Search | Engineering

订阅投稿

首页工程期刊工程焦点工程成就工程前沿关于我们 English

资源类型

期刊论文 80

会议信息 1

年份

2023 5

2022 7

2021 7

2020 3

2019 9

2018 3

2017 8

2016 1

2015 6

2014 1

2013 2

2011 1

2010 3

2009 2

2008 2

2007 2

2006 3

2005 3

2004 5

2002 2

展开︾

关键词

模式识别 8

目标识别 4

人脸识别 2

效果评估 2

特征抽取 2

语音识别 2

AR模型 1

CAD 1

CAM 1

CAPP 1

CCD影像 1

Laplacian特征映射 1

RBF神经网络 1

三维视觉知识；三维参数模型；心脏病理诊断；数据增强 1

中心聚类 1

主-客体化学 1

人口控制 1

低维流形 1

信号消噪；深度自适应阈值学习网络；多尺度特征融合；调制识别 1

展开︾

检索范围：

排序：展示方式：

Speech emotion recognitionwith unsupervised feature learning

Zheng-wei HUANG,Wen-tao XUE,Qi-rong MAO

《信息与电子工程前沿（英文）》 2015年第16卷第5期页码 358-366 doi: 10.1631/FITEE.1400323

摘要： Emotion-based features are critical for achieving high performance in a speech emotion recognition (SER) system. In general, it is difficult to develop these features due to the ambiguity of the ground-truth. In this paper, we apply several unsupervised feature learning algorithms (including -means clustering, the sparse auto-encoder, and sparse restricted Boltzmann machines), which have promise for learning task-related features by using unlabeled data, to speech emotion recognition. We then evaluate the performance of the proposed approach and present a detailed analysis of the effect of two important factors in the model setup, the content window size and the number of hidden layer nodes. Experimental results show that larger content windows and more hidden nodes contribute to higher performance. We also show that the two-layer network cannot explicitly improve performance compared to a single-layer network.

关键词： Speech emotion recognition Unsupervised feature learning Neural network Affect computing

HTML PDF 收藏

UsingKinect for real-time emotion recognition via facial expressions

Qi-rong MAO,Xin-yu PAN,Yong-zhao ZHAN,Xiang-jun SHEN

《信息与电子工程前沿（英文）》 2015年第16卷第4期页码 272-282 doi: 10.1631/FITEE.1400209

摘要： Emotion recognition via facial expressions (ERFE) has attracted a great deal of interest with recent advances in artificial intelligence and pattern recognition. Most studies are based on 2D images, and their performance is usually computationally expensive. In this paper, we propose a real-time emotion recognition approach based on both 2D and 3D facial expression features captured by Kinect sensors. To capture the deformation of the 3D mesh during facial expression, we combine the features of animation units (AUs) and feature point positions (FPPs) tracked by Kinect. A fusion algorithm based on improved emotional profiles (IEPs) and maximum confidence is proposed to recognize emotions with these real-time facial expression features. Experiments on both an emotion dataset and a real-time video show the superior performance of our method.

关键词： Kinect Emotion recognition Facial expression Real-time classification Fusion algorithm Support vector machine (SVM)

HTML PDF 收藏

多带同步模型用于噪声环境下语音识别

孙暐,吴镇扬

《中国工程科学》 2006年第8卷第3期页码 31-34

摘要：

根据人耳听觉特性，提出新的同步多带最大似然线性回归算法用于噪声环境下语音识别。该算法采用最大似然作为参数估计准则，利用各频带信号同步感知和噪声污染假定的方法进行语音模型补偿，有效地提高了识别系统在噪声环境下的识别性能。

