transport in shape analysis : a scalable numerical scheme for Riemannian manifolds
53 Georgios Arvanitidis, Lars Kai Hansen and Søren Hauberg. Maximum likelihood estimation of Riemannian metrics from Euclidean data
Session “Shape Space” (S. Allasonnière/S. Durrleman/A. Trouvé)
44 Alexander Schmeding, Elena Celledoni, Sølve Eidnes and Markus Eslitzbichler. Shape Analysis on Lie groups and homogeneous spaces
79 Alice Le Brigant, Marc Arnaudon and Frederic Barbaresco. Optimal matching between curves in a manifold
97 Boris Khesin, Gerard Misiolek and Klas Modin. Newton's Equation on Diffeomorphisms and Densities
5 Kathrin Welker. Optimization in the Space of Smooth Shapes
86 Pierre Roussillon and Joan Alexis Glaunès. Surface Matching Using Normal Cycles
Session “Optimal Transport & Applications I” (Q. Merigot/J. Bigot/B. Maury)
18 Elsa Cazelles, Jérémie Bigot and Nicolas Papadakis. Regularization of Barycenters in the Wasserstein Space
22 Giovanni Conforti and Michele Pavon. Extremal curves in Wasserstein space
57 Bruno Galerne, Arthur Leclaire and Julien Rabin. Semi-Discrete Optimal Transport in Patch Space for Enriching Gaussian Textures
66 Yunan Yang and Bjorn Engquist. Analysis of Optimal Transport Related Misfit Functions in Seismic Imaging
Session “Optimal Transport & Applications II” (J.F. Marcotorchino/A. Galichon)
20 Ryo Karakida and Shun-Ichi Amari. Information Geometry of Wasserstein Divergence
74 Damien Nogues. Anomaly detection in network traffic with a relationnal clustering criterion
54 Martin Bauer, Sarang Joshi and Klas Modin. Diffeomorphic random sampling using optimal information transport
80 Olivier Rioul. Optimal Transport to Rényi Entropies
Session “Statistical Manifold & Hessian Information Geometry” (M. Boyom/A. Matsuzoe/ Hassan Shahid)
83 Masayuki Henmi. Statistical Manifolds Admitting Torsion, Pre-contrast Functions and Estimating Functions
46 Michel Nguiffo Boyom, Mohd Aquib, Mohammad Hasan Shahid and Mohammed Jamali. Generalized Wintegen type inequality for Lagrangian submanifolds in holomorphic Statistical space forms
88 Ahmed Zeglaoui and Michel Nguiffo Boyom. The functor of Amari and Riemannian dynamics
26 Hitoshi Furuhata. Sasakian statistical manifolds II
90 Sergey Grigorian and Jun Zhang. (Para-)Holomorphic Connections for Information Geometry
111 Hideyuki Ishi. Matrix realization of a homogeneous Hessian domain
Session “Monotone Embedding in Information Geometry” (J. Zhang/ J. Naudts)
91 Jun Zhang and Jan Naudts. Information Geometry Under Monotone Embedding. Part I: Divergence Functions
68 Jan Naudts and Jun Zhang. Information Geometry Under Monotone Embedding. Part II: Geometry
67 Hiroshi Matsuzoe, Antonio M. Scarfone and Tatsuaki Wada. A sequential structure of statistical manifolds on deformed exponential family
42 Luiza Andrade, Rui Vigelis, Leidmar Vieira and Charles Cavalcante. Normalization and varphi-function: definition and consequences
45 Luigi Montrucchio and Giovanni Pistone. Deformed exponential bundle: the linear growth case
92 Atsumi Ohara. On affine immersions of the probability simplex and their conformal flattening
Session “Information Structure in Neuroscience” (P. Baudot/D. Bennequin/S. Roy)
110 Trang-Anh Nghiem, Olivier Marre, Alain Destxhe and Ulisse Ferrari. Pairwise Ising model analysis of human cortical neurons recordings
17 Jeong Joon Park, Ronnel Boettcher, Andrew Zhao, Alex Mun, Kevin Yuh, Vibhor Kumar and Matilde Marcolli. Prevalence and recoverability of syntactic parameters in sparse distributed memories
73 Majd Hawasly, Florian T. Pokorny and Subramanian Ramamoorthy. Multi-Scale Activity Estimation with Spatial Abstractions
