http:^^www.cs.wustl.edu^~sandholm^

来自「This data set contains WWW-pages collect」· EDU^~SANDHOLM^ 代码 · 共 626 行 · 第 1/2 页

Date: Thu, 21 Nov 1996 21:33:04 GMT
Server: NCSA/1.5.2
Content-type: text/html

<title>Tuomas W. Sandholm ([e-mail])</title>
<h1>Tuomas W. Sandholm</h1>
<hr>
<!WA0><img src="http://www.cs.wustl.edu/~sandholm/sandholm.big.gif"><p>
<hr>
<dl>
<dd>Assistant Professor
<dd><!WA1><a href="http://www.cs.wustl.edu/rcfdocs/newhome/index.html">Washington University</a>
<dd><!WA2><a href="http://www.cs.wustl.edu/"><!WA3><a href="http://www.cs.wustl.edu/">Department of Computer Science</a>
<dd>One Brookings Drive, Campus Box 1045
<dd>St. Louis, MO 63130
<dd>[phone] (office)
<dd>[phone] (fax)
<dd>[e-mail]
<dd>Born 12/21/1968
</dl>
<hr>
<H1>Curriculum Vitae</H1>

For the official CV, click <!WA4><a href="http://www.cs.wustl.edu/~sandholm/cv.ps">here</a>. An official
research statement is available <!WA5><a
href="http://www.cs.wustl.edu/~sandholm/research_statement.ps">here</a>.

<H2>Research Interests</H2> <P><I><!WA6><a
href="http://dis.cs.umass.edu/research/negotiation.html">Negotiation
among computationally bounded self-interested agents</a>, multiagent
systems, distributed problem solving, distributed artificial
intelligence, negotiation, coalition formation, electronic commerce,
resource-bounded reasoning, normative methods for system design,
machine learning, combinatorial optimization, scheduling.  </I>

<P>My primary interests lie in inherently distributed combinatorial
problems - resource and task allocation and scheduling among agents
with different levels of autonomy, e.g. representing different real
world enterprises that are seeking synergies by limited
cooperation. Such settings occur for example in production scheduling
in multi-enterprise agile manufacturing, in negotiations with
real-world subcontractors, in vehicle routing among dispatch centers
of different companies, in meeting scheduling, in classroom scheduling
etc. I am studying different commitment strategies and protocols in
multiagent systems, and the possibility of carrying out contracts
without external enforcement. I have studied the role of marginal
costs as a basis for negotiation and questions of bounded rationality
and deliberation scheduling when the exact computation of marginal
costs is intractable. I have also analyzed the effect of asynchrony
and message congestion among negotiators and greater domain risk
tolerance to enhance the negotiation process computationally. I am
currently studying the usefulness of distributed versions of a variety
of combinatorial optimization algorithms from the AI and OR
communities. Some other research issues that I am currently addressing
are: using machine learning techniques to alternate between different
negotiation strategies based on dynamic properties of the environment,
cost-based constraint relaxation, anticipation of future contracts in
an agent's negotiation strategy, fair on-line profit division schemes
among agents, and contracts involving multiple agents as opposed to
just two. Lately, I have also developed a normative theory of
coalitions among bounded rational agents.

<P>My secondary interest is machine learning. On the theoretical side,
I lately coauthored a paper that disproved a justification for biasing
decision tree induction algorithms towards generating trees with few
leaves. I have also coauthored an empirical paper that studied the
applicability of Q-learning to multiagent settings. On the applied
side I am currently - in cooperation with domain experts who actually
need the solution - comparing a number of symbolic induction methods
and neural network algorithms to enhance classification accuracy in a
real-world biological problem. Specifically, we use the methods to
predict morbidity and mortality in equine colic.

<hr>

<H2>Education</H2>
<UL>

<LI>Ph.D., Computer Science, University of Massachusetts at Amherst,
September 1996.  GPA 4.0.  Doctoral dissertation: Negotiation among
Self-Interested Computationally Limited Agents.

