New Interfaces and Approaches to Machine Learning When Classifying Gestures within Music

Interactive music uses wearable sensors (i.e., gestural interfaces—GIs) and biometric datasets to reinvent traditional human–computer interaction and enhance music composition. In recent years, machine learning (ML) has been important for the artform. This is because ML helps process complex biometric datasets from GIs when predicting musical actions (termed performance gestures). ML allows musicians to create novel interactions with digital media. Wekinator is a popular ML software amongst artists, allowing users to train models through demonstration. It is built on the Waikato Environment for Knowledge Analysis (WEKA) framework, which is used to build supervised predictive models. Previous research has used biometric data from GIs to train specific ML models. However, previous research does not inform optimum ML model choice, within music, or compare model performance. Wekinator offers several ML models. Thus, we used Wekinator and the Myo armband GI and study three performance gestures for piano practice to solve this problem. Using these, we trained all models in Wekinator and investigated their accuracy, how gesture representation affects model accuracy and if optimisation can arise. Results show that neural networks are the strongest continuous classifiers, mapping behaviour differs amongst continuous models, optimisation can occur and gesture representation disparately affects model mapping behaviour; impacting music practice.     

A study on the applications of the acoustic design sensitivity analysis of vibrating bodies

The determination of the sensitivity of the acoustical characteristics of vibrating systems with respect to the variation of the design parameters predicting these characteristics is a necessary and important step of the acoustic design and optimization process. Acoustic design sensitivity analysis includes the computation and evaluation of the sensitivity information required for this procedure. In this study, a boundary element code performing the sensitivity analysis of the acoustic pressure by using the matrix sensitivities with respect to different design variables has been developed. The effect of the precision of boundary element discretization on the acoustic pressure sensitivity is examined via this code. The formulation is applied to a multi-source system and the dimension sensitivity analysis of near field pressures of two-dilating-spherical source is performed. The last application is devoted to a real sound source: a washing machine sitting on the floor. Sensitivity of the field pressures to the machine’s dimensions (size), surface velocity and frequency is examined on the bases of the boundary element model of the machine and half-space condition. The impacts of these variables are compared; and a limiting speed for the machine responding both the acoustical and operational requirements is determined.     

基于SPH-FEM转换算法的7.62mm步枪弹冲击30CrMnSiA钢板的数值计算

从转换判据和时间步长控制两个方面对光滑粒子流体动力学-有限元(SPH-FEM)转换算法进行改进,采用改进的SPH-FEM转换算法对7.62 mm步枪弹冲击特殊热处理的30CrMnSiA钢板进行全尺寸三维数值计算,将子弹在冲击过程中发生严重畸变的有限单元转换为SPH粒子继续参与计算.子弹采用弹塑性模型,靶板采用修正的Johnson-Cook强度模型和Gruneisen状态方程.针对不同的子弹初速,计算靶板盘形凹陷和冲塞两种破坏模式,计算结果与实验结果吻合较好.表明改进的SPH-FEM转换算法能够合理高效地利用有限单元和SPH粒子,为低强度子弹冲击高强度靶板的计算提供一种有效途径.     

基于改进型相关向量机的高光谱图像分类

相关向量机(RVM)高光谱图像分类算法是一种基于贝叶斯概率模型的监督机器学习算法,其分类精度较高、测试时间较短。然而算法本身存在训练时间随着训练样本增加直线上升、分类效率整体降低等问题。针对这种情况,提出一种基于改进型相关向量机(VRVM)的高光谱图像分类算法。本算法在传统概率模型中引入一个新的分布,使得计算复杂度较高的积分运算可近似地拆分成两个较为简单的对数和形式。实验结果表明,VRVM高光谱图像分类算法的总体分类精度和相关向量的数量与RVM基本相同,但训练时间随样本数的增加有明显的减少。     

一种眼镜镜片缺陷自动化检测系统的研究

根据光散射原理和视觉检测的方法,研究开发了一种基于机器视觉的眼镜镜片缺陷自动化检测系统,对眼镜镜片进行图像获取、简单的图像处理与分级、分拣,实现对镜片的分类。系统采用并行机构运行,利用低角度前向照明方式获得了镜片表面和内部的斑、杂质、羽毛等缺陷图像;利用归一算法进行图像处理与分级,得到不同缺陷的识别信息。实验结果表明,系统能够检测出镜片中含有的所有缺陷,每个镜片的平均检测时间控制在2 s以内,检测精度为0.038 mm。     