21 Guido Montufar and Johannes Rauh. Geometry of Policy Improvement
29 Chenxi Li, Zelin Shi, Yunpeng Liu and Tianci Liu. Joint geometric and photometric visual tracking based on Lie group
Session “Geometric Robotics & Tracking“ (S. Bonnabel/A. Barrau)
102 Marion Pilte, Silvere Bonnabel and Frederic Barbaresco. Drone tracking with an IEKF and an innovative UKF
65 Silvere Bonnabel and Jean-Jacques Slotine. Particle observers for contracting dynamical systems
89 James Forbes and David Evan Zlotnik. Sigma Point Kalman Filtering on Matrix Lie Groups
10 Ivan Polekhin. A topological view on forced oscillations and control of an inverted pendulum
55 Pascal Morin, Alexandre Eudes and Glauco Scandaroli. Uniform observability of linear time-varying systems and application to robotics problems
78 Ioannis Sarras and Philippe Martin. Global exponential attitude and gyro bias estimation from vector measurements
Session “Geometric Mechanics & Robotics” (G. de Saxcé/J. Bensoam/ J. Lerbet)
51 Jean Lerbet, Noel Challamel, François Nicot and Félix Darve. Geometric Degree of Non Conservativeness
11 Abdelbacet Oueslati, An Danh Nguyen and Géry de Saxcé. A symplectic minimum variational principle for dissipative dynamical systems
100 Eric Bergshoeff, Athanasios Chatzistavrakidis, Luca Romano and Jan Rosseel. Torsional Newton-Cartan Geometry
98 Thomas Hélie and Fabrice Silva. Self-oscillations of a vocal apparatus: a port-Hamiltonian formulation
23 Frédéric Hélein, Joël Bensoam and Pierre Carré. Differential Geometry applied to Acoustics. Non Linear Propagation in Reissner Beams : an integrable system?
99 Maurice de Gosson. Quantum Harmonic Analysis and the Positivity of Trace Class Operators; Applications to Quantum Mechanics
Session “Geometrical Structures of Thermodynamics” (F. Gay-Balmaz/F. Barbaresco)
95 François Gay-Balmaz and Hiroaki Yoshimura. A variational formulation for fluid dynamics with irreversible processes
113 Hiroaki Yoshimura and François Gay-Balmaz. Dirac structures in nonequilibrium thermodynamics
105 Bernhard Maschke and Arjan van der Schaft. About the definition of port variables for contact Hamiltonian systems
61 Vitaly Mikheyev. Method of orbits of co-associated representation in thermodynamics of the lie noncompact groups
38 Frederic Barbaresco. Poly-Symplectic Model of Higher Order Souriau Lie Groups Thermodynamics for Small Data Analytics
112 Francesco Becattini. Thermodynamic equilibrium in relativity: Killing vectors and Lie derivatives
Session “Probability on Riemannian Manifolds” (M. Arnaudon/A.-B. Cruzeiro)
108 Yann Ollivier and Gaétan Marceau Caron. Natural Langevin Dynamics for Neural Networks
71 Birgit H. Roensch and Wolfgang Stummer. 3D insights to some divergences for robust statistics and machine learning
56 Jean-Claude Zambrini and Marc Arnaudon. A stochastic look at geodesics on the sphere
109 Matthias Glock and Thomas Hotz. Constructing Universal, Non-Asymptotic Confidence Sets for Intrinsic Means on the Circle
1 Marco Frasca. Noncommutative geometry and stochastic processes
47 Florio Maria Ciaglia, Fabio Di Cosmo and Giuseppe Marmo. Hamilton-Jacobi theory and Information Geometry
Session “Divergence Geometry” (M. Broniatowski/I. Csiszar)
60 Emmanuelle Gautherat, Patrice Bertail and Hugo Harari-Kermadec. Empirical Phi star Divergence Minimizers for Hadamard Differentiable Functionals
6 Minh Ha Quang. Log-Determinant Divergences Between Positive Definite Hilbert-Schmidt Operators
85 Wolfgang Stummer and Anna-Lena Kißlinger. Some new flexibilizations of Bregman divergences and their asymptotics
24 Zuzana Krajcovicova, Pedro Pablo Perez Velasco and Carlos Vazquez. Quantification of Model Risk: Data Uncertainty
