Plenary talks will be given by Prof. Ronald R. Yager, Prof. Hiroyuki Ozaki, and Junichi Hoshino.
Prof. Ronald R. Yager
Iona College (USA)
Uncertain Information Representation for Decision-Making: Emerging Approaches
Information used in decision making generally comes from multiple sources and is expressed in various modalities. In many cases the information available has some significant associated uncertainty. In order to address this problem there is a need to provide various methods for the representation of different types of uncertain information. Here we shall discuss some recently emerging approaches for attaining this representational capability. One approach we shall discuss is the use of Z-valuations which are based on the use of Z-numbers. These objects allow us to represent information that combines both possibilistic and probabilistic uncertainty. We shall also look at a new approach for modeling fuzzy sets with imprecise membership grades called Pythagorean fuzzy sets. One important issue that arises when using these non-standard representations in decision-making is the comparison of alternative satisfactions, that is we must compare mathematical objects that are not naturally ordered. We consider this important problem.
Prof. Hiroyuki Ozaki
Keio University (Japan)
Recursive Capacitary Kernel
We define a capacitary kernel as a function which maps a current state to a capacity (a non-additive measure)
which governs the next period's uncertainty. We then use a capacitary kernel to
define the time-consistent recursive objective
function for a dynamic optimization problem. We impose and discuss assumptions on the capacitary
kernel for this objective function to be well-defined. We also provide some decision-theoretic foundation
of this objective function. Furthermore, we develop a dynamic programming technique to solve this
optimization problem by exploiting
the recursive structure of the objective function defined by the recursive capacitary kernel.
Prof. Junichi Hoshino
Tsukuba University (Japan)
Framework of Entertainment Computing and Its Applications
Entertainment Computing is one of the promising application domains of MDAI (modeling decisions for artificial intelligence) field. In this talk, I'm going to overview the collaborative creation process of entertainment using rich examples including media art, physical exercise game, learning game, multi-agent Game AI, and social information exchange on hobby.
We also show the examples of the entertainment systems using behavioral and cognitive user modeling:
1) The Fighting Game Character Using Imitation Learning
One of the limitations of computer-based opponents in action games is that the AI character is constructed in advance, and players quickly become bored with their prepared tactics. We built an online coliseum in which a non-player character (NPC) incrementally learns action sequences and combinations of actions, allowing the NPCs to adopt different fighting strategies after fighting with different players. Indi-vidual fighting styles can be generated from a unique fighting history. We developed a new action learning engine that automatically analyzes the actions of a human player and extracts the effective fighting sequences. Action control trees are generated automatically and incrementally added to the NPC's action profile.
2) A Wellness Entertainment System using a Trampoline Interface
We describe the wellness entertainment system using a trampoline interface. In this system, we use a mini trampoline as the input device. The system enables the user to move and jump freely in VR space by exaggerated movement corresponding to walking or jumping on the mini trampoline. Improvements in exercise motivation and support for continuous exercise are achieved in our system, since it is possible to enjoy strolling through a virtual space, which is usually difficult to experience, by exercising on the mini trampoline without injury to the user's joints.
3) Disaster Experience Game in a Real World
We describe a disaster experience game system which could instruct about general knowledge and regionally specific disaster risk in a joyful way. The system does not give advice in a unilateral way; instead it helps the user, with an accurate awareness of the real world and then shows the risk information e.g., prevention plans and evac-uation maps. Additionally, introducing game elements, the user plays with some level of interaction. Using this system, we created a game application for an earthquake. An assessment experiment of the game was clearly beneficial to not only understand risk perception but support it; it also has the motivation of a muster drill.
4) Communication System for Supporting Information Gathering and Social Interac-tion in a Niche Market
We describe a communication system by which niche people can obtain cross-cutting information and communicate with other people based on each personality. The system graphically displays the degree and direction of other people's hobbies who are interested in the keyword niche people input, and relation between the knowledge e.g. movies, music, animation, history, geography using nodes. So, we can search friends who have similar interest and direction in hobbies. From a demonstra-tion experiment, we obtained good results that the system could help niche people to gain and exchange useful information.