Guidelines for proposals

The proposal shall not exceed 10 pages (not including references) on a single column page format with 12 point font. The proposal should contain the following sections: 

• Overview: A free-form description of the envisioned unit, including its mission, vision, and intended research areas; how does the proposed ELLIS unit add value to the existing infrastructure and can serve as a stepping stone towards the overarching goal of ELLIS described in the original open letter (https://ellis.eu/letter) Provide convincing evidence for excellence in machine learning-driven fields within the proposed ELLIS unit.
• Team: Planned researchers in the unit with short (no longer than 1/4 page) bios sorted into (1) existing ELLIS Fellows/Scholars/Members and (2) newly proposed ELLIS Fellows/Scholars/Members, including the director of the ELLIS unit and its organizational structure; expected number of PhD students. When preparing the team description, please see below the criteria for ELLIS units and guidelines for evaluating excellence of the team; for each proposed researcher in the team please include the year of their PhD, their h-index and link to their Google Scholar page. 
• Diversity: Diversity strategy in the ELLIS unit’s team regarding seniority, area of expertise, country of origin and underrepresented groups --e.g. gender-- of the researchers in the unit. Neither among Fellows/Scholars nor Members there should be a gender group that exceeds 85% of the unit.
• Infrastructure: Available infrastructure.
• Network: Connections with ELLIS, such as joint publications with other ELLIS Fellows/Scholars and Members, collaborations with other units, participation in ELLIS research programs, in the ELLIS PhD program, etc.
• Societal and Economic Impact: Envisioned mechanisms to promote societal and economic impact -- such as links to government and industry;
• Startups: Support for the creation of startups; 
• Dissemination: Organization of outreach events; and 
• ELLIS criteria: An explanation of how the criteria for ELLIS units described below are met.  

Criteria for ELLIS units

The creation of an ELLIS unit can leverage existing structures and funding or be created from scratch. In all cases, it requires a written agreement signed by the funding sources and relevant host institutions to commit for at least 5 years to the following tenets and processes:

• The team of the ELLIS unit consists of a director and researchers. In case there are several institutions involved it is possible to nominate one co-director for each participating institution but in general less is better. The directors of the accepted units will have the opportunity to nominate researchers from their units for evaluation to become ELLIS Fellows/Scholars. Please see the evaluation guidelines below for additional details on the expected level of excellence. Note that it is not necessary for someone to become an ELLIS Fellow/Scholar in order to be included as a researcher in the ELLIS unit. It is up to the unit director to select the researchers in the unit as long as they are at least ELLIS Members.  
• To foster the exchange of ideas and strengthen the ties in the network, ELLIS units are expected to host researchers of the ELLIS network (e.g. ELLIS Members, Fellows/Scholars, postdocs and PhD students). Researchers from the ELLIS units are encouraged to visit other ELLIS units.
• ELLIS units commit to joint PR and branding with the ELLIS society. Press releases will need to be coordinated with the central ELLIS press office.
• The director/head of each ELLIS unit will join the ELLIS Unit Directors Committee, with a solid commitment to serve as a reviewer for ELLIS and a visible engagement in supporting national and European policymakers.
• Impact:
  Machine learning is the key technology that drives a revolution of the economy and society. A crucial goal of ELLIS is to shape this process by generating innovation and impact on society. Traditionally, university professors in Europe have a very high teaching load, often much higher than 4 hours per week (the common standard in the US) such that their impact is overly constrained to classroom teaching of students. Additionally, they typically face strong administrative barriers preventing industrial impact. ELLIS units need to have a clear dedication to overcome these limitations. The researchers in ELLIS units should have maximal freedom to choose how to invest their time into high-impact projects including the spin-off or start-ups, secondary placements in industry, memberships on advisory boards, etc.
• Budget:
  To ensure a certain level of homogeneity across units, the salaries of researchers (ELLIS Fellows/Scholars/Members) of the ELLIS unit should amount to a total of at least 500k euro/year. Each unit is free to choose how to split such a budget into different positions (e.g. senior vs junior) and the number of such positions.
  The total expected budget of each ELLIS unit is at least 1 Mio euro/year (excluding salaries) including funds from industry and funding agencies. Each unit will have autonomy in its allocation of the budget, except that a contribution to the ELLIS network is required as described below .
• Contribution towards running ELLIS:
  - It is expected that from the above budget, 300k euro/year will be devoted to measures supporting the ELLIS network, such as funding ELLIS coordinator positions, hosting visitors, organizing ELLIS events or financing PhD students in the ELLIS PhD program. Financing ELLIS PhD students should represent at most half of the required 300k euro/year.
  - It is expected that each unit will take the responsibility for at least one organizational/administrative task in the ELLIS network offering coordinators reporting to the ELLIS society, or, alternatively,  will provide a direct financial contribution (current estimate is 50k Eur / year) towards organizational/administrative costs to ensure a fair distribution of the burden across the ELLIS network. Examples of these types of tasks include (but are not limited to): chairing and being responsible for the Fellow/Scholar reviewing process or the ELLIS Units reviewing process, leading or contributing to an EU network proposal (such as elise-ai.eu), leading the PR efforts of the network or hosting/maintaining the ELLIS website. These tasks will be centrally coordinated. ELLIS will take into account the preferences of each ELLIS unit. However, such preferences might not be able to be satisfied, depending on existing needs.
  - Researchers in the unit need to commit to reviewing ELLIS proposals for units, programs, Fellows/Scholars, PhD candidates or for other calls organized by the ELLIS Network. This commitment is particularly important for the directors and Fellows in the unit as they are needed for the ELLIS Fellow evaluation process.

