Subject: Engineering, Automotive Engineering Keywords: DQN Algorithm; Policy Modeling; Prior Knowledge; Intelligent Decision
Online: 31 August 2020 (04:08:04 CEST)
The reinforcement learning problem of complex action control in the Multi-player wargame is a hot research topic in recent years. In this paper , a game system based on turn-based confrontation is designed and implemented with the state-of-the-art deep reinforcement learning models. Specifically, we first design a Q-learning algorithm to achieve intelligent decision-making, which is based the DQN(Deep Q Network) to model the complex game behaviors. Then, a priori- knowledge based algorithm PK-DQN(Prior Knowledge- Deep Q Network) is introduced to improve the DQN algorithm, which accelerates the convergence speed and stability of the algorithm. The experiments demonstrate, the correctness of the PK-DQN algorithm is validated and its performance surpass the conventional DQN algorithm. Furthermore, the PK-DQN algorithm shows effectiveness in defeating the high level of rule-based opponents, which provides promising results for the exploration of the field of smart chess and intelligent game deduction.
ARTICLE | doi:10.20944/preprints201805.0017.v1
Subject: Social Sciences, Econometrics & Statistics Keywords: Bayesian approach; conjugate prior; cartel; leniency program; policy simulation
Online: 2 May 2018 (08:37:32 CEST)
Cartels cause tremendous damage to the market economy and disadvantage consumers by causing higher prices and lower quality; moreover, they are difficult to detect. We need to prevent them by scientific analysis, which includes the determination of an indicator to explain antitrust enforcement. Particularly, the probability of cartel penalization is a useful indicator for the evaluation of the competition enforcement. This study is to estimate the probability of cartel penalization by using a Bayesian approach. In the empirical study, the probability of cartel penalization is estimated by Bayesian approach from cartel data of Department of Justice in United States from 1970 to 2009. The probability of cartel penalization is seen to be sensitive to change of competition law and the results shows the usefulness of higher interpretation than other research. The result of the policy simulation shows how effective the leniency program is. From this estimation, antitrust enforcement is evaluated, and thereby, can be improved.
ARTICLE | doi:10.20944/preprints202111.0293.v1
Subject: Social Sciences, Education Studies Keywords: representations; prior knowledge; blended learning; scaling up; innovation and entrepreneurship
Online: 16 November 2021 (14:32:24 CET)
Education on Innovation and Entrepreneurship (I&E) has increased in the last two decades, specially, through MOOCs. Lately, these reusable online alternatives have tended to be revalorized by HEIs into blended learning activities, posing new challenges for instructors, specially, on how to bridge prior knowledge with in-class activities. Adopting a discursive approach to knowledge, our proposal aims to meet this challenge by identifying student’s ‘representations’, i.e., patterned constructions on disciplinary knowledge. Representations can be found across different cohorts and thus further complemented by instructors. To test this assumption and build our proposal, we analysed student’s representations in two observations. We mapped students’ representations over key I&E definitions (e.g., ‘start-up’) and, to know how prior knowledge may be complemented by instructors, we identified students’ alignment with expert disciplinary knowledge. Firstly, we found that the two cohorts tended to express representations by turning attention to several dimensions, e.g., referring to different types of features or finalities associated with concepts. Secondly, the disciplinary alignment description revealed that students tended to focus on the same components present in experts’ definitions, but with a greater level of generality. Our results have been packaged into a proposal that aims to help instructors scale their blended activities.
ARTICLE | doi:10.20944/preprints202105.0543.v1
Subject: Engineering, Electrical & Electronic Engineering Keywords: Blind Source Separation (BSS), Minimum Mean Square Error (MMSE), convolutive mixture, source Prior, generalized Gaussian distribution
Online: 24 May 2021 (08:50:37 CEST)
This paper proposes a novel efficient multistage algorithm to extract source speech signals from a noisy convolutive mixture. The proposed approach comprises of two stages named Blind Source Separation (BSS) and De-noising. A hybrid source prior model separates the source signals from the noisy reverberant mixture in the BSS stage. Moreover, we model the low and high-energy components by generalized multivariate Gaussian and super-Gaussian models, respectively. We use Minimum Mean Square Error (MMSE) to reduce noise in the noisy convolutive mixture signal in the de-noising stage. Furthermore, two proposed models investigate the performance gain. In the first model, the speech signal is separated from the observed noisy convolutive mixture in the BSS stage, followed by suppression of noise in the estimated source signals in the de-noising module. In the second approach, the noise is reduced using the MMSE filtering technique in the received noisy convolutive mixture at the de-noising stage, followed by separation of source signals from the de-noised reverberant mixture at the BSS stage. We evaluate the performance of the proposed scheme in terms of signal-to-distortion ratio (SDR) with respect to other well-known multistage BSS methods. The results show the superior performance of the proposed algorithm over the other state-of-the-art methods.
