In this paper, we present a demo named AniDance that can synthesize dance motions with melody in real-time. When users sing a song or play one in their phone to AniDance, their melody will drive the 3D-space character...
详细信息
ISBN:
(纸本)9781450356657
In this paper, we present a demo named AniDance that can synthesize dance motions with melody in real-time. When users sing a song or play one in their phone to AniDance, their melody will drive the 3D-space character to dance to create a lively dance animation. In practice, we conduct a music oriented 3D-space dance motion dataset by capturing real dance performances, using LSTM-autoencoder to identify the relation between music and dance. Based on these technologies, users can create valid choreographies that capable of musical expression, witch can promote their learning ability and interest in dance and music.
WeChat is one of social network applications that connects people widely. Huge data is generated when users conduct conversations, which can be used to enhance their lives. This paper will describe how this data is co...
详细信息
ISBN:
(纸本)9781538674475
WeChat is one of social network applications that connects people widely. Huge data is generated when users conduct conversations, which can be used to enhance their lives. This paper will describe how this data is collected, how to develop a personalized chatbot using personal conversation records. Our system will have a cognitive map based on the word2vec model, which is used to learn and store the relationship of each word that appears in the chatting records. Each word will be mapped to a continuous high dimensional vector space. Then we will adopt the sequence-to-sequence framework (seq2seq) to learn the chatting styles from all pairs of chatting sentences. Meanwhile, we will replace the traditional one-hot embedding layer with our word2vec embedding layer in the seq2seq model. Furthermore, we trained an autoencoder of seq2seq architecture to learn the vector representation of each sentence, then we can evaluate the cosine similarity between model generated response and the pre-existing response in test set, and we can also display the distance with principal component analysis (PCA) projection. As a result, our word2vec embedded seq2seq model significantly outperforms the one-hot embedded one.
The representation learning of life cycle dataset has its particularity in the correlation of different features and the dependency of adjacent sampling time. This paper addresses the difficulty of segmentation to hig...
详细信息
ISBN:
(纸本)9789811057687;9789811057670
The representation learning of life cycle dataset has its particularity in the correlation of different features and the dependency of adjacent sampling time. This paper addresses the difficulty of segmentation to high-dimensional nonlinear life cycle long CBM data, and propose a new deep learning approach based on unsupervised representation learning named autoencoder for rolling bearing diagnosis. Two kinds of autoencoder with encoder and decoder model are developed respectively using fully connected and convolutional hidden layers to automatically extract the dataset's representative features. Compared to the fully connected one, the convolutional autoencoder shows clearer in a lower dimensional feature space by preserving the local neighborhood structure, and more effective to discover subjectively the intrinsic structure of nonlinear high-dimensional data of deterioration process.
This paper introduces techniques to automatically detect driving corner cases from dashcam video and inertial sensors. Developing robust driver assistance and automated driving technologies requires an understanding o...
详细信息
ISBN:
(纸本)9781450359528
This paper introduces techniques to automatically detect driving corner cases from dashcam video and inertial sensors. Developing robust driver assistance and automated driving technologies requires an understanding of not just common highway and city traffic situations but also a plethora of corner cases that may be encountered in billions of miles of driving. Current approaches seek to collect such a catalog of corner cases by driving millions of miles with self-driving prototypes. In contrast, this paper introduces a low-cost yet scalable solution to collect such events from any dashcam-equipped vehicle to take advantage of the billions of miles that humans already drive. It detects unusual events through inertial sensing of sudden human driver reactions and rare visual events through a trained autoencoder deep neural network. We evaluate the system based on more than 120 hours real road driving data. It shows 82% accuracy improvement versus strawman solutions for sudden reaction detection and above 71% accuracy for rare visual views identification. The detection results proved useful for re-training and improving a self-steering algorithm on more complex situations. In terms of computational efficiency, the Android prototype achieves 17Hz frame rate (Nexus 5X).
We introduce SCORES, a recursive neural network for shape composition. Our network takes as input sets of parts from two or more source 3D shapes and a rough initial placement of the parts. It outputs an optimized par...
详细信息
ISBN:
(纸本)9781450360081
We introduce SCORES, a recursive neural network for shape composition. Our network takes as input sets of parts from two or more source 3D shapes and a rough initial placement of the parts. It outputs an optimized part structure for the composed shape, leading to high-quality geometry construction. A unique feature of our composition network is that it is not merely learning how to connect parts. Our goal is to produce a coherent and plausible 3D shape, despite large incompatibilities among the input parts. The network may significantly alter the geometry and structure of the input parts and synthesize a novel shape structure based on the inputs, while adding or removing parts to minimize a structure plausibility loss. We design SCORES as a recursive autoencoder network. During encoding, the input parts are recursively grouped to generate a root code. During synthesis, the root code is decoded, recursively, to produce a new, coherent part assembly. Assembled shape structures may be novel, with little global resemblance to training exemplars, yet have plausible substructures. SCORES therefore learns a hierarchical substructure shape prior based on per-node losses. It is trained on structured shapes from ShapeNet, and is applied iteratively to reduce the plausibility loss. We show results of shape composition from multiple sources over different categories of man-made shapes and compare with state-of-the-art alternatives, demonstrating that our network can significantly expand the range of composable shapes for assembly-based modeling.
Incomplete data has emerged as a prominent problem in the fields of machine learning, big data and various other academic studies. Due to the surge in deep learning techniques for problem-solving, in this paper, autho...
