A Text-to-Speech Platform for Variable Length Optimal Unit Searching Using Perception Based Cost Functions

In concatenative Text-to-Speech, the size of the speech corpus is closely related to synthetic speech quality. In this paper, we describe our work on a new corpus-based Bell Labs' TTS system. This encompasses large acoustic inventories with a rich set of annotations, models and data structures for representing and managing such inventories, and an optimal unit selection algorithm that accommodates a broad range of possible cost criteria. We also propose a new method for setting weights in the cost functions based on a perceptual preference test. Our results show that this approach can successfully predict human preference patterns. Synthetic speech using weights determined in this manner consistently demonstrates smoother transitions and higher voice quality than speech using manually set weights.     

GRADIENT-DESCENT BASED UNIT-SELECTION OPTIMIZATION ALGORITHM USED FOR CORPUS-BASED TEXT-TO-SPEECH SYNTHESIS

This paper proposes a gradient-descent based unit selection optimization algorithm for the optimization of unit-cost function weights and for improving the overall performance of the unit-selection algorithm, as used in a corpus-based text-to-speech synthesis system. Complex multidimensional and fuzzy-logic based unit-cost functions are used in the presented unit-selection algorithm. The weights used by these unit-cost functions are usually defined by heuristics or by listening tests. This can be very laborious and time consuming, and does not necessarily result in an optimal performance of the unit-selection algorithm because of multidimensional unit-cost function space, within which different database candidates’ features are evaluated. Using heuristics or listening tests is also rather rigid, especially when working with several different databases or voices. It is especially difficult, within this scope, to set up those weights used in unit-cost functions in order to achieve overall optimal performance of the unit-selection algorithm. The proposed unit-selection optimization process consists of several steps. It is fully automatic, flexible, and fast enough to enable the development of a corpus-based text-to-speech (TTS) system that uses many different voices, without any heuristics or listening tests. This optimization process can also be helpful when evaluating the performances of unit-selection cost functions, and the performance of the unit-selection algorithm itself. The obtained results “suggest” those values that the unit-selection cost-function weights should have in order to obtain smoother transitions between selected unit candidates, after the unit-selection process. The obtained results also hint at the performance level that can be achieved with a given set of unit-cost function weights, and suggest what improvements can be gained when using those additional or changed unit-cost functions included within the unit-selection algorithm.     

基于用户偏好的权重搜索及告警选择方法

用户在现有交互方式下选择最为严重的告警时完全依据其个人偏好,而未考虑处理不同告警所需成本的差异性问题.为此,提出一种基于用户偏好的权重搜索及告警选择方法.挖掘用户对不同严重程度告警的偏好值,针对问题的复杂性建立评估函数,并给出偏好权重的选择策略.对不同告警及其对应的用户偏好权重建立效用函数,确定需优先解决的告警,并在成本约束下完成基于用户偏好的告警选择,提高告警处理效率.实验结果表明,该方法能够合理有效地做出告警选择,与基于背包式和设定阈值的方法相比,其告警选择的表现更优.     

面向偏好冲突的直觉模糊多属性群体决策方法

针对偏好具有冲突性且权重信息完全未知的直觉模糊多属性群体决策问题, 提出一种基于多目标决策的求解方法. 首先, 建立以决策方案差异程度和决策成员偏好冲突程度为目标函数的多目标决策模型; 然后, 利用极小极大方法求解该模型, 得到各方案的属性权重和决策成员权重, 据此确立最优方案; 最后, 通过数值算例表明了该方法的有效性.

     

考虑多种偏好信息的多阶段方案选择方法

针对方案选择过程中专家倾向于采用不同偏好信息的问题,提出了考虑多种偏好信息的方案选择方法。专家可以从方案集中选择出子集,从效用值、偏好次序、互反判断矩阵和模糊偏好关系四种形式中选择一种对方案子集进行评价;采用转化函数将专家给出的偏好信息转化为偏好向量;根据各个专家的偏好向量集结方案分类结果;将结果用于反馈调整各专家的方案子集;最后通过多个阶段的反馈与调整,改进群体一致性以获得最优方案。将该方法用于某汽车制造商汽车零部件再制造方案的选择,验证了所提方法的有效性。     

