-------------------------------------------------------------------------------
                                       
                                       
		   B O S T O N   U N I V E R S I T Y
                                       
		      Computer Science Department
                                       
			  C O L L O Q U I U M
                                       
		       Wednesday October 12, 1994
                                       
				 3:00pm
		       (Coffee served at 2:30pm)
                                       
			 Seminar Room / MCS 135
                                       

-------------------------------------------------------------------------------
This colloquium is the first of three that survey the faculty research
interests of Boston University's Computer Science Department.

			Professor Azer Bestavros

Over the last few years, my research work has focussed on real-time
systems, for which timeliness and predictability are of prime concern.
In that respect, I have suggested and advanced two general concepts.
The first concept is that redundancy could be used to boost
timeliness. I have demonstrated this in two application domains:
real-time databases and real-time communication protocols. The second
concept is that predictability of real-time programs could be improved
if the limitations of the underlying machinery (hardware and operating
system) are made visible to programmers. I am demonstrating this by
developing compiler and programming tools for what I term as
"physically-correct" programming languages, such as Cleopatra. In this
talk, I will describe my research interests and summarize the various
research projects I have been working on recently.

			 Professor Wayne Snyder

My general area is Automated Deduction, which is the attempt to
automated various kinds of logic, such as first-order predicate
calculus, temporal logic, or higher-order logic. My particular interests
have been in unification and in equational reasoning.  Currently I am
working in two areas. The first is deduction with constraints, i.e.,
separating the deduction process into two parts, a background of a
well-understood and well-behaved logic which can be implemented
efficiently, and a foreground of a less-understood logic which requires
proof search. The second is in applying techniques from term rewriting
to proving the correctness of production rule systems (such as are used
in expert systems). In this project I am implementing an environment for
writing expert systems based on the principles of Knuth-Bendix
Completion (which is the fundamental technique used in equational
reasoning).

		      Professor Abdelsalam Heddaya

My work in distributed and parallel systems aims at obtaining high
performance and high reliability, using slow, unreliable (and hence
inexpensive) generic hardware components. I hope to see computer systems
that enable people to pool their resources in order to achieve these two
goals. For example, a typical campus community commands, a tremendous
processing and storage capacity that is often underutilized.  I work to
extract parallelism efficiently from such a system, and to enable the
aggregate to be more reliable than its components.  Each of my two
research projects focuses on one of these questions: Mermera is a
distributed memory system for parallel computing, and BURDS is a
replicated distributed object store.  Both projects have primary use of
the Distributed and Parallel Systems lab, which houses a private network
of seven Sun Sparc workstations and a server, each with a local disk.

			 Professor Steve Homer

I will discuss 3 areas in theoretical computer science, learning theory,
approximation algorithms and complexity theory, in which I am currently
doing research.