<LI>M.S., Computer Science, University of Massachusetts at Amherst, May
1994. GPA 3.95.  M.S. Thesis: A New Order Parameter for 3SAT /
Utility-Based Termination of Anytime Algorithms.

<LI>M.S., <em>with distinction</em>, Industrial Engineering and
Management Science, Helsinki University of Technology 1988-91. In the
Finnish system, the B.S. degree is included in the M.S. degree.
M.S. Thesis: Automatic Cooperation of Dispatch Centers in Vehicle
Routing.

<LI>Pilot second lieutenant (obligatory military service), Finnish
Airforce Academy, 1987-88.

</UL>
<hr>

<H2>Work Experience</H2>

<P><B>Research Assistant</B> for Prof. Victor Lesser, Fall 1992 -
present. University of Massachusetts at Amherst, Department of
Computer Science. Developed a normative theory of coalitions among
self-interested agents whose rationality is bounded by computational
complexity. Extended the contract net framework towards working among
self-interested agents whose rationality is limited by computational
complexity. Explored the advantages of different modes of commitment
and the use of marginal costs in negotiation among intelligent
agents. Invented a method that enables autonomous computational agents
to carry out exchanges without enforcement. Participated in the design
and development of a distributed scheduling system for airport ground
resource management. Discovered new structure in an abstract decision
problem (3SAT). Developed a method for calculating the value of
probabilistic information and a way to use this value for optimal
termination of anytime algorithms by conditioning the algorithm's
perfomance profile on execution so far.

<P><B>Research Scientist</B>, Fall 1990 - Summer 1992. Technical Research
Centre of Finland, Laboratory for Information Processing. Designed and
developed a prototype system for distributed optimization of vehicle
routes among autonomous dispatch centers. Studied issues of
commitment, asynchrony, message congestion, and risk attitude in this
domain using data from real dispatch centers. Built a train transport
optimization application for a large forestry company. Participated in
the development of a large scale forestry transport optimization
application involving multiple modes of transport and different time
scales.

<P><B>Researcher/Systems Analyst</B>, Spring 1990 - Fall 1990. Nokia
Research Center, Knowledge Engineering Department. Participated in the
development of an XIPlus-based expert system in the domain of
retirement and pension law. In parallel, participated in developing
and testing a compiler for transforming XIPlus rules into C-code.

<P><B>Systems Developer</B>, Fall 1988 - Spring 1990. Kielikone
Corporation. Converted the <I>Morfo</I> Finnish proofreading program
for use with different word processing programs and different
execution platforms.

<P><B>Database developer</B>, Summer 1989. Smart and Final Iris, Co.,
Los Angeles, California. Participated in the development of a customer
database.

<hr>


<H2>Publications</H2>

<H3>Refereed Journal Articles</H3>
<UL>

<LI>Sandholm, T. and Lesser, V. 1996. Coalitions among Computationally
Bounded Agents.  <I>Artificial Intelligence</I>, Special issue on
Principles of Multiagent Systems, accepted for publication.

<LI>Sandholm, T. and Crites, R. 1995. <!WA7><a
href="ftp://ftp.cs.umass.edu/pub/lesser/sandholm-biosystems95.ps">Multiagent
Reinforcement Learning in the Iterated Prisoner's Dilemma.</a>
<I>Biosystems</I> 37: 147-166, Special Issue on the Prisoner's
Dilemma.

</UL>
<H3>Refereed Journal Articles Submitted and in Preparation</H3>
<UL>

<LI>Sandholm, T. and Lesser, V. 1996. Advantages of a Leveled
Commitment Contracting Protocol. <I>Artificial Intelligence</I>,
in preparation.

<LI>Sandholm, T. and Lesser, V. 1996. Equilibrium Analysis of the
Possibilities of Unenforced Exchange in Multiagent Systems.
<I>Artificial Intelligence</I>, in preparation.

