dhf@linus.UUCP (David H. Friedman) (06/07/89)
[] I've just returned from ICASSP-89, and there are a number of impressions in my mind that I remember having had after previous ICASSP's as well, and I think it's time I tried to share them with a larger audience to see how many other people feel similarly. I would like to see this start a general discussion which I think would fulfill one purpose of this newsgroup, and which hopefully will reach and influence the organizers of future IEEE conferences. Every year I am struck by the low quality of the presentations in the traditional lecture sessions. Of course there are notable exceptions, but generally I find one or more of the following: poor speaking style, poor organization (e.g., attempting to cover the entire content of the paper in the lecture), poor design or choice of content of vugraphs (e.g., equations from the text of the paper), poor projection conditions (projector too close to screen, image not in sharp focus, etc. making it unreadable from all but the center of the first few rows), and insufficient seating or air-conditioning capacity in the lecture room. Granted, not all of these factors are within the control of the presenter or organizers. In particular, at an international conference it must be expected that many presenters will be speaking in an unaccustomed language, some coping with pronunciation difficulties (indeed, one can only credit them with courage in making the attempt!) and perhaps unsympathetic ambient acoustic conditions as well. Finally, the usual question or two (time permitting) addressed to each speaker is an travesty of the free discussion which is supposedly the goal of any such conference, and which in fact takes place (if it does) outside the sessions instead. At ICASSP-84 I attended my first poster session, and I became an instant convert. I found I could absorb the essential points of each paper at my own pace, in a moment, by reading the headlines on the author's posters, and turn from one to another at will. I could get into discussions with the authors, or eavesdrop on other ongoing discussions and thus ease into the topic until I could put together a question of my own; etc. I have since then participated as an author only in poster sessions, and have found the experience extremely worthwhile. A major factor in this is knowing how to summarize one's paper in what amounts to a graphic abstract, designed to convince the onlooker to read the full paper in the proceedings, rather than a visual account of the full paper. Where appropriate, it is possible to include live demonstrations and even (I saw this at ICASSP-89) use of a cellular radiophone to allow people to speak directly to coauthors unable to be present. Given all this, the next logical step is to ask, why not an all-poster conference? Outside of invited longer tutorial or review presentations by eminent authors, for which the lecture format is really appropriate, why have lecture sessions at all? This idea occurred to me during ICASSP-89, and when I raised it with a few people active in the ASSP Society whom I met on the floor, I found that they already felt the same way. In these conver- sations I raised an impression I had formed at earlier ICASSP's, that poster sessions are considered (at some level) to be less prestigious than the lecture sessions, in the sense that the "better" papers are assigned to lecture sessions while the poster sessions are implicitly seen as a means to absorb the "overflow" of lesser papers. In some instances poster sessions have been held in less accessible or otherwise less favorable locations, e.g., on a different floor or even in a different building, or (at ICASSP-88, which I did not attend) in a crowded corridor - all of which reinforces the impression I mention. I was assured that, at least as far as deliberate policies of paper selection and assignment are concerned, this was definitely not the case, although - tradition being what it is - some such feeling might still exist in the professional community, such that some authors might still feel slighted at being assigned to a poster session, or affronted at being challenged to present their work "while standing on one foot" in the image of a well-known Talmudic story. Insofar as this is the case, my response to it is best illustrated by relating the story itself in full: A heathen approached the two greatest rabbis then living, Shammai and Hillel, asking each to explain the essence of Judaism while standing on one foot. Shammai immediately threw him out. Hillel, on the other hand, raised one foot, cited the Golden Rule, and said, "The rest is just commentary, you can go read it yourself" - in the conference proceedings, so to speak. This is also, IMHO, a good recipe for what a conference presentation should be. One other minor pet peeve I have about conferences concerns name badges. In most cases they can't be read from a distance, as they should be if you're trying to meet people on the fly. Now that the technology has advanced from Selectrics with ORATOR type balls to dot-matrix printers, it should be possible to print in a large, clear expanded font. But even that won't help if people don't wear their badges so they can be read. In any event, as I said at the top, I'd like to get a discussion going, and hopefully this will lead to more useful conferences in the future. dhf@linus (David H. Friedman, The MITRE Corporation, Bedford, MA, USA)