ARTICLE | doi:10.20944/preprints201805.0045.v1
Subject: Engineering, Electrical & Electronic Engineering Keywords: robust principal component analysis; video separation; compressive measurements; prior information; optical flow; motion estimation; motion compensation
Online: 2 May 2018 (13:19:49 CEST)
In the context of video background-foreground separation, we propose a compressive online Robust Principal Component Analysis (RPCA) with optical flow that separates recursively a sequence of video frames into foreground (sparse) and background (low-rank) components. This separation method can process per video frame from a small set of measurements, in contrast to conventional batch-based RPCA, which processes the full data. The proposed method also leverages multiple prior information by incorporating previously separated background and foreground frames in an n-l1 minimization problem. Moreover, optical flow is utilized to estimate motions between the previous foreground frames and then compensate the motions to achieve higher quality prior foregrounds for improving the separation. Our method is tested on several video sequences in different scenarios for online background-foreground separation given compressive measurements. The visual and quantitative results show that the proposed method outperforms other existing methods.
ARTICLE | doi:10.20944/preprints202109.0399.v2
Subject: Mathematics & Computer Science, Applied Mathematics Keywords: perturbed nonlinear schrodinger equation; conformal fractional derivative; prior estimate; bifurcation method; complete discrimination system for polynomial method
Online: 24 November 2021 (13:04:55 CET)
The main idea of this paper is to investigate the exact solutions and dynamic properties in optical nanofibers, which is modeled by space-time fractional perturbed nonlinear schr\"odinger equation involving Kerr law nonlinearity with conformal fractional derivative. Firstly, by the complex fractional traveling wave transformation, the traveling wave system of the original equation is obtained, then a conserved quantity, namely the Hamiltonian is constructed, and the qualitative analysis of this system is conducted via this quantity by classifying the equilibrium points. Moreover, the prior estimate of the existence of the soliton and periodic solution is established via the bifurcation method. Furthermore, all exact traveling wave solutions are constructed to illustrate our results explicitly by the complete discrimination system for polynomial method.
SHORT NOTE | doi:10.20944/preprints202206.0032.v1
Subject: Life Sciences, Other Keywords: Conformity assessment; lot inspection; acceptance sampling; Quality level; sample size; Bayesian statistics; prior distribution; posterior distribution; consumer risk; producer risk
Online: 2 June 2022 (10:59:47 CEST)
The ISO 2859 and ISO 3951 series provide acceptance sampling procedures for lot inspection, allowing both sample size and acceptance rule to be determined, starting from a specific value either for the consumer or producer risk. However, insufficient resources often make it difficult to implement “ISO sampling plans.” In cases where the sample size is determined by external constraints, the focus shifts from determining sample size to determining consumer and producer risks. Moreover, if the sample size is very low (e.g. one single item), prior information should be included in the statistical analysis. For this reason, it makes sense to work within a Bayesian theoretical framework, such as that described in JCGM 106. Accordingly, the approach from JCGM 106 is adopted and broadened so as to allow application to lot inspection. The discussion is based on a “real-life” example of lot inspection on the basis of a single item. Starting from simple assumptions, expressions for both the prior and posterior distributions are worked out, and it is shown how the concepts from JCGM 106 can be reinterpreted in the context of lot inspection. Conceptual differences regarding the definition of consumer and producer risks in JCGM 106 and in the ISO acceptance sampling standards are elucidated and a numerical example is provided.
ARTICLE | doi:10.20944/preprints201705.0043.v1
Subject: Mathematics & Computer Science, Information Technology & Data Management Keywords: Bayesian reliability analysis; Bayesian hierarchical model; MCMC method; scale mixtures of log-normal failure time model; stochastic constraint; two-stage MaxEnt prior.
Online: 4 May 2017 (16:26:10 CEST)
This paper develops Bayesian inference in reliability of a class of scale mixtures of log-normal failure time (SMLNFT) models with stochastic (or uncertain) constraint in their reliability measures. The class is comprehensive and includes existing failure time (FT) models (such as log-normal, log-Cauchy, and log-logistic FT models) as well as new models that are robust in terms of heavy-tailed FT observations. Since classical frequency approaches to reliability analysis based on the SMLNFT model with stochastic constraint are intractable, the Bayesian method is pursued utilizing a Markov chain Monte Carlo (MCMC) sampling based approach. This paper introduces a two-stage maximum entropy (MaxEnt) prior, which elicits a priori uncertain constraint and develops Bayesian hierarchical SMLNFT model by using the prior. The paper also proposes an MCMC method for Bayesian inference in the SMLNFT model reliability and calls attention to properties of the MaxEnt prior that are useful for method development. Finally, two data sets are used to illustrate how the proposed methodology works.