关键词：隐马尔可夫模型最大似然多带同步模型语音识别

HTML PDF 收藏

基于语音识别的电磁调控智能超表面 Article

柏林, 刘元可, 徐亮, 张政, 王强, 蒋卫祥, 仇成伟, 崔铁军

《工程（英文）》 2023年第22卷第3期页码 185-190 doi: 10.1016/j.eng.2022.06.026

摘要：

本研究提出并实现了一种基于人类语音识别的智能超表面平台，用于对电磁波束进行可编程调控。该智能超表面平台由数字编码超表面、语音识别模块、单片机和数模转换器（DAC）电路组成，可根据预先存储的语音指令对电磁波进行智能控制。所构建的数字编码超表面包含6 × 6 个超级子单元，每个超级子单元由4 × 4 个嵌入了变容二极管的有源数字单元组成。语音识别模块配合DAC和单片机对语音指令进行识别，并生成对应的电压序列来控制超表面。此外，在超表面的设计过程中引入遗传算法，可有效优化超表面相位分布。为了验证智能超表面平台的性能，实验展示了雷达散射截面积缩减、涡旋波束生成和波束分裂三种典型功能。所提出的方案为调控电磁波提供了一种新途径，并在电磁和声学通信之间架起了一座桥梁。

关键词：语音识别可编程超表面遗传算法智能电磁调控

HTML PDF 收藏

用于语音识别的二值神经网络 Regular Papers

Yan-min QIAN, Xu XIANG

《信息与电子工程前沿（英文）》 2019年第20卷第5期页码 701-715 doi: 10.1631/FITEE.1800469

摘要：近年来，在语音识别的声学建模中，深度神经网络(DNNs)明显优于高斯混合模型。然而，推断阶段巨大的计算量使其难以部署在低功耗的嵌入式模型上。为此，稀疏性和低精度定点量化技术被广泛使用。为降低推理阶段计算量，本文开发了用于语音识别的二进制神经网络，并实现了高速的二值矩阵乘法。在中央处理器(CPU)和图形处理单元(GPU)上，二值矩阵乘法的运行速度是浮点矩阵乘法的5–7倍。针对大规模连续语音识别的声学建模，提出多种二值神经网络及相关模型优化算法。为提高二值模型的精度，探索了从浮点模型到二值模型的知识蒸馏技术。在标准的Switchboard语音识别任务上，该二值神经网络模型比浮点神经网络模型速度提高3–4倍。借助知识蒸馏技术，二值深度神经网络或卷积神经网络相对其浮点神经网络的词错误率增加可以保持在15%以内。若只二值化卷积神经网络的卷积层，词错误率增加几乎可忽略。

关键词：语音识别；二值神经网络；二值矩阵乘法；知识蒸馏；位1计数

HTML PDF 收藏

基于心理生理信号的个人音乐情感体验识别 Research Articles

Le-kai ZHANG, Shou-qian SUN, Bai-xi XING, Rui-ming LUO, Ke-jun ZHANG

《信息与电子工程前沿（英文）》 2019年第20卷第7期页码 964-974 doi: 10.1631/FITEE.1800101

摘要：重要节点识别对电网安全意义重大。但电网音乐能激发人的情感，这是一个心理生理过程。因此心理生理特征可用于识别个人音乐情感体验。提出一种新的基于人体生理特征的个人音乐情感识别方法。首先，建立一个基于音乐情感特征的数据库和一个基于听音乐产生的生理信号的数据库，包括心电、脉搏、皮温、呼吸和瞳孔直径变化等生理信号。然后，分别采用线性回归、岭回归、三种不同核的支持向量机、决策树、K近邻算法、多层感知器和Nu支持向量回归（NuSVR）方法，通过音乐特征和人体生理特征识别音乐情感。结果显示，NuSVR性能优于其他方法，其唤醒相关系数为0.7347（均方差为0.02323），效价相关系数为0.7902（均方差为0.01485）。最后，对不同数据集进行比较，结果表明所有特征（音乐特征和所有生理特征）数据集在识别中表现最好，唤醒相关系数为0.6499（均方差为0.02932），效价相关系数为0.7735（均方差为0.01576）。本文提供了一种有效的个人音乐情感体验识别方法，可用于个性化音乐推荐。