96 Eric Grivel and Léo Legrand. Process comparison combining signal power ratio and Jeffrey's divergence between unit-power signals
Session “Non-parametric Information Geometry” (N. Ay/J. Armstrong)
9 John Armstrong and Damiano Brigo. Ito Stochastic Differential Equations as 2-Jets
15 Van Le, Juergen Jost and Lorenz Schwachhoefer. The Cramer-Rao inequality on singular statistical models
27 Shinto Eguchi and Katsuhiro Omae. Information Geometry of Predictor Functions in a Regression Model
41 Giovanni Pistone. Translations in the exponential Orlicz space with Gaussian weight
58 Marina Santacroce, Paola Siri and Barbara Trivellato. On Mixture and Exponential Connection by Open Arcs
Session “Optimization on Manifold” (P.A. Absil/R. Sepulchre)
50 Benjamin Eltzner and Stephan Huckemann. Applying Backward Nested Subspace Inference
75 Pierre-Yves Gousenbourger, Laurent Jacques and P.-A. Absil. Fast method to fit a C1 piecewise-Bézier function to manifold-valued data points: how suboptimal is the curve obtained on the sphere S2?
76 Ronny Bergmann and Daniel Tenbrinck. Nonlocal Inpainting of Manifold-valued Data on Finite Weighted Graphs
82 Cyrus Mostajeran and Rodolphe Sepulchre. Affine-invariant orders on the set of positive-definite matrices
104 Geert Verdoolaege. Geodesic Least Squares Regression on the Gaussian Manifold: Baryonic Tully-Fisher Scaling in Disk Galaxies
Session “Computational Information Geometry” (F. Nielsen/O. Schwander)
2 Salem Said and Yannick Berthoumieu. Warped metrics for location-scale models
12 Frank Nielsen and Richard Nock. Bregman divergences from comparative convexity
32 Philippe Regnault, Valérie Girardin and Loïck Lhote. Weighted Closed Form Expressions Based on Escort Distributions for Rényi Entropy Rates of Markov Chains.
106 Remy Boyer and Frank Nielsen. On the Error Exponent of a Random Tensor with Orthonormal Factor Matrices
33 Tomonari Sei. Coordinate-wise transformation and Stein-type densities
Session “Probability Density Estimation” (S. Said/E. Chevallier)
14 Paolo Zanini, Salem Said, Yannick Berthoumieu, Marco Congedo and Christian Jutten. Riemannian Online Algorithms for Estimating Mixture Model Parameters
52 Florent Chatelain, Nicolas Le Bihan and Jonathan Manton. Density estimation for Compound Cox processes on hyperspheres
69 Hatem Hajri, Salem Said and Yannick Berthoumieu. Maximum likelihood estimators on manifolds
107 Stephane Puechmorel and Florence Nicol. Von Mises-like probability density functions on surfaces
13 Salem Said, Nicolas Le Bihan and Jonathan Manton. Riemannian Gaussian distributions on the space of positive-definite quaternion matrices
37 Emmanuel Chevallier. A family of anisotropic distributions on the hyperbolic space
Session “Geometry of Tensor-Valued Data” (J. Angulo/Y. Berthoumieu/ G. Verdoolaege/ A.M. Djafari)
77 Aleksei Shestov and Mikhail Kumskov. A Riemannian Approach to Blob Detection in Manifold-Valued Images
40 Ioana Ilea, Lionel Bombrun, Salem Said and Yannick Berthoumieu. Co-occurrence matrix of covariance matrices: a novel coding model for the classification of texture images
19 Reiner Lenz. Positive Signal Spaces and the Mehler-Fock Transform
81 Simon Apers, Alain Sarlette and Francesco Ticozzi. Bounding the convergence time of local probabilistic evolution
31 Estelle Massart and Sylvain Chevallier. Inductive means and sequences applied to online filtering and classification of EEG
Session “Geodesic Methods with Constraints” (J.-M. Mirebeau/L. Cohen)
35 Erik Bekkers, Remco Duits, Alexey Mashtakov and Yuri Sachkov. Vessel Tracking via Sub-Riemannian Geodesics on Projective Line Bundle