Rebuilding of metal components with laser cladding forming

Laser cladding forming (LCF) is a novel powerful tool for the repairing of metal components. Rebuilding of V-grooves on medium carbon steel substrates has been carried out with laser cladding forming technique using stainless steel powder as the cladding material. Microstructure of the deposited layers has been characterized using optical microscopy (OM), scanning electron microscopy (SEM), energy dispersive X-ray microanalysis (EDAX), electron probe microanalysis (EPMA) and X-ray diffraction (XRD). Mechanical properties of the rebuilt V-groove samples have been evaluated by tensile and impacting tests and microhardness measurement. Experimental results show that good fusion bonding between the rebuilt layers and the substrate has been formed, and the microstructure of the cladding layers is mainly composed of fine, dense and defect-free epitaxial columnar dendrites. Due to the effect of grain size refinement, the tensile strength, impacting toughness, elongation and microhardness of the rebuilt samples have been greatly enhanced compared to those of the substrate. Microhardness is also very uniform throughout the rebuilt regions. With the growth of the deposited layers, the microhardness increases gradually. The good ductility of the deposited regions is verified by the SEM fracture analysis.     

Ein Beitrag zur radiometrischen Bestimmung des Ziehsteinverschleißes beim Ziehen von Stahldraht

Es wird ein zuvelässiges und sehr empfindliches radiometrisches Verfahren zur Bestimmung des Ziehsteinverschleiβes beim Ziehen von Stahldraht beschrieben. Im Vergleich zu den in der Literatur angegebenen radiometrischen Meβverfahren ist das behandelte Verfahren weniger zeitaufwending und gestattet es, bei einem vertretbaren Aufwand gröβere Verscuhsreihen durchzuführen. Mit Hilfe dieser Methode ist es möglich, langwierige groβtechnische Versuche zu umgehen und gleichzeitig eine wesentlich exaktere Aussage ūber den Verschleiβ unter verschiedenen Ziehbedingungen zu erhalten.     

Determining Optimum Conditions for Measuring by Gamma-ray Backscattering

The paper is concerned on the calculation of the optimal thickness of measured material from the point of view of measurement error. The minimum relative error of a measurement is determined as a function of the gamma radiaiton linear attenuation coefficient and the thickness of the material used. Various types of fluctuations are considered and attention is paid to the background detected in the same arrangement without measured material. The results are summarized into simple figures. The conclusion is made that the optimal thickness increases with poorer quality of the measuring instrument.     

Externally applied stress sign and film elastic properties effects on brittle film fracture

Rectangular stainless steel samples with TiN film deposited on the front lateral surface were loaded in three-point bending to the maximum normal strain of 6%. Scanning electron microscopy showed that vertical cracks appeared in the tension zone when the tensile strain exceeded 1.5%, while horizontal cracks appeared in the compression zone when the compressive strain exceeded –2.9%. Film cracks in the compressive zone originate from the tensile stress imposed by the plastically deformed substrate due to the Poisson’s expansion. Taking plastic deformation and Poisson’s expansion of the substrate in compression into account, theoretical analysis of normal stress distribution along the cracked film segment in compression is presented. Substrate strain and film elastic properties affect film cracking in the compressive zone. At larger compressive strain, some transverse cracks along with buckling cause the film spallation. The presented method is useful for studying brittle film fracture with variable strain levels in a single sample.     

基于AHP权重和灰色服务器负载预测的云计算On-line迁移策略

为了克服云计算环境下由于实时用户任务的不确定性到来和服务器性能差异而导致的云计算环境的负载不均衡问题,提出了一种AHP权重获取和灰度算法预测服务器负载的云计算on-line虚拟机迁移策略。首先,设计了基于AHP和灰色服务器预测的虚拟机on-line迁移模型,提出了采用AHP获取虚拟机各资源需求权重,然后,采用灰色模型预测下一时刻的服务器负载,采用此权值向量与各无需迁移的服务器的空闲资源向量进行加权得到加权和,将具有最小加权和的物理服务器作为迁移的目标宿主机。最后,定义了基于AHP权重和灰色服务器负载预测的云计算on-line迁移算法。在CloudSim环境下进行实验,结果表明文中的迁移策略使得云计算在响应用户任务时,具有任务失败次数少、SLA违约率低和迁移成功率高的优点,同时与其它方法相比,具有负载均衡程度高的优点,具有较强的可行性。