Statements for the ACP elections - 2018

Carlos Ansotegui

I am currently an Associate Professor in the Department of Computer Science and Industrial Engineering at the Universitat de Lleida (UdL) and the leader of the Logic Optimization Group (LOG, ulog.udl.cat) that brings together researchers from UdL and Universitat Politècnica de Catalunya (PC). Previously, I was a postdoctoral researcher at Cornell University.

My research interests include: modeling and solving decision and optimization problems, through the application of CP techniques. I have coauthored several papers on efficient solving techniques for real-world problems. I have served on many program committees at CP, AAAI, IJCAI and other conferences, have reviewed for several journals on CP topics and have been local chair of CPAIOR’15 and program chair of the workshop MODREF’14.

My main motivation is to bring our technology to other research areas and industrial domains. The essence of this interest can be distilled from the following words in the Handbook of Constraint Programming: "Constraint programming has proven useful -indeed, it is often the method of choice in important applications from industry, business, manufacturing, and science... As well, in the best traditions of application driven research, the chapters describe how focusing on real-world applications has led to basic discoveries and improvements to existing constraint programming techniques. In a fruitful cycle, these discoveries and improvements then led to new and more successful applications." The technology of our community is already a key tool in many processes. The need for this technology is explicit in many research calls, for example, “Efficiency” appears as one of the keywords in several areas of the H2020 European Program, in reference to efficient energy distribution, logistics applications, etc.

The currently elected members of the ACP Executive Committee have worked very hard for the visibility of CP but, of course, this is an endless task. In this sense, I advocate the following measures:

  • Promote synergies with other research areas and the industrial field. This can be implemented by supporting invited talks from these domains to explain their problems and then present a selection as challenges for our community. Finally, work with conferences and journals to create special tracks on these challenges.

  • Increase our support to international competitions. These events have become in our community a catalyst of research. Work with the organizing teams to enforce the creation of standard interfaces and repositories to promote the access to CP tools and benchmarks. Establish the conditions to reach a higher level of software maturity.

  • Promote the participation of our members in international projects. In this sense, we can offer technical sessions on the existent international calls, success stories of our members, lessons learnt, etc.

David Bergman

I am excited by the prospect of serving on the ACP Executive Committee. Since attending my first CP conference in 2011 I have been an active and enthusiastic member of the CP community, assisting with organizing various programs and conferences during my tenure as an ACP member. My past positions include Publicity Chair at CPAIOR 2015, Doctoral Program co-Chair at CP 2015, co-Chair of the 2016 ACP Summer School, Operations Research Track co-Chair at CP 2017 and CP 2018, Master Class Chair at CPAIOR 2018, and I have recently been announced as the Conference Chair of CP 2019. My service to the community is complemented by my research contributions. I was honored to receive the 2014 ACP Doctoral Research Award, and I frequently publish in Constraints as well as in the major annual CP-focused conferences. I also serve on the Editorial Board for Constraints.

My focus as an ACP Executive Committee member, should I have the opportunity to serve in this capacity, will be to strengthen our industry outreach and collaboration. Within our research community, in CP and more broadly in operations research, we are well aware of the tremendous potential of the work we do, and I find our integration with industry to be a tremendous growth opportunity. As an active consultant for a premier management consulting firm, I am exposed to problems that are top of mind in industry, and I know that the application of CP is still in its infancy. As a concrete example, I built a training scheduling optimization algorithm for a large military force - a simple CP model brought force readiness from 60% to near 100% and allowed for the closing of several facilities. This model resulted in significant long term impact which, had the organization known of CP, would have been implemented years ago.

By cultivating relationships with industrial partners, we will learn about problems that can benefit from CP, as well as drive real-world prescriptive decision making that will help organizations run more effectively and strengthen our community’s research impact. One mechanism for generating interest from industry is to design a high-quality, professionally edited presentation that highlights high impact use cases of CP technology. We can either make this presentation available on the ACP website, or have this as a resource that ACP members can bring to organizations, so that others can get a flavor of what CP can do for them. I would be happy to lead this effort should I be elected to the Executive Committee – CP has landed on a comet (https://www.a4cp.org/node/1058), now let’s expand on earth!