43 Da Chen and Laurent Cohen. Anisotropic Edge-based Balloon Eikonal Active Contours
94 Jean-Marie Mirebeau and Johann Dreo. Automatic differentiation of non-holonomic fast marching for computing most threatening trajectories under sensors surveillance
30 Rui Vigelis, Luiza Felix and Charles Cavalcante. On the existence of paths connecting probability distributions
48 Michel Nguiffo Boyom, Aliya Naaz Siddiqui, Wan Ainun Mior Othman and Mohammad Hasan Shahid. Classification Of Totally Umbilical CR-Statistical Submanifolds In Holomorphic Statistical Manifolds With Constant Holomorphic Curvature
Session "Applications of Distance Geometry" (A. Mucherino/D. Gonçalves)
16 Antonio Mucherino and Douglas Gonçalves. An Approach to Dynamical Distance Geometry
36 Claudia D'Ambrosio and Leo Liberti. Distance geometry in linearizable norms
59 Philippe Jacquet and Dalia-Georgiana Herculea. Self-similar Geometry for Ad-Hoc Wireless Networks: Hyperfractals
4 Radmila Pribic. Information Distances in Stochastic Resolution Analysis
93 Frank Nielsen, Ke Sun and Stéphane Marchand-Maillet. k-Means Clustering with Hölder divergences
39 Imsoon Jeong, Gyu Jong Kim and Young Jin Suh. Real hypersurfaces in the complex quadric with certain condition of normal Jacobi operator
We are Covestro. We are curious. We are courageous. We are colorful. We redefine chemical material solutions with game-changing products. Let us empower you to push boundaries. Join us and our 16,000 colleagues now and together we will MAKE the world a brighter place.
Head (m/f) of Advanced Analytics
YOUR TASKS AND RESPONSIBILITIES
In your new role, you will build and lead a central Advanced Analytics (Artificial Intelligence, Machine Learning, Deep Learning etc.) team that will act as a Covestro competence center and work with all functions (e.g. manufacturing, R&D, commercial) across Covestro on use cases
You will also act as the Head (m/f) of the global Covestro Advanced Analytics community
You will advise on the most suitable approaches and methods to achieve the desired business outcomes based on the analysis of large volumes of structured and unstructured data
Thereby, you will help Covestro to move from generating insights through statistical analysis (descriptive), over formulating predictions for future events (predictive) towards generating actionable intelligence (prescriptive)
You will build and maintain an external network, set up and manage collaborations with third parties
WHO YOU ARE
You hold a master ́s degree or PhD in mathematics, statistics, machine learning, scientific computing or a related discipline
Your profile shows long-term experience in data analytics preferably in a commercial environment and you are familiar with virtual organizations
Besides that you are a Team Player (m/f) and you have a very innovative driven personality
You are fascinated by new ideas and the development of Covestro
Very good networking and communication skills are among your key competencies
You are familiar with working in an multinational interdisciplinary environment
Excellent English language skills, written and spoken, round up your profile
We offer a competitive salary in an international environment as well as excellent opportunities for professional and personal development. If your background and personal experience fits this profile, please apply online via the 'Go to application page' button, submitting a cover letter, your CV and references as well as your salary expectation. Please also indicate your availability/notice period.
First questions will be answered by Ms Isabell Ramrath Phone: +49 214 6009 8246.