关键词：音乐；情感识别；生理信号；小波变换

HTML PDF 收藏

一种面向机器情感智能的人格引导型情感脑机接口 Research Article

李少杰,李伟,邢泽健,袁文杰,韦香玉,张晓炜,胡斌

《信息与电子工程前沿（英文）》 2022年第23卷第8期页码 1158-1173 doi: 10.1631/FITEE.2100489

摘要：情感脑机接口（brain–computer interfaces, BCIs）已成为在人机协作中实现情感智能的一个重要途径。然而，由于脑电图（electroencephalogram, EEG）信号的复杂性和情绪反应的个体差异性，设计一个可靠和有效的模型仍然是一个巨大挑战。考虑到不同人格特征的个体在情绪感知和反应过程中的差异，整合人格信息和脑电信号对情绪识别是有帮助的。鉴于此，提出一种人格引导的注意力神经网络，其可以利用人格信息学习更为有效的EEG表征以用于情感识别。具体来说，我们首先利用卷积神经网络提取脑电信号的时域和空域表征，进而设计一种特殊的卷积核同时学习大脑头皮不同区域间和区域内的EEG导联相关关系。其次，考虑到不同大脑头皮区域在情绪识别中可能发挥不同的作用，提出一种人格引导的区域注意力机制，以进一步探索区域内和区域间EEG导联的贡献。最后，设计一种基于注意力的长短期记忆网络（long short-term memory, LSTM）建模EEG信号的时域动态特征。在AMIGOS数据集（一个用于个人和群体的情感、人格特征和情绪多模态研究的数据集）的实验结果表明，本研究所提方法可以显著提升被试独立策略下情感识别的性能，并优于现有情感识别方法。

关键词：脑电图；情感识别；注意力机制；人格特征

HTML PDF 收藏

在国际工程科技大会午宴上的致辞

宋健

《中国工程科学》 2000年第2卷第10期页码 3-4

HTML PDF 收藏

一种基于子带激励替换的窄带语音水印算法 Article

Wei LIU, Ai-qun HU

《信息与电子工程前沿（英文）》 2017年第18卷第5期页码 627-643 doi: 10.1631/FITEE.1601503

摘要：本文建议了一种新的窄带语音水印算法，该算法将部分语音信号替换为一个经过幅度调节和频谱成形的隐蔽信号。文中从理论上证明了若仅对一小部分载体语音作修改，则不但可以建立一个用于隐蔽通信的理想信道模型，而且还能确保隐藏信号的不可感知性和带水印信号的可懂度。在此基础上，文中建议了一个实际的系统模型。在嵌入器中，使用了能量规范化准则，使得通带水印信号的能量与原始通带语音激励信号的能量相等，然后使用合成滤波器来对该水印信号进行频谱成形。在提取器中，先用带通滤波器去除带外信号，然后用分析滤波器对失真进行补偿。实验结果显示，本文建议算法的数据速率可达400 bits/s，具有更佳的带宽效率，且有良好的不可感知性。此外，该算法对实际应用中的各种攻击也是顽健的。

关键词：分析滤波器；线性预测；窄带语音水印；通带激励替代；能量规范化；谱包络成形；合成滤波器

HTML PDF 收藏

在中国工程院第六次院士大会闭幕式上的讲话

徐匡迪

《中国工程科学》 2002年第4卷第7期页码 6-7

HTML PDF 收藏

Review of sentiment analysis: An emotional product development view

《工程管理前沿（英文）》页码 592-609 doi: 10.1007/s42524-022-0227-z

摘要： Conveying consumers’ specific emotions in new products, referred to as emotional product development or emotional design, is strategically crucial for manufacturers. Given that sentiment analysis (SA) can extract and analyze people’s opinions, sentiments, attitudes, and perceptions regarding different products/services, SA-based emotional design may provide manufacturers with real-time, direct, and rapid decision support. Despite its considerable advancements and numerous survey and review articles, SA is seldom considered in emotional design. This study is among the first efforts to conduct a thorough review of SA from the view of emotional design. The comprehensive review of aspect-level SA reveals the following: 1) All studies focus on extracting product features by mixing technical product features and consumers’ emotional perceptions. Consequently, such studies cannot capture the relationships between technical and emotional attributes and thus cannot convey specific emotions to the new products. 2) Most studies use the English language in SA, but other languages have recently received more interest in SA. Furthermore, after conceptualizing emotion as Kansei and introducing emotional product development and Kansei Engineering, a review of the data-driven emotional design is then conducted. A few efforts start to study emotional design with the help of SA. However, these studies only focus on either analyzing consumers’ preferences on product features or extracting emotional opinions from online reviews, thus cannot realize data-driven emotional product development. Finally, some research opportunities are provided. This study opens a broad door to aspect-level SA and its integration with emotional product development.