Sophie Demassey

I have navigated between Operational Research (OR) and Constraint Programming (CP) since 2000. I have served on many program committees in both domains and been actively involved in both communities, especially at the national level: I took part to the last executive board of ROADEF, the active french OR society, for two years, and I animated from 2004 to 2009 the "CP & OR working group" which acts as a bridge between ROADEF and AFPC, its CP counterpart. I now want to put my experience and my ideas to contribute to the international CP community.

One role of the ACP is to involve its members. In this regard, ACP membership should perhaps be rethought to give its members the sense of belonging, and to give others the envy to join, for example by offering fee-for-service facilities. To drive this discussion, openly, it would be instructive to review the conditions in national or international societies in OR and AI, like the fairly comparable Mathematical Optimization Society, to name but one.

Beside animating the community, promoting CP is what I aim to commit to, first because the tools of CP deserve to be known by the many potential users, then because such a cross-disciplinary scientific field needs a broad base of competences and fruitful exchanges with adjacent fields. Promotion is communication and, among the many options today, a website is the best medium to index and archive knowledge and to make it public. Pragmatically, I would contribute to add online resources to the ACP website developed by the previous executive committees, by updating or setting up directories of success stories, available solvers, research labs and practitioners, institutions and training courses, forums, or any material that could make know what CP is and where to learn more. In parallel, I would advertise it and other CP-related events. Data science and Artificial Intelligence are bubbling at the moment, and we should took the opportunity to claim that, beyond predictive analytics, machine learning and neural networks, there are CP, optimization and decision prescription.

Sophie Demassey,
Associate Professor, Mines ParisTech, Sophia-Antipolis, France

Emmanuel Hebrard

http://homepages.laas.fr/ehebrard

I am researcher at CNRS (French acronym for "national center for scientific research") and ACP member since 2003. I have served in the program committee of IJCAI, AAAI and CP regularly, in the organizing committee of CP 2016 in Toulouse, and as program chair of JFPC 2016 (the French CP conference).

Given its rolling election mechanism, it seems adequate to initially learn "by doing" about the mission and practice of the executive committee before assuming a more proactive approach. This suits my conception of the role of a member of the executive committee which is to serve the community and implement the ideas and decisions emanating from a majority of the members of the ACP rather than impose mine. If I am elected, I would actually like to consider setting up some form of forum to collect initiatives, discuss them, and select the most deserving ones through open deliberation. One can see that as an electronic permanent extension of the annual general assembly of the ACP.

This does not mean, however, that I do not have personal opinions about pressing issues and priorities. For instance, I recently tried to substantiate the feeling I had, from my limited experience as attendee and PC member, that CP is losing ground in general AI conferences over time. The data from IJCAI on DBLP seems to confirm this observation: from about 15% of the accepted papers from 2001 to 2005, the "Constraints and SAT" track* went down to less than 10% of the total from 2007 to 2013 and is now steadily decreasing and under 5% since the 2015 edition. Whether this is an artifact of the recent boom of machine learning or a more profound problem is yet unclear to me. However, this is probably something we should be aware of, and maybe concerned about.

I would like to conclude by stating that I am committed to promote an open definition of constraint programming, ranging from industrial applications to the most theoretical questions, and as inclusive as possible with respect to related fields. Moreover, I am deeply convinced that an important role of the ACP is to support existing as well as new initiatives to share resources (problems, data sets, languages, solvers, teaching material, etc.) and I will do my best to help in this respect.

(*) or similar keywords

Michele Lombardi

CP has an strong potential to address real world problems (like city-wide traffic optimization, or large-scale management of renewable power sources) that may become key to deliver societal benefits in the near future. This is due to its ability to integrate radically different approaches within the same framework. Examples are plentiful and include (just to mention a few) CP/OR hybrids, Lazy Clause Generation, sampling based search methods, CP approaches for itemset mining, and CP techniques employing Multivalued Decision Diagrams. In many cases, these integrated approaches represent the current state of the art for one or several application domains. Additionally, this flexibility makes CP uniquely well suited to tackle complex problems that are hard to model via a traditional interaction between the domain and the modeling expert, but for which an approximate model can be obtained via Machine Learning.

