Ensemble Based Temporal Weighting and Pareto Ranking (ETP) Model for Effective Root Cause Analysis

Root-cause identification plays a vital role in business decision making by providing effective future directions for the organizations. Aspect extraction and sentiment extraction plays a vital role in identifying the root-causes. This paper proposes the Ensemble based temporal weighting and pareto ranking (ETP) model for Root-cause identification. Aspect extraction is performed based on rules and is followed by opinion identification using the proposed boosted ensemble model. The obtained aspects are validated and ranked using the proposed aspect weighing scheme. Pareto-rule based aspect selection is performed as the final selection mechanism and the results are presented for business decision making. Experiments were performed with the standard five product benchmark dataset. Performances on all five product reviews indicate the effective performance of the proposed model. Comparisons are performed using three standard state-of-the-art models and effectiveness is measured in terms of F-Measure and Detection rates. The results indicate improved performances exhibited by the proposed model with an increase in F-Measure levels at 1%–15% and detection rates at 4%–24% compared to the state-of-the-art models.     

Analysing the roles of CEO's financial literacy and financial constraints on Spanish SMEs technological innovation

Drawing on human capital and upper echelons theories, this study analyses how CEO's financial literacy influences a firm's technological innovation and investigates the mediating role of alleviating financial constraints of Small and Medium-sized Enterprises (SMEs) in the former relationship. We develop and test hypotheses applying a Structural Equation Model to a sample of 310 Spanish SMEs. The results show that CEO's financial literacy exerts both a direct and an indirect impact, through alleviating financial constraints, on a firm's technological innovation.     

Structure constrained image completion by geometric transformation model and dynamic patches

Image completion aims to fill in a damaged region (hole) using texture form known regions. In this study, a novel image completion method is proposed that can repair a hole region using preserved structure. A geometric transformation model is first proposed to determine the relation between patches. Then, a multi-layer dynamic patch solution is obtained that can both effectively recover damaged structure and expedite the algorithm. To obtain satisfactory results, image completion is reduced to an energy optimization problem. An objective function is established, and the optimal solution is obtained. The proposed method is applied in various scenes, and the results demonstrate that this method has strong potential as an image completion technique.     

A stress-influence-function with adaptive strength feature approach for stress constrained continuum topology optimization via small-loop sequential strategy

Recently, the stress-influence-function (SIF) approach is presented for stress constrained continuum topology optimization. The SIF approach provides an alternative for continuum topology optimization with stress constraints. However, the SIF approach is not good at controlling the maximum stress of the elements compared to the conventional approach. In the study, the stress-influence-function with adaptive strength feature (SIF-ASF) approach via small-loop sequential strategy is proposed to achieve better control on the maximum elemental stress. First, the stress constrained continuum topology optimization formulation is given and the SIF approach is briefly introduced. Then the SIF-ASF approach is proposed for stress constrained continuum topology optimization, in which the strength feature in the stress influence function is adjusted in each iterative step of the optimization process. The adjoint-vector based sensitivity analysis to the design variables is further discussed. Three numerical examples are given to illustrate the applicability and validity of the proposed SIF-ASF approach. It is shown that the proposed SIF-ASF approach can achieve better control on the maximum elemental stress than the SIF approach. Moreover, the proposed SIF-ASF approach may obtain a lighter structure than the conventional approach.     

A multiobjective topology optimization approach for cost and time minimization in additive manufacturing

The ever-present drive for increasingly high-performance designs realized on shorter timelines has fostered the need for computational design generation tools such as topology optimization. However, topology optimization has always posed the challenge of generating difficult, if not impossible to manufacture designs. The recent proliferation of additive manufacturing technologies provides a solution to this challenge. The integration of these technologies undoubtedly has the potential for significant impact in the world of mechanical design and engineering. This work presents a new methodology which mathematically considers additive manufacturing cost and build time alongside the structural performance of a component during the topology optimization procedure. Two geometric factors, namely, the surface area and support volume required for the design, are found to correlate to cost and build time and are controlled through the topology optimization procedure. A novel methodology to consider each of these factors dynamically during the topology optimization procedure is presented. The methodology, based largely on the use of the spatial gradient of the density field, is developed in such a way that it does not leverage the finite element discretization scheme. This work investigates a problem that has not yet been explored in the literature: direct minimization of support material volume in density-based topology optimization. The entire methodology is formulated in a smooth and differentiable manner, and the sensitivity expressions required by gradient based optimization solvers are presented. A series of example problems are provided to demonstrate the efficacy of the proposed methodology.     

Co-Evolution in the Successful Learning of Backgammon Strategy

Following Tesauro's work on TD-Gammon, we used a 4,000 parameter feedforward neural network to develop a competitive backgammon evaluation function. Play proceeds by a roll of the dice, application of the network to all legal moves, and selection of the position with the highest evaluation. However, no backpropagation, reinforcement or temporal difference learning methods were employed. Instead we apply simple hillclimbing in a relative fitness environment. We start with an initial champion of all zero weights and proceed simply by playing the current champion network against a slightly mutated challenger and changing weights if the challenger wins. Surprisingly, this worked rather well. We investigate how the peculiar dynamics of this domain enabled a previously discarded weak method to succeed, by preventing suboptimal equilibria in a meta-game of self-learning.     

折棍变分贝叶斯图像分割算法

为提高图像分割的抗噪鲁棒性并解决分割数目的自适应确定问题,通过在聚类标签先验概率的折棍构造过程中建立Markov随机场,将空间相关性约束引入Dirichlet过程混合模型的概率建模,使聚类的空间平滑性得以增强,并采用变分推断方法获得聚类标签的收敛解析解,提出一种基于折棍变分贝叶斯推断的图像分割算法,实现了对像素聚类标签和分割数目的同步自适应学习,避免了传统方法中因引入空间相关性约束而出现的计算复杂问题.基于Berkeley BSD500图像测试数据集的数值实验结果表明,该算法具有比现有的混合模型聚类图像分割算法更高的PRI值,且在低于0.1的噪声方差条件下表现出了更优的抗噪鲁棒性.     

复杂背景下基于深度学习的视频动作识别

快速有效地识别视频中的人体动作,具有广泛的应用前景及潜在的经济价值。但目前的视频动作识别方法易受到运动人体晃动、背景变化、摄相机抖动、运动人体阴影等背景因素影响。为解决上述问题,本文提出一种非局域时间段网络方法。该方法在双流网络的基础上,通过加入非局域计算使网络能关注到更大时空范围的信息,并进一步融入光流信息使网络更精确地将注意力放在动作区域,从而增强对视频复杂静态背景的鲁棒性。此外,为了融合双流分段网络的多路判别结果,本文使用可学习的加权平均取代简单平均来融合多模态信息。经过在TDAP数据集上的实验验证,本文的模型可在复杂背景下较为精确地识别出人体动作,与原有模型相比在几乎不增加时间复杂度的前提下提升了识别性能。     

A particle filter via constrained sampling for nonlinear dynamic systems

This article is concentrated on the particle filtering problem for nonlinear systems with nonlinear equality constraints. Considering the constraint information incorporated into filters can improve the state estimation accuracy, we propose an adaptive constrained particle filter via constrained sampling. First, in order to obtain particles drawn from the constrained important density function, we construct and solve a general optimization function theoretically fusing equality constraints and the importance density function. Furthermore, to reduce the computation time caused by the number of particles, the constrained Kullback‐Leiler distance sampling method is given to online adapt the number of particles needed for state estimation. A simulation study in the context of road‐confined vehicle tracking demonstrates that the proposed filter outperforms the typical constrained ones for equality constrained dynamic systems.