关键词： sentiment analysis emotion product development Kansei Engineering

HTML PDF 收藏

Multiobjective image recognition algorithm in the fully automatic die bonder

JIANG Kai, CHEN Hai-xia, YUAN Sen-miao

《机械工程前沿（英文）》 2006年第1卷第3期页码 313-316 doi: 10.1007/s11465-006-0026-y

摘要： It is a very important task to automatically fix the number of die in the image recognition system of a fully automatic die bonder. A multiobjective image recognition algorithm based on clustering Genetic Algorithm (GA), is proposed in this paper. In the evolutionary process of GA, a clustering method is provided that utilizes information from the template and the fitness landscape of the current population. The whole population is grouped into different niches by the clustering method. Experimental results demonstrated that the number of target images could be determined by the algorithm automatically, and multiple targets could be recognized at a time. As a result, time consumed by one image recognition is shortened, the performance of the image recognition system is improved, and the atomization of the system is fulfilled.

关键词： clustering different recognition algorithm Algorithm multiobjective

HTML PDF 收藏

Advances in tissue state recognition in spinal surgery: a review

Hao Qu, Yu Zhao

《医学前沿（英文）》 2021年第15卷第4期页码 575-584 doi: 10.1007/s11684-020-0816-3

摘要： Spinal disease is an important cause of cervical discomfort, low back pain, radiating pain in the limbs, and neurogenic intermittent claudication, and its incidence is increasing annually. From the etiological viewpoint, these symptoms are directly caused by the compression of the spinal cord, nerve roots, and blood vessels and are most effectively treated with surgery. Spinal surgeries are primarily performed using two different techniques: spinal canal decompression and internal fixation. In the past, tactile sensation was the primary method used by surgeons to understand the state of the tissue within the operating area. However, this method has several disadvantages because of its subjectivity. Therefore, it has become the focus of spinal surgery research so as to strengthen the objectivity of tissue state recognition, improve the accuracy of safe area location, and avoid surgical injury to tissues. Aside from traditional imaging methods, surgical sensing techniques based on force, bioelectrical impedance, and other methods have been gradually developed and tested in the clinical setting. This article reviews the progress of different tissue state recognition methods in spinal surgery and summarizes their advantages and disadvantages.

关键词： spinal surgery tissue state recognition image force sensing bioelectrical impedance

HTML PDF 收藏

View-invariant human action recognition via robust locally adaptive multi-view learning

Jia-geng FENG,Jun XIAO

《信息与电子工程前沿（英文）》 2015年第16卷第11期页码 917-920 doi: 10.1631/FITEE.1500080

摘要： Human action recognition is currently one of the most active research areas in computer vision. It has been widely used in many applications, such as intelligent surveillance, perceptual interface, and content-based video retrieval. However, some extrinsic factors are barriers for the development of action recognition; e.g., human actions may be observed from arbitrary camera viewpoints in realistic scene. Thus, view-invariant analysis becomes important for action recognition algorithms, and a number of researchers have paid much attention to this issue. In this paper, we present a multi-view learning approach to recognize human actions from different views. As most existing multi-view learning algorithms often suffer from the problem of lacking data adaptiveness in the nearest neighborhood graph construction procedure, a robust locally adaptive multi-view learning algorithm based on learning multiple local L1-graphs is proposed. Moreover, an efficient iterative optimization method is proposed to solve the proposed objective function. Experiments on three public view-invariant action recognition datasets, i.e., ViHASi, IXMAS, and WVU, demonstrate data adaptiveness, effectiveness, and efficiency of our algorithm. More importantly, when the feature dimension is correctly selected (i.e.,>60), the proposed algorithm stably outperforms state-of-the-art counterparts and obtains about 6% improvement in recognition accuracy on the three datasets.

关键词： View-invariant Action recognition Multi-view learning L1-norm Local learning