Geoff Gordon

Parallel Stochastic Hill- Climbing with Small Teams

We address the basic problem of coordinating the actions of multiple robots that are working toward a common goal. This kind of problem is NP-hard, because in order to coordinate a system of n robots, it is in principle necessary to generate and evaluate a number of actions or plans that is exponential in n (assuming P ≠ NP). However, we suggest that many instances of coordination problems, despite the NP-hardness of the overall class of problems, do not in practice require exponential computation in order to arrive at good solutions. In such problems, it is not necessary to consider all possible actions of the n robots; instead an algorithm may restrict its attention to interactions within small teams, and still produce high-quality solutions.

Dynamics and anchoring of heterochromatic loci during development

Positioning a euchromatic gene near heterochromatin can influence its expression. To better understand expression-relevant changes in locus positioning, we monitored in vivo movement of centromeres and a euchromatic locus (with and without a nearby insertion of heterochromatin) in developing Drosophila tissue. In most undifferentiated nuclei, the rate of diffusion and step size of the locus is unaffected by the heterochromatic insertion. Interestingly, although the movement observed here is non directional, the heterochromatic insertion allows the flanking euchromatic region to enter and move within the heterochromatic compartment. This study also finds that a constraint on chromatin movement is imposed which is a factor of distance from the centric heterochromatic compartment. This restraint prevents the heterochromatic locus from moving away from the centric heterochromatin compartment. Therefore, because of the constraint, even distinct and non-random nuclear organizations can be attained from random chromatin movements. We also find a general constraint on chromatin movement is imposed during differentiation, which stabilizes changes in nuclear organization in differentiated nuclei.

Inter-cultural opponent behavior modeling in a POMDP based Automated Negotiating Agent

As the world gets increasingly networked, business and political negotiations take place between people of different cultures. Cross-cultural negotiations have been mainly studied empirically and there is a dearth of computational models of negotiation that incorporate the culture of the negotiators. In this chapter, we take the first steps towards building a partially observable Markov decision process (POMDP) based automated negotiation (PAN) agent, that takes the culture of the negotiators into account. We move away from the offer-counteroffer paradigm that is usually used in computational modeling of negotiation. We assume that apart from making offers, the agents can take other actions for seeking/providing information during negotiation. Our approach is motivated by the experimental findings that (a) during negotiation, people communicate their preferences and justification of their preferences apart from making direct offers and (b) cultural distinctions can be made between negotiating agents based on an abstract coding of their conversation. We show that in accordance with an existing cognitive theory of inter-cultural negotiation from behavioral psychology literature, we can construct a POMDP model of negotiation. A key challenge in developing the PAN agent is in obtaining the state transition function for the POMDP. We demonstrate that the state transition function can be built from transcripts of actual negotiations between people.