详细信息
ISBN:
(纸本)9783319938189;9783319938172
Incomplete data has emerged as a prominent problem in the fields of machine learning, big data and various other academic studies. Due to the surge in deep learning techniques for problem-solving, in this paper, authors have proposed a deep learning-metaheuristic approach to combat the problem of imputing missing data. The proposed approach (DL-GSA) makes use of the nature inspired metaheuristic, Gravitational search algorithm, in combination with a deep-autoencoder and performs better than existing methods in terms of both accuracy and time. Owing to these improvements, DL-GSA has wider applications in both time and accuracy sensitive areas like imputation of scientific and research datasets, data analysis, machine learning and big data.
The piecewise-linear activation functions such as ReLU become the catalyst that revolutionizes the training of the deep neural networks. Common nonlinear activation functions used in neural networks such as the tanh a...
详细信息
ISBN:
(纸本)9781538643990
The piecewise-linear activation functions such as ReLU become the catalyst that revolutionizes the training of the deep neural networks. Common nonlinear activation functions used in neural networks such as the tanh and the sigmoid activation functions suffer from saturation during training. The saturation behavior causes the problem of vanishing stochastic gradient decent. We propose a fast activation function, namely the Adaptive Linear Function (ALF) to increase the convergence speed and accuracy of the deep leaning structure for real-time applications. The ALF reduces the saturation effects caused by the soft activation functions and the vanishing gradient caused by the negative values of the ReLU. We evaluate the training method for an online anomaly intrusion detection system using Deep Belief Network (DBN) and simulating four benchmark datasets. The activation function increases the convergence speed of the DBN, with the entire training time reduced 80% compared to the sigmoid, ReLU, and tanh activation functions. The method achieves an accuracy rate of 98.59% on the total 10% KDDCUP'99 test dataset, 96.2% on the NSL-KDD dataset, 98.4% on the Kyoto dataset, and 96.57% on the CSIC HTTP dataset. The proposed activation function outperformed the results obtained when any of the three activation functions-sigmoid, ReLu, or tanh-was used on the test stream of the four datasets. Furthermore, the DBN structure outperforms state-of-the-art networks such as the Stacked Sparse autoencoder Based Extreme Learning Machine (SSAELM) in both accuracy and convergence speed.
The exploration of oil and gas is a vital part of today's increasing power demands to meet the energy we need to power our homes, businesses, and transportation. Oil and gas explorers use seismic surveys, both ons...
详细信息
The exploration of oil and gas is a vital part of today's increasing power demands to meet the energy we need to power our homes, businesses, and transportation. Oil and gas explorers use seismic surveys, both onshore and offshore, to produce detailed images of the various rock types, layers, and their locations beneath the Earth's subsurface. The acquired data undergo a series of processing steps, which require powerful computing hardware, sophisticated software, and specialized manpower. To extract useful information from seismic data, interpreters manually delineate important geological structures, which contain hints about petroleum and gas reservoirs such as salt domes, faults, channels, fractures, and horizons. These structures typically span over several square kilometers and are delineated based on correlation, changes in illumination, intensity, contrast, and texture of seismic data. There are limited tools available for automatic detection and manual interpretation is becoming extremely time consuming and labor intensive. In this dissertation, we propose novel seismic attributes based on texture dissimilarity, visual-attention theory, the modeling of human visual system, and machine learning to quantify changes and highlight geological features in a three-dimensional space. To automate the process of seismic interpretation, we develop interpreter-assisted, fully-, and semi-automated workflows that are interactive and easy-to-use for the delineation of important geological structures within seismic volumes. Experimental results on real and synthetic datasets show that our proposed algorithms outperform the state-of-the-art methods for seismic interpretation. In a nutshell, this dissertation introduces novel seismic attributes and automated, interactive, and interpreter-assisted workflows, which have a very promising future in effective seismic interpretation. The proposed research is computationally inexpensive and is expected to not only reduce the time fo
汽车电子控制系统(Automotive electronic control system)在汽车领域的重要性日益增强,它让汽车变得更加安全、舒适和节能。现阶段汽车电子控制系统正变得越来越复杂,而汽车电子控制系统中以动力总成模块PCM最为复杂和核心,为了解决PC...
详细信息
汽车电子控制系统(Automotive electronic control system)在汽车领域的重要性日益增强,它让汽车变得更加安全、舒适和节能。现阶段汽车电子控制系统正变得越来越复杂,而汽车电子控制系统中以动力总成模块PCM最为复杂和核心,为了解决PCM控制功能复杂化与快速开发之间的矛盾,V模式开发流程开始被各大汽车公司应用到汽车电子控制系统中。HIL是V模式开发流程中非常重要的环节,通过HIL仿真测试系统就可以在实验室对PCM进行测试,极大的缩短了开发的时间周期、降低了开发的成本。对驾驶行为进行定性或定量的建模与评估是无人驾驶、智能交通等领域的重要课题和关键技术,在车辆性能测试、辅助驾驶和异常驾驶行为的诊断中都有十分重要的应用价值。从汽车制造厂商的角度来看,了解汽车驾驶员的驾驶行为偏好对优化汽车的驾驶行为体验和个性化的提升车辆的动力学性能都有十分重要的意义。本文通过建立驾驶行为的标准化模型对驾驶员的驾驶行为偏好进行学习,将不同路段、不同环境的驾驶行为映射到联邦车辆测试曲线中,实现对不同驾驶行为的标准化。本文在充分了解了汽车测试的HIL系统发展现状的基础上,结合自身对PCM的开发需求,设计制作了基于NIPXI平台的硬件系统和上位机管理软件NI Verstand的HIL仿真测试系统。然后用autoencoder算法对不同驾驶风格的驾驶员采集到的驾驶数据进行驾驶行为标准化,最后用经过标准化后的驾驶数据对HIL系统进行整机测试,证明了该系统的有效性和准确性。
暂无评论