Invited Talks
Invited Talk I
Title
AI Research in Japan – Towards Empathic Agents
Speaker
Host
Abstract
I will talk about the past, present and future of AI in Japan. First, I will overview a historical background and the contemporary AI-research activities in Japan. Then, I will highlight some recent prominent results from the industry. Finally, I will discuss how much information and communication technology can contribute to enhance empathy, with a particular emphasis on potential contributions from AI for creating empathy between humans and machines.
Download Presentation Slide (PDF File)
Invited Talk II
Title
Unlabeled Data and Multiple Views
Speaker
Host
Abstract
In many real applications there are abundant unlabeled data but the amount of labeled training examples are limited, since labeling the data requires extensive human effort and expertise. To effectively utilize unlabeled data to help improve performance, semi-supervised learning and active learning have attracted much attention during the past decade. It is noteworthy that when the data have multiple views, such as data with multi-modalities, helpful information contained in the multi-views will be able to enable strong process of semi-supervised learning and active learning. In this talk we will introduce some advances in this line of research.
Download Presentation Slide (PDF File)
Invited Talk III
Title
Common Sense Communications : A Considerate Supervisor for a Multiparty Conference Call
Speaker
Host
Abstract
Interactions between people occur in a social realm. On the other hand, “things”, including devices for communication and computation, are generally socially deficient. Imagine socially aware systems moving from an interruption model of communication to an introduction model. To create considerate systems, there is a need to model social context, social behavior, and communication goals.
This talk describes early systems that work to understand and eliminate the socially disruptive qualities of the ubiquitous systems people increasingly use and rely on in all aspects of their personal, educational, social and business lives. We show performance improving systems: an instant message arrives after you have finished typing a sentence, not while you are forming it; a car waits for you to complete a difficult maneuver before giving you distracting feedback
This work relies on dynamic task, user, system, and communication models. The goal is to stimulate more work to understand and create considerate systems. Such systems will improve people’s experience and performance.
Social responsiveness can become the norm for the technology that pervades our lives.
Invited Talk IV
Title
Graphical Models and Flexible Classifiers: Bridging the Gap with Boosted Regression Trees
Speaker
Host
Abstract
There have long been two parallel paths of research in machine learning. One path has studied flexible, non-parametric models such as decision tree ensembles and support vector machines. These methods can adapt the complexity of the model to the amount and complexity of the training data, which helps them avoid over-fitting (and under-fitting) to achieve highly accurate predictions. In addition, these methods are easy to use, because they do not require the analyst to choose the form of the model.
The other path has studied probabilistic graphical models (Bayesian networks, Markov random fields). These models have three advantages that non-parametric models lack: (a) the analyst can incorporate domain knowledge by specifying the structure of the graphical model, (b) the models can contain hidden (latent) variables, and (c) the models have well-understood probabilistic semantics. This talk will describe efforts at Oregon State to bridge the gap between these two paths so that we can obtain some of the advantages of graphical models (incorporating domain knowledge, using hidden variables) with the advantages of non-parametric methods (flexible, easy-to-use algorithms).
Our work builds on Jerome Friedman’s L2-TreeBoost method to represent conditional probability distributions and MRF potentials using boosted regression trees. I will discuss applications of this method to conditional random fields and ecological models.
Download Presentation Slide (PDF File)
IWCG Invited Talk
Title
Computer Shogi-- History and Techniques
Speaker
Host
Prof. Shi-Jim Yen
Abstract
Shogi is a game with the same origin as the chess and xiangqi , played in Japan. Computer shogi studies started in the 1970s. Nowadays, it has reached professional player's strength. Many technologies, peculiar to computer Shogi, have developed as well as application of chess knowledge, such as: end-game search, realization probability search, evaluation function learning, etc.
I will talk about this history of this computer Shogi with the technical development.