Ongoing technological revolutions continually bring forth new research challenges and myriad research communities within Computer Science usually respond to them on their own. It has been generally recognized that we need tighter collaboration across these communities: researchers from different communities working jointly "in the field," constantly informing each other, inventing in their respective areas in response to their experience in the field, and forging systems and solutions that are simultaneously validated by both algorithmic and systems communities. Such an experience will produce algorithms that are practical as well as powerful, and systems and applications that are efficient and effective, without working to make independent research threads meet ex post. We believe that the time is appropriate for "Algorithms in the Field." We propose to initiate discussions and collaborations on understanding the challenges and promises of algorithmic foundations in various CS fields, and further establish the vision of algorithmic foundations forged in these fields in practice. The development of successful examples will increase awareness that collaborations across the subdisciplines should be customary, and not exceptional. The main goal of the workshop is to provide a working vision for such examples of Algorithms in the Field, and a prioritization of near term goals and directions for scientific investigation.
Computer Science (CS) is rapidly evolving with new computing, communication and storage technologies breaking barriers and enabling new worlds. We need a consolidated effort to understand the algorithmic foundations of these revolutions and develop practical algorithms. This requires collaboration between the different fields of Computer Science, and researchers in algorithmic foundations. This workshop developed a vision of "Algorithms in the Field". Generally, Systems researchers conceive of, build and demonstrate powerful worlds and apply algorithmic techniques that are available. Foundational algorithms researchers abstract theories and propose new algorithms. To a limited extent, there is collaboration: algorithms researchers attempt to be faithful in modeling new technologies, and systems researchers strive to test out emerging algorithmic solutions. The alternative, envisioned by this workshop, is that researchers from different communities should work jointly "in the field" (like in field sciences), constantly informing each other, inventing in their respective areas in response to their experience in the field, and forging systems and solutions that are simultaneously validated by both algorithmic and systems communities (without needing to make independent research threads meet ex post). The Workshop on Algorithms In The Field (W8F) was held at DIMACS May 16 – 18 with close to 60 invited researchers and 90 in all, including local students and researchers. Research leaders from networking to databases, machine learning, data mining, statistics, and other areas were represented, in addition to algorithms/theory researchers. Besides selected talks, the attendees were formed into 6 breakout groups respectively addressing: How to acquire, manage data and support data applications? How to analyze data, mine and predict? How to connect and manage devices, locations and applications? How to understand, enable and optimize social interactions? What are core algorithmic models and methods, past, ongoing, and for the next two decades? How to define and develop "Algorithms in the Field" as a research practice? The workshop produced scintillating interaction and debate, and led to indentifying potential areas of joint field research, challenges and bottlenecks to overcome and concrete suggestions for future. All the talks, slides and videos, as well as reports of the breakout groups and overall directions can be found at the website: https://sites.google.com/site/algorithmsinthefield/ Among the key areas identified were Sensed Data problems, parallelism, Autotuning algorithms, new sublinear approximation approaches, Algorithmic Statistics, Supervised+X learning, etc. Some of the concrete suggestions for the future included the need to build Field teams of researchers, algorithms foundry and office hours, as well as guidance to support not only medium teams of algorithmic and systems researchers on joint proposals, but also, support for setting up algorithms help desks, implementation contests, collaboration sparks and platforms for playing with data. Challenges included validation of the approaches, besides the need to train across disciplines. Nearly all groups emphasized educational aspects including ``field courses'' in the nexus of algorithms and other areas, and identified exisiting courses that could serve as models. The mental exercise of putting researchers from Systems and algorithmicists itself was a ``Field exercise'' and it led to teams which found common problems to work in the future and also followup meetings, with one on Geometry in the Field and Distributed Data workshops materializing.