Pricing strategy in a dual-channel and remanufacturing supply chain system

This article addresses the pricing strategy problems in a supply chain system where the manufacturer sells original products and remanufactured products via indirect retailer channels and direct Internet channels. Due to the complexity of that system, agent technologies that provide a new way for analysing complex systems are used for modelling. Meanwhile, in order to reduce the computational load of searching procedure for optimal prices and profits, a learning search algorithm is designed and implemented within the multi-agent supply chain model. The simulation results show that the proposed model can find out optimal prices of original products and remanufactured products in both channels, which lead to optimal profits of the manufacturer and the retailer. It is also found that the optimal profits are increased by introducing direct channel and remanufacturing. Furthermore, the effect of customer preference, direct channel cost and remanufactured unit cost on optimal prices and profits are examined.     

Application of Genetic Algorithm in unit selection for Malay speech synthesis system

Corpus based speech synthesis can produce high quality synthetic speech due to it high sensitivity to unit context. Large speech database is embedded in synthesis system and search algorithm (unit selection) is needed to search for the optimal unit sequence. Speech feature which served as target cost is estimated from the input text. The acoustic parameters which served as join cost are derived from mel frequency cepstral coefficients (MFCCs) and Euclidean distance. In this paper, a new method which is Genetic Algorithm is proposed to search for optimal unit sequence. Genetic Algorithm (GA) is a population based search algorithm that is based on the biological principles of selection, reproduction, crossover and mutation. It is a stochastic search algorithm for solving optimization problem. The speech unit sequence that has minimum join cost will be synthesized into complete waveform data.     

On compatibility of uncertain multiplicative linguistic preference relations based on the linguistic COWGA

The aim of this work is to develop a new compatibility for the uncertain multiplicative linguistic preference relations and utilize it to determine the optimal weights of experts in the group decision making (GDM). First, the compatibility degree and compatibility index for the two multiplicative linguistic preference relations are proposed. Then, based on the linguistic continuous ordered weighted geometric averaging (LCOWGA) operator, some concepts of the compatibility degree and compatibility index for the two uncertain multiplicative linguistic preference relations are presented. We prove the property that the synthetic uncertain linguistic preference relation is of acceptable compatibility under the condition that the uncertain multiplicative linguistic preference relations given by experts are all of acceptable compatibility with the ideal uncertain multiplicative linguistic preference relation, which provides a theoretic basis for the application of the uncertain multiplicative linguistic preference relations in GDM. Next, an optimal model is constructed to determine the weights of experts based on the criterion of minimizing the compatibility index in GDM. Moreover, an approach to GDM with uncertain multiplicative linguistic preference relations is developed, and finally, an application of the approach to supplier selection problem with uncertain multiplicative linguistic preference relations is pointed out.     

大群体应急决策中考虑属性关联的偏好信息融合方法

针对大群体应急决策中属性之间相互关联的问题,提出一种新的偏好信息融合方法.首先,利用最优离散拟合模型测度专家的风险偏好,提出一种考虑专家风险偏好的犹豫模糊元补充方法;其次,运用TF-IDF算法获得相互关联的事件属性集;再次,结合传统的主成分分析法与误差理论,提出基于犹豫模糊语言的主成分分析模型,获得几个互不关联的主属性及其权重,进而进行信息集结和方案择优;最后,通过广西洪涝灾害事件验证所提方法的可行性和有效性.     

对方案有偏好的最小最大偏差多属性决策方法

针对部分属性权重信息且对方案有偏好的多属性决策问题,提出了一种新的决策方法。该方法通过求解最小化最大的主、客观偏差的数学优化模型,较合理地确定属性的权重向量。克服已有方法中不能保证每一方案的主、客观偏差都最小的问题。产品开发应用实例表明方法的有效性和实用性。     

Integrating multicriteria PROMETHEE II method into a single-layer perceptron for two-class pattern classification

PROMETHEE methods based on the outranking relation theory are extensively used in multicriteria decision aid. A preference index representing the intensity of preference for one pattern over another pattern can be measured by various preference functions. The higher the intensity, the stronger the preference is indicated. In contrast to traditional single-layer perceptrons (SLPs) with the sigmoid function, this paper develops a novel PROMETHEE II-based SLP using concepts from the PROMETHEE II method involving pairwise comparisons between patterns. The assignment of a class label to a pattern is dependent on its net preference index, which the proposed perceptron obtains. Specially, this study designs a genetic-algorithm-based learning algorithm to determine the relative weights of respective criteria in order to derive the preference index for any pair of patterns. Computer simulations involving several real-world data sets reveal the classification performance of the proposed PROMETHEE II-based SLP. The proposed perceptron performs well compared to the other well-known fuzzy or non-fuzzy classification methods.     

