Venus AUEB Forum

Geia se olous,

opws eixa pei,
ksekiname authn thn Paraskeuh.

Tha dwsw mia omilia gia cooperative games.
Stis prohgoumenes parousiaseis eidame arketa themata panw se non-cooperative games,
paixnidia dhladh opou kathe paikths einai on his own kai krinei panta me bash to diko tou utility.

Thn Paraskeuh tha doume ena diaforetiko montelo paigniwn opou epitrepetai h sunergasia metaksu paiktwn.
Tha kanoume prwta mia mikrh eisagwgh gia cooperative games kai meta tha parousiasw kapoia themata apo mia prosfath ergasia mas (akolouthei titlos kai abstract).
Opws panta sthn A36 stis 4.

Thn epomenh ebdomada, to tea tha ginei Pempth stis 11 kathws o omilhths (pou einai enas apo tous symmetexontes sto sunedrio SAGT thn epomenh ebdomada)
den mporouse na meinei Ellada parapanw meres.


Algorithmic Aspects of the Core in Cooperative Games over Graphs
(joint work with Georgios Chalkiadakis and Nick R. Jennings)

Abstract:
In many real-world settings, the structure of the environment might
constrain the formation of certain coalitions among agents. For instance,
this is the case in sensor and telecommunication networks,
or multiagent settings with restricted inter-agent communication.
Examining the stability of formed coalition structures in such settings
has not received much attention to date. We address this by considering various
models of cooperative games defined on a graph structure. First, we focus
on characteristic function games (CFGs) defined on graphs. These
are regular transferable utility games, along with a graph that determines
which coalitions are feasible. In particular, a coalition S
can emerge only if S is a connected set in the graph.
We study the notion of the core, which is the set of payoff allocations such that no
subset of players has an incentive to deviate.
More precisely, we look at the modified version, in which it suffices to check
only deviations that are feasible (connected sets). We investigate the non-emptiness
of the core as well as the complexity of computing stable configurations.
We then present some extensions to the more general class of partition
function games (PFGs), where the value of a coalition depends
on which other coalitions are present in the environment. Finally, we investigate “Bayesian” extensions,
in which information regarding the success of a deviation is provided
in the form of a probability distribution describing the possible reactions
of non-deviating agents.

sandra wrote:Eλατε μωρε, ποτε δεν τα καταλαβαινεις ολα εντελως, και μονο να αρχισεις να βλεπεις πως κινειται η ερευνα και τα ενδιαφεροντα των συναδελφων αξιζει:)

sandra wrote:Σήμερα ήταν πολύ πιο καλή στο theory tea

Abstract:

Hyperlinks on the worldwide web were originally used to increase the intrinsic quality of a page, mainly by aiding navigation or identifying relevant resources. In recent years, the exploitation of web's link structure by the search engines has given new incentives to link-placement: strategic authors now explicitly attempt to boost their own PageRank by careful placement of links. Moreover, such an environment favors the purchase of links. Potential buyers are not restricted to spammers in our days; every page that deserves reputation or taking advantage of its reputation can do it by purchasing the proper links. Given this selfish behavior of the web page authors, Game Theory seems to provide the appropriate framework for studying the evolution of web. Our work investigates the consequences of such strategic behavior via a network formation game. In our model, authors may place outgoing links as well as purchase ingoing links, and their objective is to maximize reputation combined with monetary payoff. Is the web that evolves through this process desired, in terms of total welfare of its users? Can we give the page authors incentives that will result to a web of high total welfare? In this talk we will present some directions and advances of this work in progress.