Change Or Die? Revisiting The Signal Processing Publication Model

It has long been known that research in electrical engineering and computer science communities often follows a different publication philosophy (i.e., "conference plus journal" versus "preprint plus conference"). While in the past these differences were mainly formal, recently these publication models, along with reproducible research policies, have started to significantly affect the way the research is conducted. Indeed, early disclosure of ideas through preprints can speed up the pace of research. This has worked tremendously well in some areas, e.g. deep learning and artificial intelligence, where the increased pace of research has led to increased impact. The objective of the proposed panel is to bring these issues to the attention of the signal processing community that gathers at ICASSP since it is the historically and scientifically most important signal processing conference. The overarching question to be discussed is whether the publication model currently adopted in signal processing is still working well, or if changes should be made. A discussion on this topic is very timely and could bring out some interesting ideas, possible solutions and ways forward, in order to keep signal processing research relevant under these changing research scenarios. To address the issues above, this panel will host researchers from academia and industry, who have published in both signal processing and computer science venues, covering both signal and image/video processing, as well as editors of IEEE journals.