高显著性的时空金字塔精简描述符算法研究

当传统时空金字塔层数较多时,特征描述符的维数会非常高,使得此类描述符在训练和测试阶段计算效率非常低。此外,在时空金字塔的分层及每层立方体单元的划分中,至今仍然采用手动划分视频,使得视频划分策略没有强的理论依据。鉴于以上缺点,提出一种高显著性的时空金字塔精简描述符算法。形成的新描述符是所有时空金字塔层中每个立方体单元局部特征的权重和,而不是把所有立方体单元局部特征描述符串联起来形成一个巨大的特征描述符,每个立方体单元的权重可以通过偏最小二乘法自动获取,由此产生的视频全局描述符精简并且具有高的显著性。此外通过观测其精细立方体单元的权重,还可以展现出显著性时空金字塔每个立方体单元及每层金字塔的贡献,由此,可以根据权重自动划分视频。采用HMDB51和YouTube两个动作数据库进行实验验证,与时空金字塔描述符和超稀疏编码向量相比,此描述符精简并能在低维度下取得较好的识别效果。     

融合自动检错的单元挑选语音合成方法

提出了一种融合自动检错的单元挑选语音合成方法。本文方法旨在设计与主观听感更加一致的单 元挑选准则,以提高合成语音的自然度。首先利用众包网络平台快速大量地收集测听人对于合成语音的主观评价数据,取代了传统的利用具备语言学知识的专家收集主观评价数 据的方法;然后基于这些主观评价数据,提取对应语音的音节时长、单元代价以及声学参数距 离等特征,构建基于支持向量机的合成错误检测器;在合成阶段,该检测器被用来对传统单元 挑选输出的N条路径行重打分,以确定最优的单元挑选序列。倾向性测听结果表明本文方法可以有效地提高合成语音的自然度。     

一种基于服务关系与效能的Web服务选择方法

当前Web服务选择是一个研究热点,很多研究人员提出根据用户对QoS（Quality of Service）属性的偏爱,把指定属性权值计算QoS属性综合值作为服务选择依据。实际应用中,用户对服务的偏爱是主观的,在多备选服务,多QoS属性参照的条件下主观性被进一步放大,会导致所选择的服务并不是最适合的。同时,服务之间在社会层次的相互关系并未在服务选择中被考虑。针对上述两个问题,文中首先设计了Web服务选择二维度量属性模型,引入了社会域的服务作用关系;其次通过QoS约束条件排除不满足用户要求的服务;最后设计了一种基于服务效能偏爱的服务选择方法,并通过实验数据验证了该服务选择方法的可行性、有效性和客观性。     

基于物理规划的航空保障多目标优化模型

根据作战需求确定优化权衡目标,以使用保障需求和技术可行性为约束条件,以保障能力指标集合,以备件储备量的组合为决策变量,建立多个单目标优化模型和多目标优化模型,提出了基于物理规划的航空保障能力多目标优化方法,结合粒子群算法得到了满足设计者偏好的最佳航材备件方案.设计了航空保障能力系数的满意等级,构造了各优化目标的偏好函数和综合偏好函数,使整个设计过程更加灵活地反映决策者偏好,减轻大规模多目标设计问题的计算负担,使军用机群的保障能力更适合实际作战要求,同时通过对比单目标优化的结果,验证了算法的有效性.     