<LI>Sandholm, T. and Lesser, V. 1996. Optimal Information Value Based
Termination of Anytime Algorithms with Conditional Performance
Profiles. <I>Real-Time Systems</I>, in preparation.

<LI>Sandholm. T., Brodley, C., Vidovic, A. and Sandholm,
M. 1996. Comparison of Regression Methods, Symbolic Induction Methods
and Neural Networks in Morbidity Diagnosis and Mortality Prediction in
Equine Gastrointestinal Colic. <I>AI in Medicine</I>, in preparation.

<LI> Sandholm, T. 1996. TRACONET: An Implementation of the Contract
Net Protocol Based on Marginal Cost Calculations. Extended
version. <I>Group Decision and Negotiation</I>, in preparation.

<LI>Sandholm, T. and Lesser, V. 1996. Issues in Extending the Contract
Net Framework for Self-interested Resource-bounded Reasoning
Agents. Extended version. <I>Group Decision and Negotiation</I>, in
preparation.

</UL>
<H3>Highly Refereed Conference Publications</H3>
<UL>

<LI>Sandholm, T. 1996.  Limitations of the Vickrey Auction in
Computational Multiagent Systems. <I>Second International Conference
on Multiagent Systems (ICMAS-96)</I>, Keihanna Plaza, Kyoto, Japan.

<LI>Sandholm, T. and Lesser, V. 1996. Advantages of a Leveled
Commitment Contracting Protocol. <I>Thirteenth National Conference on
Artificial Intelligence (AAAI-96)</I>, Portland, OR, pp. 126-133.

<LI>Sandholm, T. 1996. A Second Order Parameter for
3SAT. <I>Thirteenth National Conference on Artificial Intelligence
(AAAI-96)</I>, Portland, OR, pp. 259-265.

<LI>Sandholm, T. and Lesser, V. 1995. <!WA8><a
href="ftp://ftp.cs.umass.edu/pub/lesser/sandholm-ijcai95-coalition.ps">Coalition
Formation among Bounded Rational Agents.</a> <I>14th International
Joint Conference on Artificial Intelligence (IJCAI-95)</I>, Montreal,
Canada, pp. 662-669.

<LI>Sandholm, T. and Lesser, V. 1995. <!WA9><a
href="ftp://ftp.cs.umass.edu/pub/lesser/sandholm-ijcai95-equilibrium.ps">Equilibrium
Analysis of the Possibilities of Unenforced Exchange in Multiagent
Systems.</a> <I>14th International Joint Conference on Artificial
Intelligence (IJCAI-95)</I>, Montreal, Canada, pp. 694-701.

<LI>Sandholm, T. and Lesser, V. 1995. <!WA10><a
href="ftp://ftp.cs.umass.edu/pub/lesser/sandholm-icmas95-issues.ps">Issues
in Automated Negotiation and Electronic Commerce: Extending the
Contract Net Framework.</a> <I> First International Conference on
Multiagent Systems (ICMAS-95)</I>, San Fransisco, pp. 328-335.

<LI>Neiman, D., Hildum, D., Lesser, V. and Sandholm, T.
1994. Exploiting Meta-Level Information in a Distributed Scheduling
System. <I>Twelfth National Conference on Artificial Intelligence
(AAAI-94),</I> Seattle, Washington, pp. 394-400.

<LI>Sandholm, T. 1993. <!WA11><a
href="ftp://ftp.cs.umass.edu/pub/lesser/sandholm-aaai93-traconet.ps">An
Implementation of the Contract Net Protocol Based on Marginal Cost
Calculations.</a> <I>Eleventh National Conference on Artificial
Intelligence (AAAI-93),</I> Washington DC, pp. 256-262.