Reference Code: 0000010888
Covestro welcomes applications from all individuals, regardless of racial or ethnic origin, skin color, nationality, religion, philosophy, gender, age, disability, appearance or sexual identity. We are committed to treating all applicants fairly and avoiding discrimination.
Optimus Data is working on behalf of one of the world’s biggest brands. Do you want to work in a new Data Science team in their brand new open plan offices? Be part of a social and interactive company
Master or Ph.D. in computer science, statistics, operations research, engineering or related Field
Theoretical knowledge and 2+ years hands-on experience in data mining, optimisation, machine learning, NLP, image processing, intelligent agents or other specialisations of Artificial Intelligence
Proficiency with SQL, R and/or Python
Experience in programming languages like Java, C++ or Scala
Preferably experience in developing, deploying and running analytics applications
A team player who communicates effectively with colleagues from technical and business areas
Work together with internal business partners to show how Data Analytics can improve their business.
Identify the most suitable method to solve business cases with the right data and algorithms
Work together with Software Engineers to develop prototypes and bring your solution to life
Demonstrate the value of your solution by measuring performance and communicating its benefits to all stakeholders
Strive to continuously enrich our applications by improving model performance but also by deriving business recommendations from your data insights
Be creative in exploring and mining data to generate insights, discover connections and innovate new business cases
keep track of evolving technologies and methods that create new business cases or new solutions
Start date: Immediately
Perks: Amazing relocation support, Phone, Laptop, 32 days vacations, Home Office, On-site/ off-site training
If this role sounds interesting to you, then send your updated CV to cowen [at] optimussearch.com.
Entropy Young Investigator Award 2018
WEBSITE - APPLICATION
The award will consist of：
- 2500 Swiss Frances;
- the book Probability Theory: The Logic of Science by Edwin T. Jaynes;
- a commemorative plaque;
- a waiver of registration fees and a talk for the conference “Entropy
2018: From Physics to Information Sciences and Geometry”.
- The nominee has contributed outstanding research in the fields
covered by the Entropy journal, see details at
- The nominee should have received his/her PhD within the last eight
years (by 31 December 2017), and not yet hold a permanent professorship.
- The nominee should be 40 years of age or under (by 31 December 2017).
Required application documents:
- CV (including date on which the PhD degree was awarded and a list of
- Research description (up to two pages)
- Nomination letter from his/her supervisor, research director or
Please submit your application online at
by 31 December 2017. The winner will be announced by the end of
Time Schedule & Deadlines:
- Abstract Submission: 6 Oct. 2017
- Acceptance Notification: 20 Oct. 2017
- Full Paper Submission: 3 Nov. 2017
Welcome from the Chair of the 4th ECEA
You are cordially invited to participate in the 4th International Electronic Conference on Entropy and Its Applications. Building on the success of the past three events in this sequence, the conference is designed to bring together researchers working in the field, to present and discuss their recent contributions, without needing to leave the comfort of their home.
Conserved quantities such as energy, and monotonic quantities such as entropy, are fundamental to our understanding of high-dimensional dynamical systems. Since the foundations of irreversibility were laid in 19th century thermodynamics, and the conceptual analogy between statistical mechanics and information theory (both classical and quantum) was made in the 20th century, inter-related concepts of entropy have been fruitfully applied to a large and expanding list of fields, including economics, biology, computer science, operations research using maximum entropy estimates, linguistics and the social sciences.
