Speaker: Adam Milstein
Localization is the problem of determining a s location in an environment. Monte Carlo Localization (MCL) is a method of solving this problem by using a partially observable Markov decision process to find the s state based on its sensor readings, given a static map of the environment. MCL requires a model of each sensor in order to work properly. One of the most important sensors involved is the estimation of the s motion, based on its encoders that report what motion the robot has performed.
Future students
Speaker: Edward Chrzanowski
This talk will present the path I traversed from metaheuristics to parallel metaheuristics in the search of a better scheduler for High Performance Computing.
Speaker: Milen Pavlov
When applying reinforcement learning to real world problems it is desirable to make use of any prior information to speed up the learning process. Unfortunately, there does not exist any framework for encoding prior knowledge in RL nor developing general algorithms that make use of it.
Speaker: Pascal Poupart
In this presentation, I will talk about the oldest form of Reinforcement Learning (RL): Bayesian Reinforcement Learning (BRL).
Speaker: Harold Boley (University of New Brunswick and Institute for Information Technology, National Research Council of Canada)
The Semantic Web contributes to making AI mainstream. AI knowledge representations such as ontologies and rules can be used to formalize Web metadata vocabularies, where constants, classes, and relations carry (Uniform Resource Identifier) URIs. This permits AI reasoning over metadata both for direct question answering and high-precision information (URI) retrieval.
Speaker: Salvador Garcia-Martinez (McMaster University)
Through the development of Multi-Agent Systems, different architectures for agent systems have emerged. They follow different models and standards, but at the same time they share something in common: elements and relations. In this talk, I will discuss different multi-agent architectures that have been developed, focusing mainly on their similarities and how they have been implemented.
Speaker: Saeed Hassanpour
Recognition of cursive handwritings such as Persian script is a hard task as there is no fixed segmentation and simultaneous segmentation and recognition is required. We presents a novel comparison method for such tasks which is based on a Multiple States Machine to perform robust elastic comparison of small segments with high speed through generation and maintenance of a set of concurrent possible hypotheses.
Speaker: Jie Zhang
Familiarity between agents is often considered to be an important factor in determining the level of trust. In electronic marketplaces, trust is modeled, for instance, in order to allow buying agents to make effective selection of selling agents. In previous research, familiarity between two agents has been simply assumed to be the similarity between them, which is fixed for the two agents.
Speaker: Jiye Li
Clickstream data collected across multiple websites (user-centric data) captures users' browsing behaviors, interests and preferences more than clickstream data collected from individual websites (site-centric data). For example, we would expect that we could better model and predict the intentions of users who we know not only searched on google but also visited certain shopping websites, than if we know only one of these pieces of information. Current research on clickstream data analysis is mostly centered around site-centric data.
Speaker: Feng Jiao
We present a new semi-supervised training procedure for conditional random fields (CRFs) that can be used to train sequence segmentors and labelers from a combination of labeled and unlabeled training data.