</UL>
<H3>Other Refereed Conference Publications</H3>
<UL>

<LI>Sandholm, M., Sandholm. T., Brodley, C., and Vidovic,
A. 1996. Linear and logistic regression, symbolic induction methods,
and neural networks in morbidity diagnosis and mortality prediction in
equine gastrointestinal colic: An extended abstract. <I>The Second
Annual SEPSIS / SIRS Conference: Reducing Mortality to Patients &
Suppliers,</I> Washington, D.C. Poster presentation.

<LI>Sandholm, T. and Lesser, V. 1995. <!WA12><a
href="ftp://ftp.cs.umass.edu/pub/lesser/sandholm-manufacturing.ps">On
Automated Contracting in Multi-enterprise Manufacturing.</a>
<I>Proc. Improving Manufacturing Performance in a Distributed Enterprise:
Advanced Systems and Tools</I>, Edinburgh, Scotland, pp. 33-42, July
13-14.

<LI>Sandholm, T. 1992. Automatic Cooperation of Area-Distributed
Dispatch Centers in Vehicle Routing.<I> International Conference on
Artificial Intelligence Applications in Transportation
Engineering,</I> San Buenaventura, California, pp. 449-467.

<LI>Sandholm, T. 1992. Automatic Cooperation of Factorially
Distributed Dispatch Centers in Vehicle Routing. <I>Joint
International Conference on Operational Research / Management Science
(EURO / TIMS -92),</I> (Collection of abstracts), Helsinki, Finland.

<LI>Sandholm, T. 1992. A Bargaining Network for Intelligent
Agents. <I>Finnish Artificial Intelligence Conference (STeP-92), New
Directions in Artificial Intelligence,</I> Espoo, Finland. Vol. 3,
pp. 173-181.

<LI>Linnainmaa, S., Jokinen, O., Sandholm, T. and Vepsalainen,
A. M. 1992. Advanced Computer Supported Vehicle Routing for Heavy
Transports. <I>Finnish Artificial Intelligence Conference (STeP-92),
New Directions in Artificial Intelligence,</I> Espoo, Finland. Vol. 3,
pp. 163-172.

<LI>Sandholm, T. 1991. A Strategy for Decreasing the Total
Transportation Costs Among Area-Distributed Transportation
Centers. <I>Nordic Operations Analysis in Cooperation (NOAS-91): OR in
Business,</I> Turku, Finland. Publications of the Turku School of
Economics and Business Administration.


</UL>
<H3>Refereed Conference Publications Submitted and in Preparation</H3>
<UL>

<LI>Sandholm, T. and Lesser, V. 1996. A Paradox of Limited Rationality
in Multiagent Systems. In preparation.


</UL>
<H3>Refereed Workshop Publications</H3>
<UL>

<LI>Sandholm. T., Brodley, C., Vidovic, A. and Sandholm, M. 1996. <!WA13><a
href="ftp://ftp.cs.umass.edu/pub/lesser/sandholm-aaai96ss.ps">Comparison
of Regression Methods, Symbolic Induction Methods and Neural Networks
in Morbidity Diagnosis and Mortality Prediction in Equine
Gastrointestinal Colic.</a> AAAI Spring Symposium Series, Artificial
Intelligence in Medicine: Applications of Current Technologies,
pp. 154-159, Stanford University, CA.

<LI>Sandholm, T. and Crites, R. 1995. <!WA14><a
href="ftp://ftp.cs.umass.edu/pub/lesser/sandholm-ijcai95ws-learning.ps">On
Multiagent Q-Learning in a Semi-competitive Domain.</a> <I>14th
International Joint Conference on Artificial Intelligence (IJCAI-95),
Workshop on Adaptation and Learning in Multiagent Systems</I>,
Montreal, Canada, pp. 71-77.

<LI>Sandholm, T. and Lesser, V. 1994. An Exchange Protocol without
Enforcement. <I>13th International Workshop on Distributed Artificial
Intelligence (DAI-94),</I> Seattle, Washington, pp. 305-319.