Jim Boerkoel

Home
Research
Links

Ph.D., Computer Science and Engineering - University of Michigan - Ann Arbor, MI, ABD
M.S., Computer Science and Engineering - University of Michigan - Ann Arbor, MI, August 2008
B.S., Mathematics, Computer Science - Hope College - Holland, MI, December 2005 (Summa Cum Laude)

My research focuses on building foundational technologies for computational agents that can assist people in successfully managing their schedules, as well as coordinate their schedules with those of other agents. In particular, I am working on developing algorithms for efficiently solving constraint-based representations of scheduling problems both in single-agent (centralized) and multiagent (distributed) frameworks.

Download my CV [PDF]

Boerkoel, J. and Durfee, E. 2012. A Distributed Approach to Summarizing Spaces of Multiagent Schedules, To Appear in AAAI 2012.
Boerkoel, J. and Planken, L. 2012. Distributed Algorithms for Incrementally Maintaining Multiagent Simple Temporal Networks, To Appear in ARMS 2012.
Boerkoel, J. and Durfee, E. 2011. Distributed Algorithms for Solving the Multiagent Temporal Decoupling Problem, In Proc. of AAMAS 2011.
Boerkoel, J.; Durfee, E.; Purrington, K. 2010. Generalized Solution Techniques for Preference-Based Constraint Optimization with CP-nets, In Proc. Of AAMAS 2010.
Boerkoel, J. and Durfee, E. 2010. A Comparison of Algorithms for Solving Multiagent Simple Temporal Problem, In Proc. Of ICAPS 2010.
Boerkoel, J. and Durfee, E. 2009. Evaluating Hybrid Constraint Tightening for Scheduling Agents, In Proc. Of AAMAS 2009.
Boerkoel, J. and Durfee, E. 2008. Hybrid Constraint Tightening for Hybrid Constraint Scheduling, In Proc. Of AAAI-2008.

Algorithms for Solving the Multiagent Simple Temporal Problem
James C. Boerkoel Jr.

My work focuses on building computational agents that assist people in managing their activities in environments in which tempo and complexity outstrip people's cognitive capacity, such as in coordinating rescue teams in the aftermath of a disaster, or in helping people with dementia manage their everyday lives. A critical challenge faced in such environments is that individuals must factor complicated constraints into deciding how and when to act on their own goals. A compounding challenge is that their decisions are further constrained by choices made by others with whom they interact and vice versa, such as between cooperating teams in disaster relief or between patients and caregivers in an assisted-living facility. A final challenge is that the interests of individuals, such as privacy and autonomy, along with slow, costly, uncertain, or otherwise problematic communication, may further limit individuals' abilities to work together. My work assumes that a computational agent is associated with each individual, and that these agents will work together efficiently to manage individual and joint activities, while maintaining autonomy and privacy to the extent possible.

My idea is to exploit problem structure by naturally decomposing the problem into conditionally-independent subproblems that agents can reason over separately, thus limiting agent coordination to a much smaller portion of the collective scheduling problem. Each agent, then, has the complete freedom to solve its conditionally-independent subproblem concurrently, privately, and autonomously. To decompose the problem, first each agent summarizes how its local subproblem affects other agents' subproblems in the form of new constraints for those agents, such as a caregiver's agent that sends available appointment times in the form of new constraints to a patient's agent. Summarization constraints promote privacy and autonomy by compactly capturing all necessary coordination information in terms of mutually-known aspects of the problem, which then allows each agent to manage the remaining, potentially sensitive details of its individual's schedule independently and privately. Secondly, agents propose additional decoupling constraint commitments, which, if agreed upon, safely allow those agents to ignore constraints between their subproblems, eliminating any need for further coordination. For example, consider the case where a caregiver's agent commits to providing a drug or exercise regimen prescription by 1:00 PM and a patient's agent agrees to wait to perform the prescribed regimen until after 1:00 PM. These commitments guarantee the coordination of both the prescription and execution of the regimen, and still give both agents significant flexibility to solve the remaining aspects of their individuals' scheduling problems independently and privately. These two ideas are complementary. Summarization hides private details and helps agents make commitments that are more likely to succeed, while commitments help simplify otherwise complex interactions so that an agent can better summarize other aspects of its problem.

One of the reasons I first chose multiagent coordination research is its immense potential for improving how people live. Imagine an application that not only helped students manage and prioritize their academic and social endeavors, but also proactively notified struggling students of opportunities for impromptu study groups. With the proliferation of cell phones, even in developing regions of the world, my work may be foundational in addressing many difficult infrastructural and coordination challenges faced by disadvantaged populations, for example, by facilitating health-care distribution to people who previously could not gain access. Scheduling agents can play a key role in increasing the efficiency of healthcare delivery by achieving better coordination and more efficient resource allocation while protecting patient privacy and keeping pace in an up-tempo environment. Scheduling agents can also provide better allocation of other continuous resources, such as energy. All of these applications point to my larger goal as a researcher: to harness my unique experiences and ideas towards developing socially relevant technology.

Boerkoel, J. and Durfee, E. 2011. Distributed Algorithms for Solving the Multiagent Temporal Decoupling Problem, In Proc. of AAMAS 2011.
Boerkoel, J.; Durfee, E.; Purrington, K. 2010. Generalized Solution Techniques for Preference-Based Constraint Optimization with CP-nets, In Proc. Of AAMAS 2010.
Boerkoel, J. and Durfee, E. 2010. A Comparison of Algorithms for Solving Multiagent Simple Temporal Problem, In Proc. Of ICAPS 2010.
Boerkoel, J. and Durfee, E. 2009. Evaluating Hybrid Constraint Tightening for Scheduling Agents, In Proc. Of AAMAS 2009.
Boerkoel, J. and Durfee, E. 2008. Hybrid Constraint Tightening for Hybrid Constraint Scheduling, In Proc. Of AAAI-2008.

Algorithms for Solving the Multiagent Simple Temporal Problem
James C. Boerkoel Jr.