We are located at a nexus between traditionally separated research fields: we have the expertise, tools, and mindset to act as a bridge between Machine Learning, Logic, Operations Research, and Statistics, Planning, and more. This gives us the opportunity to draw from such fields and improve our framework, but also to bring contributions in those related areas, both by enabling successful ideas to cross-over, and by providing novel insights. We may not be a large community, but we can be a catalyst for far-reaching scientific advancements.

In this context, we all stand to gain by ensuring that our interactions with related research communities are frequent, significant, and fruitful. Strategically, this is especially true for areas that are in the spotlight in the current wave of interest in AI, such as Machine Learning and its applications. This could fostered by supporting the organization of clusters/tracks within conferences in related fields, or via specific seminars or workshop. We also need to make sure that CP gets some exposure to students pursuing a degree in Computer Science, Operations Research, AI, or Business Analytics: this could be achieved via the organization of summer schools on the integration of CP with different fields (e.g. CP and OR, CP and ML, CP and planning), or by activating courses that include CP among the covered topics. Finally, achieving a practical impact will require to strengthen our connections with the business world, e.g. by encouraging partnerships between companies and research labs, collecting success stories, or giving visibility to spin-offs and companies using CP. The EC is already coordinating efforts in these directions: I would be glad to be part of such an endeavor and to contribute to the best of my possibilities.

Ian Miguel

I have been a member of the CP community for over twenty years, since beginning my PhD studies on dynamic and soft constraint satisfaction in 1997. In that time, I have been active in service to the community, as a member of the CP conference programme committee over ten times, as CP publicity chair twice (and also for CPAIOR in 2006), as CP sponsorship chair twice, as a member of the CP local organisation committee at St Andrews in 2010 (and also for SAT 2005), and as a mentor in the CP Doctoral Programme several times. I have also been actively involved in the success of the long-running ModRef workshop, participating in the Formul’01 that preceded the ModRef series and then regularly serving on the ModRef programme committee as well as co-chairing the workshop in 2004 (with Alan Frisch) and 2006 (with Steve Prestwich). In addition I contributed a talk on modelling to the ACP Summer School in 2008 and, with Ken Brown, the chapter on Uncertainty and Change to the Handbook of Constraint Programming. For a number of years I was, with Brahim Hnich, the editor of CSPLib.

Although I have worked in several areas of CP the principal focus of my research has been in automated modelling. Although modelling is important to the performance of our solvers, it is also of central importance in making connections outside our immediate field. Easier modelling widens the set of potential users of constraint solving, and model transformation to related formalisms such as SAT, SMT, and MIP builds bridges to other research communities. As a member of the ACP executive I would look for every opportunity to foster such connections, which I think are vital to the growth of our field. The CP and CPAIOR conferences provide annual opportunities to reach outwards to other communities and potential collaborators by means of invited talks and tutorials, focused workshops and conference tracks. Given the huge popularity of AI at present, we must also strive for a more consistent presence at the major AI conferences, perhaps by establishing a regular ACP workshop. We should be proactive in presenting at other conferences and industrial/public events where CP can make an impact. The ACP could sponsor such activity and adopt the responsibility of engaging with these external partners.

A second mechanism for ensuring the health of the field is to maintain and strengthen its foundations. When I joined St Andrews in 2004, I established a module on Constraint Programming for the undergraduate (and later taught postgraduate) degree programme. This module remains popular and serves both as a fertile source of PhD students to help to grow the field and to educate those who choose other careers in the theory and practice of CP. There is a wealth of similar experience in the field and therefore a great opportunity exists to share this experience with each other and those wishing to set up new teaching modules in the subject. As a member of the executive I would be happy to coordinate such activity.

Thierry Petit

I am an associate professor of computer science. My research area is constraint programming. I have been at IMT Atlantique, Nantes, France, for over 15 years. I have also served as an affiliate associate professor at the Worcester Polytechnic Institute (USA) from 2014 to 2017. I have been a member of many program committees: AAAI, CP, CPAIOR, IJCAI (senior PC), among others. I was the program chair of the francophone conference in Constraint Programming (JFPC) in 2014. I published several articles in constraint programming, including fourteen papers at CP and six at IJCAI.