The conference will be organized into six sessions, which reflect the inter-disciplinary nature of entropy and its applications:
- Section A
Statistical Physics: Statistical Mechanics, Irreversibility, Fluctuation Theorems, Equilibrium and Non-Equilibrium Distributions, Phase Transitions, Stochastic Systems, Chaos and Nonlinear Dynamics, Population Dynamics and Genetics
- Section B
Information and Complexity: Simplicity and Complexity, Shannon Entropy, Kullback–Leibler Divergence, Channel Capacity, Alternative Entropies, Coding, Operations Research, Forecasting, Symmetry Breaking, Similarity
- Section C
Thermodynamics in Materials: Chemical thermodynamics, the Second Law, Free Energy, Enthalpy, Self-Organization, Biochemical Networks, Nano-Scale Physics, Molecular Theory of Fluids
- Section D
Quantum Information and Foundations: Quantum Foundations, Quantum Probability and non-Kolmogorov Models, Quantum-Like Models in Cognition, Decision Making, Psychology, Social and Political Science, Economics and Finances, Bell's Inequality, Quantum Nonlocality, Contextuality, Pure and Mixed States, Superposition, De-coherence, Quantum Computing, Born Rule and Generalizations, Causality and Randomness, Role of Conscious Observer
- Section E
Machine Learning: Artificial Intelligence, Neural Networks, Cybernetics, Robotics, Man–Machine Interfaces, Bio-mimic Algorithms
- Section F
Astrophysics and Cosmology: Evolution of Stars, Gravitating Systems, Black Holes, The Universe
- Section P
Posters: Posters can be presented stand-alone, i.e., without an accompanying proceedings paper or conference presentation. Posters will be available online on this website during and after the e-conference. However, posters will not be added to the proceedings of the conference.
Accepted papers will be published in the proceedings of the conference, and selected/extended papers will be considered for publication in Entropy with a 20% discount off the APC. Entropy is an open access publication journal of MDPI in the field of entropy and information theory.
Editorial Board of the Journal Entropy
Department of Mathematics and Statistics
La Trobe University, Melbourne
- Dr. Antonio M. Scarfone (Politecnico di Torino, Torino, Italy)
- Prof. Dr. Miguel Rubi (Universitat de Barcelona, Barcelona, Spain)
- Prof. Dr. Raúl Alcaraz Martínez (University of Castilla-La Mancha, Cuenca, Spain)
- Dr. Dawn E. Holmes (University of California, Oakland, CA, USA)
- Prof. Dr. Alexander Gorban (University of Leicester, Leicester, UK)
- Dr. Andriy Olenko (La Trobe University, Melbourne, Australia)
- Prof. Dr. Leslie Glasser (Curtin University, Perth, Australia)
- Prof. Giacomo Mauro D'Ariano (University of Pavia, Pavia, Italy)
- Prof. Dr. Andrei Khrennikov (Linnaeus University, Växjö, Sweden)
- Dr. Michael J. Way (NASA Goddard Institute for Space Studies, New York, NY, USA)
Scientific Advisory Committee Members
- Prof. Anne Humeau-Heurtier (University of Angers, Angers cedex, France)
- Dr. Takuya Yamano (Kanagawa University, Kanagawa, Japan)
- Prof. Dr. Carlo Cattani (Engineering School (DEIM) University of Tuscia, Viterbo, Italy)
- Dr. Renaldas Renaldas Urniezius (Kaunas University of Technology, Lithuania)
- Dr. Robert Niven (The University of New South Wales at ADFA, Canberra, Australia)
Descriptif du poste
Dans le cadre de sa croissance, Lincoln renforce ses équipes et recrute cinq consultants Data Scientist.
Vous travaillerez au sein d’une de nos équipes de consultants et participerez aux différentes phases de la mission :
Analyse du problème posé (marché, marketing mix…)
Diagnostic de la situation et des données disponibles
Préparation de données
Modélisation prédictive et explicative (scoring…)
Mise en place d’indicateurs pertinents
Analyse des résultats (segmentation, ciblage…)
Diplômé(e) d'une formation supérieure (avec une majeure en machine learning/big data), vous justifiez d’une première d’expérience en statistiques / économétrie / data-mining
Vous maitrisez les principaux logiciels statistiques (R, SAS …), les langages SQL et programmation(C, VBA, Python…), les méthodes de machine learning (SVM, Random Forest, Kmeans…)
La connaissance des technologies Big Data (Hadoop, Pig, Hive, Aster…) est un plus
Dynamique et disponible, vous aimez communiquer et travailler en équipe
Presentation TGSI2017 - GSI - SEE - entropy journal and conferences
Download Presentation TGSI