Real-time three-dimensional visualization for seismic data is difficult when seismic data are large-scale and usually exceed the limitation of host memories. This paper proposed a dynamic caching framework based on OC...
详细信息
We consider a modular method to reinforcement learning that represents uncertainty of model parameters by maintaining probability distributions over them. The algorithm we call MBDP (model-based Bayesian dynamic progr...
详细信息
ISBN:
(纸本)9781509001644
We consider a modular method to reinforcement learning that represents uncertainty of model parameters by maintaining probability distributions over them. The algorithm we call MBDP (model-based Bayesian dynamic programming) can be decomposed into two parallel types of inference: model learning and policy learning. During learning a model, we update posterior distributions of a model over observations after taking an action in each state. During learning a policy, we solve MDPs by dynamic programming with greedy approximation to make an agent choose behaviors which maximize return under the estimated model. Furthermore, we propose a principled method which utilizes the variance of Dirichlet distributions for determining when to learn and relearn the model. We demonstrate that MBDP can find near optimal policies with high probability by sufficient model learning and experimental results show that MBDP performs better compared with current state-of-the-art methods in reinforcement learning.
In this paper, we investigate the problem of image de-noising. Here, the theory of morphological component analysis is employed to separate the image to be de-noised into some layers with different morphological compo...
详细信息
In this paper, a new image de-noising algorithm based on directional bi-dimensional empirical mode decomposition. Attractive features of this algorithm include its data driven mechanism and its ability of capturing di...
详细信息
In this paper, a new image segmentation algorithm based on Otsu thresholding. One of attractive feature of this algorithm is its ability of processing noised images. The framework contains three steps: the image to be...
详细信息
Sensor optimization is the problem of minimizing sensor activation in a controlled discrete event system. During the evolutionary process, the available resources are supposed to be limited. Therefore, sensors are act...
详细信息
Sensor optimization is the problem of minimizing sensor activation in a controlled discrete event system. During the evolutionary process, the available resources are supposed to be limited. Therefore, sensors are activated by the agent when it is necessary. Sensor activation policies are the functions that determine which sensors are to be activated. One policy is considered to minimal, if any strictly less activation decided by the agent satisfies the feasibility. In this paper, a new algorithm is proposed to compute the minimal sensor activation policy. The algorithm, based on the operation of Reverse Change and the property of the Label-reached, calculates the minimal solution of sensor activation and achieves a lower complexity of computation relatively.
Epistasis or the interaction of single nucleotide polymorphisms (SNPs) at different loci plays a significant role on the mechanisms and pathogenesis of many common complex, multifactorial diseases and may be responsib...
详细信息
Data compression does not only save space for data storage, but also improve its safety and efficiency during data transport. As any data can be saved in the form of an integer directly or indirectly, it is a meaningf...
详细信息
Classification in networked data is a popular research of complex network. Because of the large scale of networked data and the shortage of training data, active learning, which is an effective classification method f...
详细信息
ISBN:
(纸本)9781510819085
Classification in networked data is a popular research of complex network. Because of the large scale of networked data and the shortage of training data, active learning, which is an effective classification method for sparse data in machine learning, is often applied to networked data classification problems. Introduced in this paper are classification methods based on active learning for networked data classification problems with basic concepts and algorithms. Finally, according to the existing research, some problems for the future developing and research of networked data classification issues is presented.
The application of intelligent geophysical interpreting, technology of mapping and database building based on GIS have been discussed under the guidance of the theory and the method of the mineral resources prediction...
详细信息
暂无评论