I run for this election to support the constraint programming community.

We are at a turning point in the development of constraint programming. In 2018, journalists and technologists sometimes assert that Artificial Intelligence (AI) could make analytical technologies obsolete, including technologies considered initially as subfields of AI, such as constraint programming. The reason is that what is being called AI in the public sphere is what has been called “machine learning” in the past. Although machine learning techniques gave impressive results, they are subject to limitations. These limitations relate to expressiveness, solution provability, and reasoning reproducibility. Our great fortune is that constraint programming is probably the best technology for addressing these issues. In industry, people are likely to pay attention to the gap between the real applications and the solutions provided by software. As expressed by the renowned statistician George Box, “The most that can be expected from any model is that it can supply a useful approximation to reality: all models are wrong; some models are useful.”. This comment also applies to combinatorial optimization. Thanks to its unique ability to integrate various constraints and solving techniques, constraint programming is the best technology for stating and solving useful models. Convincing all computer scientists about the validity of these arguments remains a challenge.

In this context, I consider that the future of our community lies in the ability of the ACP to strengthen its communication actions further and perform more aggressive advertising towards industries and universities worldwide. As an elected member of the ACP, I would strive to shape the strategic decisions of the community in that direction. Actions proposed to promote our community include:

  • Implement a systematic popularization policy concerning applications. We should not restrict our dissemination of applications of constraint programming to the “application track” of CP and outstanding “success stories” on the ACP website. Especially, constraint programming applications on critical humanistic challenges should be spotlighted outside from our community; for example, applications related to sustainable development, ecology, ethics, medical research, struggle against ethnic and racial discriminations, and ableism. We need to think about alternative policies of scientific popularization.

  • Centralize information. Information about constraint programming on the internet is split into many sites. The ACP website could be expanded to create a unique platform grouping links to: slides, online courses, modelers/solvers, tutorials, examples of code, catalogs, as well as some recordings of conference talks. Marginally, this project could be an additional vector for facilitating the emergence of new ideas for expanding our community.

Christian Schulte

I am a professor of computer science at KTH Royal Institute of Technology, Sweden. I have been part of the CP community since 25 years and have been serving the community in most roles: chair, organizer, reviewer, educator, examiner, author, system builder, competition contributor, executive committee member, panellist, you name it. This of course includes venues other than CP such as CPAIOR, Constraints, IJCAI, AAAI, AIJ, JAIR, PADL, ICLP, PPDP, AI, ….

I am an active member of the programming language community and love building major, high-impact open-source systems that are based on sound principles such as Gecode (gecode.github.io) and Unison (unison-code.github.io.), a system using CP for code generation in compilers. I have more than 20 years of experience in successfully collaborating with industry and have been part of non-university research organizations targeting transfer and innovation.

What I intend to do, if elected, is to contribute to making the ACP stronger, more diverse, and more inclusive with a better display of what we are good at. The key question for the ACP is: whose home should it be?

Today, the ACP offers home and shelter mostly to academics making contributions within CP. Ideally, the ACP should feel home for anyone with an interest in CP, be they from "core" CP, SAT, OR, from any academic community, industry, the public sector, …. In order to convince people to call the ACP home, we need to stress that CP is in fact very often a transformative technology in how problems are approached, modelled, and solved. And yes, we need to aggressively communicate this.

Some areas where I think we can improve and where I want to contribute to the improvement by ideas, discussion, effort, and experience are:

  • We have many success stories that go beyond simple applications but where CP has been transformative such as power systems, embedded systems, security, and programming languages. The ACP must give these stories a venue and act as an facilitator and amplifier of the stories to be heard so that the fundamental impact of CP becomes clear.

  • We have successful competitions, unfortunately we have two rather than one which conveys that we actually do not really know what we want to do. I will push for a single ACP-backed competition which takes into account how the competition can have impact beyond the core CP community.

  • We have a successful CP conference but are we really sure that it is the best format for the community and for inclusion of others? What about SAT, SMT, OR? What about ML, recently having become the new definition of AI?

I do not claim to have the right answers but I am committed to serve, again, to help all of us getting closer to something we and many others love to call home.