The third Asian Computing Science Conference (ASIAN 97) was held in Kathmandu, Nepal, from December 9 to 11, 1997. The first two conferences had met in Bangkok (1995) and Singapore (1996).
ASIAN is a meeting specifically aimed at bringing together computer science researchers from Asia, with an emphasis on programming language theory and formal reasoning about systems. The constitution of the Steering Committee reflects this: Dines Bjorner (Macau), Shigeki Goto (Japan), Joxan Jaffar (Singapore), Gilles Kahn (France), Kanchana Kanchanasut (Thailand), Jean-Jacques L\'evy (France), R.~K.~Shyamasundar (India) and Kazunori Ueda (Japan). Shyamasundar and Ueda were Co-chairpersons of the Programme Committee for ASIAN 97. The organization was mainly done by Kanchanasut and Roland Yap (Singapore).
The choice of Kathmandu was a surprise as many of us are ignorant of computer science research in Nepal. The group which did the bulk of the local arrangements work was ICIMOD, the International Centre for Integrated Mountain Development, and they have a very active group working mainly in image processing. There were also some participants from the Computer Association of Nepal, but I rather suspect that they were turned off by the theory stuff.
The conference had a keynote address by Michael Rabin (Harvard Univ and Hebrew Univ), and two invited lectures: Nicholas Ayache (INRIA) and Randal Bryant (Carnegie Mellon Univ). There were 24 contributed papers and 10 poster presentations (selected from 122 submissions).
Rabin opened the conference with a beautiful lecture on proving correctness of protocols through randomization. The key idea is simple: instead of accounting for all possible behaviours of a protocol, the task is one of showing that certain random variables used in the randomized solution are independent. The advantage is this: while deterministic solutions get very messy in trying to anticipate various possible control flows, the randomized ones (when you find them !) are elegant and simple. Rabin showed this for the multiprocessor mutual exclusion problem and for the Byzantine agreement problem. The idea is greatly appealing, but it is hard to accept this as a verification methodology, as claimed by the speaker.
Ayache spoke on medical image processing and showed what a range of challenges lie there for computer scientists, ranging from design of algorithms to formal verification of designs. Bryant's talk was on verification of pipelined microprocessors. While it is easy to comprehend that verifying arrays of processors should involve a symbolic approach, based on single processor verification and properties of abstraction functions, the details are forbiddingly hard. Bryant and co-workers, as well as researchers in Dill's group (Stanford) have been addressing these problems in the last few years and have come up with some solutions. Bryant's talk concentrated on one technical issue here and gave a detailed comprehensible solution. An admirable efoort.
Among the contributed papers, there were about 3 papers in logic, 4 papers in concurrency theory, 5 papers in logic programming and the rest rather diffuse, ranging from recursive path orderings to visual specifications. The posters showed specific programming systems, either at the language specification or implementation level.
On December 12, there was a mini-workshop with 4 talks: Doug Tygar of Carnegie-Mellon spoke on auction protocols in computer security. The area seems to have many simply-stated and hard-to-solve problems, and he made the subject very inviting. Jean-Jacques Levy from Paris spoke on a language for process migration he is working on, and showed the basic difficulties in this, particularly in saving state and checkpointing. Catuscia Palamidessi (Italy, now in Penn State, USA) spoke on process calculi with process export, and I gave a talk on problems in local model checking.
An interesting poster (not part of the conference posters at all !) was from John Plaice (Sydney). It advertised a system called {\tt Omega}, a text processor meant for multi-script texts. There were examples from most Indian languages, and the fonts were captivatingly beautiful. In the same sentence you can freely mix several languages and generate lovely text. It is going to be available in TEX-integrated form.
The organization was excellent, with five-star food at every meal. There was also an enjoyable outing to Bhaktapur, despite non-cooperative weather, which was cold and rainy almost throughout the conference. However, it did clear afterwards, and the magnificence of the Himalayas made the waiting very worthwhile. And, of course, Kathmandu is itself a great tourist location, and visiting the Buddhist stupas, Shaivite temples and Royal courts was enchanting.
ASIAN 98 will be in Manila, Philipines, with hopefully lower registration fees !