In multitask reinforcement learning, tasks often have sub-tasks that share the same solution, even though the overall tasks are different. If the shared-portions could be effectively identified, then the learning process could be improved since all the samples between tasks in the shared space could be used. In this paper, we propose a Sharing Experience Framework (SEF) for simultaneously training of multiple tasks. In SEF, a confidence sharing agent uses task-specific rewards from the environment to identify similar parts that should be shared across tasks and defines those parts as shared-regions between tasks. The shared-regions are expected to guide task-policies sharing their experience during the learning process. The experiments highlight that our framework improves the performance and the stability of learning task-policies, and is possible to help task-policies avoid local optimums.
This paper discusses and proposes a rough set model for an incomplete information system, which defines an extended tolerance relation using frequency of attribute values in such a system. It first discusses some rough set extensions in incomplete information systems. Next, "probability of matching" is defined from data in information systems and then measures the degree of tolerance. Consequently, a rough set model is developed using a tolerance relation defined with a threshold. The paper discusses the mathematical properties of the newly developed rough set model and also introduces a method to derive reducts and the core.
The original rough set theory deals with precise and complete data, while real applications frequently contain imperfect information. A typical imperfect data studied in rough set research is the missing values. Though there are many ideas proposed to solve the issue in the literature, the paper adopts a probabilistic approach, because it can incorporate other types of imperfect data including imprecise and uncertain values in a single approach. The paper first discusses probabilities of attribute values assuming different type of attributes in real applications, and proposes a generalized method of probability of matching. It also discusses the case of continuous data as well as discrete one. The proposed probability of matching could be used for defining valued tolerance/similarity relations in rough set approaches.
The paper introduces a rough set model to analyze an information system in which some conditions and decision data are missing. Many studies have focused on missing condition data, but very few have accounted for missing decision data. Common approaches tend to remove objects with missing decision data because such objects are apparently considered worthless from the perspective of decision-making. However, we indicate that this removal may lead to information loss. Our method retains such objects with missing decision data. We observe that a scenario involving missing decision data is somewhat similar to the situation of semi-supervised learning, because some objects are characterized by complete decision data whereas others are not. This leads us to the idea of estimating potential candidates for the missing data using the available data. These potential candidates are determined by two quantitative indicators: local decision probability and universal decision probability. These potential candidates allow us to define set approximations and the definition of reduct. We also compare the reducts and rules induced from two information systems: one removes objects with missing decision data and the other retains such objects. We highlight that the knowledge induced from the former can be induced from the latter using our approach. Thus, our method offers a more generalized approach to handle missing decision data and prevents information loss.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations鈥揷itations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.