Guidelines for evaluation

Excellence

The most important criterion is the excellence of the team in machine learning-driven fields. Nurturing, retaining and attracting top talent is the key focus for ELLIS. At the time of submission, each unit proposal should have at least one Fellow or someone clearly on the Fellow level. The proposal should list additional (candidates for) Scholars and Fellows and a convincing strategy on how to attract additional researchers on the Scholar/Fellow level.  The rest of the proposed researchers in the unit should be at least ELLIS Members. There is a fast and light-weight procedure to become an ELLIS Member, which should be completed by the proposed researchers in the unit before the unit proposal is submitted. Please see ellis.eu/members. Note that proposals might have incomplete teams if they plan to use the ELLIS unit to attract and hire new talent. In this case, only the scientific excellence of existing researchers should be evaluated. New researchers will be evaluated later on.

For Fellows and Scholars the level of expected excellence would be similar to what ERC grants or prestigious research institutions (both private and public) would look for. If the unit is accepted, candidates to become ELLIS Fellows and Scholars in the unit are nominated by the Unit director for evaluation. This is a separate evaluation process from evaluating the unit but we provide below the guidelines for evaluating Fellows and Scholars as a reference that can be also used by the unit evaluators. 

Overview of the proposal (Guideline questions)

• Does the ELLIS unit proposal correctly represent the research strengths of the team?
• How relevant are these strengths for the machine learning-driven AI community?
• Do they have competitive/impactful publications on the topics?  Highly cited papers, relevant patents?
• Are they in the PC and organizing committee of relevant conferences?
• Have they won competitive best paper awards? ACM/IEEE distinctions etc...? Memberships in academies of science, etc; --Have they been selected to give keynotes/tutorials in top-tier international conferences and/or international advanced schools?
• Have they worked in relevant industries and/or created any companies? Have they been advisors to governments, the European Commission, the WEF?
• [For senior researchers in the team] Have they mentored students/junior scientists who have become prominent contributors to the machine learning field?

Organizational structure (Guideline questions)

• How well suited is the director of the proposed unit to ensure its sustainability and success?
• Is there a balance between the number of junior and senior researchers?
• Does the unit’s governance give a voice to junior researchers?

Diversity Strategy (Guideline questions)

• How diverse is the proposed team?
• How comprehensive and ambitious is the diversity strategy?
• Are there concrete actions to achieve diversity?

Infrastructure (Guideline questions)

• Does the proposed unit have the necessary infrastructure to carry out the proposed activities?
• Relationship with ELLIS Guideline questions:
• Are there ELLIS Members/Scholars/Fellows? Are they linked to ELLIS research programs, or collaborating with other ELLIS Fellows?
• Have they volunteered time for ELLIS?

Impact on society and economy (Guideline questions)

• Do they have a plan to have an impact on society? How feasible and ambitious is the plan? What kind of impact(s) do they envision?

Startups and Entrepreneurship (Guideline questions)

• Do they have a program for spinning off companies?
• Do they provide flexibility to ELLIS researchers to get involved with startup companies?

Dissemination and Outreach (Guideline questions)

• How feasible and ambitious are the proposed dissemination and outreach activities?

ELLIS criteria (Guideline questions)

• Are all the ELLIS criteria met? If not, how easy/feasible would be to address the missing criteria within a reasonable time frame?

Structure of the evaluation form for the reviewers:

Each proposal will be reviewed by at least 4 international reviewers. To avoid conflicts of interest, all the reviews are done by scientists of countries different from the units being evaluated. The reviews are organized into four sections:

• Evaluation of each section in the proposal
• ELLIS criteria -- Excellence
• ELLIS criteria -- Enabling factors
• Overall evaluation

The criteria and guidelines are meant to set the level of expectations as objectively as possible. However, it is the responsibility of the ELLIS sites committee and the ELLIS board to assess to what degree each individual proposal sufficiently addresses the goals and vision of ELLIS.

 

Guidelines for evaluating scientific excellence for Fellow/Scholar nominations 

ELLIS Fellows are outstanding senior researchers who have made significant contributions to science and commit to acting as ambassadors of ELLIS. They are expected to provide strategic advice and leadership not just scientifically, but also in terms of how to build and grow ELLIS.  ELLIS Scholars are outstanding junior scientists, often assistant professors, who do not yet have the seniority of a Fellow but are on a clear trajectory to reach this level. The level of expected excellence would be similar to what ERC grants or prestigious research institutions (both private and public) would look for. Examples include (but are not limited to): regular publications in top-tier venues, research impact (h-index / citations / real-world applications), prestigious awards, honors and grants, editorial board memberships, invitations to serve as AC/PC for top-tier conferences, membership in editorial boards of top-tier journals, and invited talks/tutorials at top-tier conferences. 

As a reference:

• Scholars: Junior scientists with up to 10 years of post-PhD experience. We also consider extensions for maternity or other types of leaves. The h-index of the current group of Scholars is 14 or higher. The median is 19 and scholars reaching 10 years of post-PhD experience have an h-index of 25 on average. 
• Fellows: Senior scientists with typically more than 10 years of post-PhD experience and h-index which typically is higher than 25. The median h-index of the current Fellows is 53.
• Depending on the research areas there may be systematic differences that should be taken into account. Theoretical research may not generate the same amount of impact as computer vision research or other more applied disciplines.

Note the eligible career breaks should be taken into consideration, following the ERC criteria described here: https://erc.europa.eu/content/according-conditions-erc-starting-grant-2020-call-what-are-eligible-career-breaks-extension

Each candidate is evaluated by a number of existing Fellows working in a different country and with a different nationality each providing a score according to the following 6-point scale:  5 - outstanding; 4 - excellent; 3 - very good; 2 - good; 1 - fair; 0 - poor. As a reference, Scholars/Fellows are expected to have a score of at least 4.