• KIISE Transactions on Computing Practices
• /
• v.23 no.10
• /
• pp.599-604
• /
• 2017

A useful application of smart assistants is to predict and suggest users' daily behaviors the way real assistants do. Conventional methods to predict behavior have mainly used explicit schedule information logged by a user or extracted from e-mail or SNS data. However, gathering explicit information for smart assistants has limitations, and much of a user's routine behavior is not logged in the first place. In this paper, we suggest a novel approach that combines explicit schedule information with patterns of routine behavior. We propose using inference based on a Markov decision process and learning with a reward function based on inverse reinforcement learning. The results of our experiment shows that the proposed method outperforms comparable models on a life-log dataset collected over six weeks.

• Nuclear Engineering and Technology
• /
• v.25 no.1
• /
• pp.154-165
• /
• 1993

A stochastic method using continuous time Markov process is presented to model the one-dimensional convective nuclide transport in geologic media, which have usually heterogeneous feature in physical/geochemical parameters such as velocity, dispersion coefficient, and retardation factor resulting poor description by conventional deterministic advection-dispersion model. The primary desired quantities from a stochastic model are the mean values and variance of the state variables as a function of time. The time-dependent probability distributions of nuclides are presented for each discretized compartment given the volumetric groundwater flux and the intensity of transition. Since this model is discrete in medium space, physical/geochemical parameters which affect nuclide transport can be easily incorporated for the heterogeneous media as well as remarkably layered media having spatially varied parameters. Even though the Markov process model developed in this study was shown to be sensitive to the number of discretized compartments showing numerical dispersion as the number of compartments are increased, this could be easily calibrated by comparing with the analytical deterministic model.

• Journal of Internet Computing and Services
• /
• v.10 no.3
• /
• pp.85-94
• /
• 2009

In this work we propose an energy-efficient transmission strategy for wireless sensor networks that operate in a strict energy-constrained environment. Our transmission algorithm consists of two components: a binary-decision based transmission and a channel-aware backoff adjustment. In the binary-decision based transmission, we obtain the optimum threshold for successful transmission via Markov decision process (MDP) formulation. A channel-aware backoff adjustment, the second component of our proposal, is introduced to favor sensor nodes seeing better channel in terms of transmission priority. Extensive simulations are performed to verify the performance of our proposal over fading wireless channels.

• Nuclear Engineering and Technology
• /
• v.25 no.4
• /
• pp.529-538
• /
• 1993

A stochastic way using continuous time Markov process is presented to model the one-dimensional nuclide transport in fractured rock matrix as an extended study for previous work ［1］. A nuclide migration model by the continuous time Markov process for single planar fractured rock matrix, which is considered as a transient system where a process by which the nuclide is diffused into the rock matrix from the fracture may be no more time homogeneous, is compared with a conventional deterministic analytical solution. The primary desired quantities from a stochastic model are the expected values and variance of the state variables as a function of time. The time-dependent probability distributions of nuclides are presented for each discretized compartment of the medium given intensities of transition. Since this model is discrete in medium space, parameters which affect nuclide transport could be easily incorporated for such heterogeneous media as the fractured rock matrix and the layered porous media. Even though the model developed in this study was shown to be sensitive to the number of discretized compartment showing numerical dispersion as the number of compartments are decreased, with small compensating of dispersion coefficient, the model agrees well to analytical solution.

• Nuclear Engineering and Technology
• /
• v.25 no.4
• /
• pp.539-547
• /
• 1993

A stochastic approach using continuous time Markov process is presented to model the one-dimensional nuclide transport in fractured rock media as a further extension for previous works［1-3］. Nuclide transport of decay chain of arbitrary length in the single planar fractured rock media in the vicinity of the radioactive waste repository is modeled using a continuous time Markov process. While most of analytical solutions for nuclide transport of decay chain deal with the limited length of decay chain, do not consider the case of having rock matrix diffusion, and have very complicated solution form, the present model offers rather a simplified solution in the form of expectance and its variance resulted from a stochastic modeling. As another deterministic way, even numerical models of decay chain transport, in most cases, show very complicated procedure to get the solution and large discrepancy for the exact solution as opposed to the stochastic model developed in this study. To demonstrate the use of the present model and to verify the model by comparing with the deterministic model, a specific illustration was made for the transport of a chain of three member in single fractured rock medium with constant groundwater flow rate in the fracture, which ignores the rock matrix diffusion and shows good capability to model the fractured media around the repository.