An integrated model for closed-loop supply chain configuration and supplier selection: Multi-objective approach

Reverse logistics consists of all operations related to the reuse of products. External suppliers are one of the important members of reverse logistics and closed loop supply chain (CLSC) networks. However in CLSC network configuration models, suppliers are assessed based on purchasing cost and other factors such as on-time delivery are ignored. In this research, a general closed loop supply chain network is examined that includes manufacturer, disassembly, refurbishing, and disposal sites. Meanwhile, it is managed by the manufacturer. We propose an integrated model which has two phases. In the first phase, a framework for supplier selection criteria in RL is proposed. Besides, a fuzzy method is designed to evaluate suppliers based on qualitative criteria. The output of this stage is the weight of each supplier according to each part. In the second phase, we propose a multi objective mixed-integer linear programming model to determine which suppliers and refurbishing sites should be selected (strategic decisions), and find out the optimal number of parts and products in CLSC network (tactical decisions). The objective functions maximize profit and weights of suppliers, and one of them minimizes defect rates. To our knowledge, this model is the first effort to consider supplier selection, order allocation, and CLSC network configuration, simultaneously. The mathematical programming model is validated through numerical analysis.     

Utility-driven solution for optimal resource allocation in computational grid

Optimal resource allocation is a complex undertaking due to large-scale heterogeneity present in computational grid. Traditionally, the decision based on certain cost functions has been used in allocating grid resource as a standard method that does not take resource access cost into consideration. In this paper, the utility function is presented as a promising method for grid resource allocation. To tackle the issue of heterogeneous demand, the user's preference is represented by utility function, which is driven by a user-centric scheme rather than system-centric parameters adopted by cost functions. The goal of each grid user is to maximize its own utility under different constraints. In order to allocate a common resource to multiple bidding users, the optimal solution is achieved by searching the equilibrium point of resource price such that the total demand for a resource exactly equals the total amount available to generate a set of optimal user bids. The experiments run on a Java-based discrete-event grid simulation toolkit called GridSim are made to study characteristics of the utility-driven resource allocation strategy under different constraints. Results show that utility optimization under budget constraint outperforms deadline constraint in terms of time spent, whereas deadline constraint outperforms budget constraint in terms of cost spent. The conclusion indicates that the utility-driven method is a very potential candidate for the optimal resource allocation in computational grid.     

A maximizing consensus approach for alternative selection based on uncertain linguistic preference relations

In alternative selection problems managed by multiple experts in uncertain situations achieving consensus is a desirable objective as incorrect selection may adversely affect stakeholder outcomes. This paper develops an approach to solve consensus problems when expert preference information is in the form of uncertain linguistic preference relations. First, definitions for aggregation operators and group consensus level based on a 2-tuple linguistic representation model are provided. Then, in order to obtain the weights of the experts under the assumption of incomplete weights information, an optimization model is developed which seeks maximum consensus from the current expert preferences in the group. If the consensus level reached does not meet predefined requirements, a consensus reaching algorithm is presented which can automatically achieve the goal. To determine the parameters for the proposed algorithm, a simulation procedure is presented. Finally, an investment company optimal selection example is provided to show the properties of the proposed approach. A comparative study and discussion of the proposed approach are also conducted.     

Convergence analysis of a computational method for optimal control of non-linear differential-algebraic systems

The convergence analysis of a computational method for optimal control problems of non-linear differential-algebraic systems is considered. The class of admissible controls is taken to be the class of piecewise smooth functions. A control parametrizution technique is used to approximate the optimal control problem into a sequence of optimal parameter selection problems. The solution of each of these approximate problems gives rise to a suboptimal solution to the original optimal control problem in an obvious way. The gradients of the cost functional with respect to parameters are derived. Furthermore, the error bounds between the suboptimal costs and the true optimal cost are derived.     

Optimal system selection in fuzzy environment

The problem of the selection of an optimal decision from a set of finite number of alternatives is considered, when there is a single or multiple decision-makers present. In the single decision-maker case, the optimal decision is the one that satisfies pre-specified multiple objectives. A method is given to select the optimal decision based on membership grade. However, in the multiple decision-makers case, the problem of arriving at group decision, when each decision-maker gives his preference, has been discussed. Nevertheless, in both the situations decision-makers have the freedom to modulate their decision functions by some measure of their satisfaction or preference. Results from the theory of fuzzy sets have been applied to develop the solution methods for different situations. Illustrations are included to show the usage of the methodology.