malcolm@Apple.COM (Malcolm Slaney) (06/09/89)
In article <55337@linus.UUCP> dhf@linus.UUCP (David H. Friedman) writes: > Every year I am struck by the low quality of the presentations in the >traditional lecture sessions. You mean you go to see the lectures? Everbody knows the good stuff is always in the hallway. I'm on the organizing committee for the 1992 ICASSP and we have made sure that the hallways are all very large :-). We will provide chairs in the lecture halls so people can rest between hallway talks. :-) Seriously though....we have talked alot about how to make the lectures a bit more useful. We can't do much about the speaker quality but one thing we have talked about is having the session chairman lead a discussion after all the papers comparing and contrasting the material discussed. This way you could ask one speaker why he disagrees with the material in the other sessions. Given the cross cultural differences this is going to be difficult to make work. The lecture format has a big advantage. It is very efficient. Can you imagine 1000 people all trying to gather around the posters from IBM, BBN, AT&T, SRI, etc to see the latest Connected Speech Recognition results? It just wouldn't work. I wasn't able to go to Glasgow but all of us on the 92 ICASSP committee were agast at the overcrowding in the poster sessions. We are planning to use a space that is about 50% as large as the exhibit hall for posters. Also, as it turns out the poster session will be the closest to the hotel lobby. Behind the posters (as you come down the stairs) will be the exhibits and in front of the posters (and through the walkway) will be all of the technical sessions. Of course all of this is subject to change in the next three years. Also, we have suggested to the ASSP conference committee that we have an all day video session running. People could submit video tapes (some work is better shown as video then slides) and a series of video tapes would be shown continuously (but on a schedule so you could plan ahead.) Ideally you would be able to see them in your hotel room but we would also have to have them in a room available for people not staying at the hotel. A hard problem would be figuring out some way for people to ask questions of the authors. Finally, I agree about the larger type on the name badges. I've passed your note on to the rest of the committee. Malcolm Slaney malcolm@apple.com
fede@ethz.UUCP (F. Bonzanigo) (06/14/89)
I am very pleased that David Friedman in Article 73 of comp.org.ieee raised again the point about the format of IEEE conferences, in particular of ICASSP. He proposes an invited review lecture + poster session format and I second it. Actually this idea is not new, but is the one used for ages in biology and physics, at least in Europe. This means that conferences can be successfully organized this way: just ask your friends in the physics or biology departement how they do! From what I heard from them I can add the following rules: * The posters are exposed not only during the poster session, but for the whole conference if possible (this needs a lot of floor space, but the conference needs fewer lecture rooms), or at least long time before the official poster session. In this way one should have enough time to look at the posters by himself, pick the ones most appropriate to his interests, maybe read the papers, and think about good questions to be asked. Enough time (breaks) should be allotted for this purpose. * The poster sessions are when the author has to be present at his poster. They are not organized by topic, but the authors of a given subject are evenly distributed among the poster sessions. In this way they have more chances to discuss with people with the same interests. During the poster sessions no lectures take place. In this way the poster sessions would cease to be the Cinderellas of the conferences and the latter would fulfill better their purpose to forster discussion and information exchange between researchers. A personal note: I remember a discussion on this subject with 4 or 5 ASSP leaders (clearly Al Oppenheim was there, maybe Russ Mersereau, I do not remember who the others were) where I exposed how biologists and physicists organize their conferences. It was on a train from Florence to Milano after the 1978 Florence Conference on DSP. The IEEE 1980 L'Aquila Workshop on DSP has been basically organized in this way and was a great success. I do not know why the half-hearthed kind of poster sessions described by David Friedmann has been adopted afterwards both by IEEE and EURASIP conferences. Federico Bonzanigo Institut fuer Elektronik Swiss Federal Institute of Technology (ETH) CH-8092 Zurich, Switzerland SWITCH: bonzanigo@nimbus.ethz.ch EARN/BITNET: BONZANIGO@CZHETH5A EUNET/UUCP: ...!mcvax!cernvax!ethz!fede Phone: +41 (1) 256-5134 (+ = whatever you have to dial Fax: +41 (1) 251-2172 to call outside your country) Telex: 817115 vaw ch