Voice conversion using General Regression Neural Network

The objective of voice conversion system is to formulate the mapping function which can transform the source speaker characteristics to that of the target speaker. In this paper, we propose the General Regression Neural Network (GRNN) based model for voice conversion. It is a single pass learning network that makes the training procedure fast and comparatively less time consuming. The proposed system uses the shape of the vocal tract, the shape of the glottal pulse (excitation signal) and long term prosodic features to carry out the voice conversion task. In this paper, the shape of the vocal tract and the shape of source excitation of a particular speaker are represented using Line Spectral Frequencies (LSFs) and Linear Prediction (LP) residual respectively. GRNN is used to obtain the mapping function between the source and target speakers. The direct transformation of the time domain residual using Artificial Neural Network (ANN) causes phase change and generates artifacts in consecutive frames. In order to alleviate it, wavelet packet decomposed coefficients are used to characterize the excitation of the speech signal. The long term prosodic parameters namely, pitch contour (intonation) and the energy profile of the test signal are also modified in relation to that of the target (desired) speaker using the baseline method. The relative performances of the proposed model are compared to voice conversion system based on the state of the art RBF and GMM models using objective and subjective evaluation measures. The evaluation measures show that the proposed GRNN based voice conversion system performs slightly better than the state of the art models.     

多粒度支持直觉模糊粗糙集的多属性决策方法

针对多属性决策中多个相互冲突的属性信息使决策者很难做出决策判断的问题,文中从支持直觉模糊集的角度研究该问题.首先,在支持直觉模糊集的基础上,结合多粒度粗糙集理论,构造乐观、悲观两种多粒度支持直觉模糊粗糙集模型,分析两种模型之间的相互关系,讨论相关性质.然后,利用t-模和t-余模定义拟合函数,提出多粒度支持直觉模糊粗糙集的多属性决策求解方法,同时定义得分函数和精确函数排序决策结果,提取相应的决策规则,设计算法.实例分析表明,文中方法使决策者在处理信息冲突的多属性决策问题时可根据实际需求选择最优决策方案     

TrendPerceptor: A property-function based technology intelligence system for identifying technology trends from patents

Technology intelligence systems are vital components for planning of technology development and formulation of technology strategies. Although such systems provide computation supports for technology analysis, much effort and intervention of experts, who may be expensive or unavailable, is required in gathering processes of information for analysis. As a remedy, this paper proposes TrendPerceptor, a system that uses a property-function based approach. The proposed system assists experts (1) to identify trends in invention concepts from patents, and (2) to perform evolution trend analysis of patents for technology forecasting. For this purpose, a module of the system uses grammatical analysis of textual information to automatically extract properties and functions, which show innovation directions in a given technology. Using the identified properties and functions, a module for invention concept analysis based on network analysis and a module for evolution trend analysis based on TRIZ (Russian acronym of the Theory of Inventive Problem Solving) trends are suggested. This paper describes the architecture of a system composed of these three modules, and illustrates two case studies using the system.     

Function evaluation with feedforward neural networks

Much effort has previously been spent in investigating the decision making/object identification capabilities of feedforward neural networks. In the present work we examine the less frequently investigated abilities of such networks to implement computationally useful operations in arithmetic and function evaluation. The approach taken is to employ standard training methods, such as backpropagation, to teach simple three-level networks to perform selected operations ranging from one-to-one mappings to many-to-many mappings. Examples considered cover a wide range, such as performing reciprocal arithmetic on real valued inputs, implementing particle identifier functions for identification of nuclear isotopes in scattering experiments, and locating the coordinates of a charged particle moving on a surface. All mappings are required to interpolate and extrapolate from a small sample of taught exemplars to the general continuous domain of possible inputs. A unifying principle is proposed that looks upon all such function constructions as expansions in terms of basis functions, each of which is associated with a hidden node and is parameterized by such techniques as gradient descent methods.     

设计的力量——从无印良品看设计的社会职能

随着商品经济竞争的复杂化,设计在社会中所扮演的不同角色和人们对设计作用的认识的不同决定了不同国家的设计风格及其发展方向。而在现今迷茫的中国,设计逐渐"高端化"的误解日益加重,设计究竟应该服务于大众借以引导和提高大众的生活品位和质量,还是作为一种奢侈商品。当我们把设计人格化,那么设计的职能是什么?这是当今中国"畸形"经济需要我们的思考。     

Estimating the Expected and the Optimal Probabilities of Misclassification

The problem is to estimate the expected and the optimal probabilities of misclassification in the context of the two group p-dimensional normal classification problem with means and common covariance matrix unknown and a rule based on the linear discriminant function. Performance of several estimators is compared by means of a computer sampling study. For larger p (p = 20) certain estimators are definitely superior for each of the probabilities, while for small p there is less differentiation in performance.     

Cloud-DPP for distributed process planning of mill-turn machining operations

Today, the dynamic market requires manufacturing firms to possess a high degree of adaptability to deal with shop-floor uncertainties. Specifically targeting SMEs active in the metal cutting sector who normally deal with intensive process planning problems, researchers have tried to address the subject. Among proposed solutions, Cloud-DPP elaborates a two-layer distributed adaptive process planning based on function-block technology and cloud concept. One of the challenges of companies is to machine as many part features as possible in a single setup on a single machine. Nowadays, multi-tasking machines are widely used due to their various advantages such as reducing setup times and increasing part accuracy. However, they also possess programming challenges because of their complex configuration and multiple machining functions. This paper reports the latest state of design and implementation of Cloud-DPP methodology to support parts with a combination of milling and turning features, and process planning for multi-tasking machining centers with special functionalities to minimize the number of setups. The contributions of this work are: representation of machining states and part transfer functionality, support of multi-tasking machines in adaptive setup merging, development of special function blocks to handle sub-setups and transitions, and finally generation of function-block network for the merged setups. The developed prototype is validated through a case study.     

Function projective synchronization between integer-order and stochastic fractional-order nonlinear systems

In this paper, the function projective synchronization between integer-order and stochastic fractional-order nonlinear systems is investigated. Firstly, according to the stability theory of fractional-order systems and tracking control, a controller is designed. At the same time, based on the orthogonal polynomial approximation, the method of transforming stochastic error system into an equivalent deterministic system is given. Thus, the stability of the stochastic error system can be analyzed through its equivalent deterministic one. Finally, to demonstrate the effectiveness of the proposed scheme, the function projective synchronization between integer-order Lorenz system and stochastic fractional-order Chen system is studied.     

A parallel function evaluation approach for solution to large-scale equation-oriented models

The equation-oriented (EO) approach is widely used for process simulation and optimization. Nevertheless, large-scale EO models consist of a huge number of nonlinear equations and make the solution procedure a challenging and time-consuming task. For most gradient-based numerical algorithms, function evaluations are the dominant step during the solution procedure. Here, a parallel computation method is developed for function evaluations within EO optimization strategies. After dividing the equations into several groups, function evaluations are calculated by using multiple threads on a parallel hardware platform simultaneously. Theoretical analysis for the speedup ratio is conducted. The implementation of the proposed method on a multi-core processor platform as well as a graphics processing unit (GPU) platform is then presented with several case studies. Numerical results are compared and discussed to show that the multi-core processor implementation has good computational performance, whereas the GPU implementation only achieves computational acceleration under relatively specific conditions.