The origin of artificial intelligence is investigated, based on which the concepts of hybrid intelligence and parallel intelligence are presented. The paradigm shift in Intelligence indicates the ''new normal&...
详细信息
With the development of hyperspectral remote sensing information processing, hyperspectral image classification becomes a hot topic. The algorithm of kernel sparse representation classification based on spatial-spectr...
详细信息
ISBN:
(纸本)9781509036783
With the development of hyperspectral remote sensing information processing, hyperspectral image classification becomes a hot topic. The algorithm of kernel sparse representation classification based on spatial-spectral graph regularization and sparsity concentration index (SSGSCI-KSRC) gains a good result. Due to the big scale of hyperspectral image data, time-critical requirement in the practical application makes it impossible to use the original SSGSCI-KSRC algorithm. This paper proposes a parallelization method for SSGSCI-KSRC algorithm. The optimization method achieves the efficient calculation operations of hyperspectral image matrix data, coalesces memory accesses to reduce the time of transferring data to the GPU devices, and designs proper kernel functions in the classification algorithm. The experimental results demonstrate that the parallel SSGSCI-KSRC algorithm obtains a better result in terms of computational performance when the accuracy stays the same.
Gliomas are the most common primary brain malignancies, with different degrees of aggressiveness, variable prognosis and various heterogeneous histologic sub-regions, i.e., peritumoral edematous/invaded tissue, necrot...
详细信息
Convolutional Neural Networks (CNNs) have delivered impressive state-of-the-art performances for many vision tasks, while the computation costs of these networks during test-time are notorious. Empirical results have ...
详细信息
ISBN:
(纸本)9781467399623
Convolutional Neural Networks (CNNs) have delivered impressive state-of-the-art performances for many vision tasks, while the computation costs of these networks during test-time are notorious. Empirical results have discovered that CNNs have learned the redundant representations both within and across different layers. When CNNs are applied for binary classification, we investigate a method to exploit this redundancy across layers, and construct a cascade of classifiers which explicitly balances classification accuracy and hierarchical feature extraction costs. Our method cost-sensitively selects feature points across several layers from trained networks and embeds non-expensive yet discriminative features into a cascade. Experiments on binary classification demonstrate that our framework leads to drastic test-time improvements, e.g., possible 47.2× speedup for TRECVID upper body detection, 2.82× speedup for Pascal VOC2007 People detection, 3.72× for INRIA Person detection with less than 0.5% drop in accuracies of the original networks.
Transformative technologies are enabling the construction of three dimensional (3D) maps of tissues with unprecedented spatial and molecular resolution. Over the next seven years, the NIH Common Fund Human Biomolecula...
详细信息
Accurate pitch extraction from speech is important but challenging problem for speech synthesis. However, the additive nature and long-term suprasegmental property of pitch features have not been fully exploited in mo...
详细信息
Accurate pitch extraction from speech is important but challenging problem for speech synthesis. However, the additive nature and long-term suprasegmental property of pitch features have not been fully exploited in most of the existing pitch estimators as they are operated frame by frame. As a result, they would cause some inherent discontinuities, such as double/half F0 errors and unvoiced/voiced(U/V) error. This would adversely affect the quality of synthetic speech as well as the expressiveness of the prosody information. In this paper, we explore the novel use of multi-tasks(Task 1: U/V;Task 2: Pitch) bidirectional long short-term memory recurrent neural network(BLSTM) to model the pitch and voicing decision simultaneously in a unified framework. The features used in this study are extracted from the frequency domain. We compute the log-frequency power spectrogram and then normalize to the long-term speech spectrum to attenuate noises. A filter is then used to enhance the harmonicity. Experiments show that the proposed approach substantially outperforms RAPT, which behaves the best in clean condition. Besides, our proposed approach can even work well with a certain level of background noise.
Recent years have witnessed the rapid development of intelligent transportation systems (ITS), which urgently call for highly reliable and effective communication technologies for vehicle-to-vehicle and vehicle-to-inf...
详细信息
ISBN:
(纸本)9781467385800
Recent years have witnessed the rapid development of intelligent transportation systems (ITS), which urgently call for highly reliable and effective communication technologies for vehicle-to-vehicle and vehicle-to-infrastructure (V2X) applications. Orthogonal frequency division multiplexing (OFDM) is widely believed to be the most promising candidate among these applications. However, the orthogonality of subcarriers in OFDM can be easily destroyed by the Doppler-induced inter-carrier interference (ICI), which is very common in V2X channels. In this paper, we propose a novel scheme which integrates the ICI self-cancellation technique into the index modulated (IM-) OFDM framework. Via careful designs, the proposed scheme can not only inherit the advantages of IM-OFDM but also suppress the ICI from active subcarriers. Simulations validate that in V2X channels, the proposed scheme significantly outperforms existing IM-OFDM and more importantly shows better performance than the conventional OFDM with ICI self-cancellation without sacrificing the spectral efficiency.
Orthogonal frequency division multiplexing (OFDM) is prone to frequency offset which gives rise to intercarrier interference (ICI). Recently, a novel OFDM transmission scheme called OFDM with index modulation (IM-OFDM...
详细信息
ISBN:
(纸本)9781509016990
Orthogonal frequency division multiplexing (OFDM) is prone to frequency offset which gives rise to intercarrier interference (ICI). Recently, a novel OFDM transmission scheme called OFDM with index modulation (IM-OFDM) is proposed, which outperforms conventional OFDM in the absence of frequency offset. As in IM-OFDM, partial subcarriers are set to be idle by spatial modulation, the potential of IM-OFDM in ICI reduction is expected. In this paper, we find that this potential vanishes in some real scenarios with ICI, which leads to IM-OFDM exhibiting even worse performance than conventional OFDM. To improve this situation, a novel ICI cancellation scheme is designed, which integrates the ICI self-cancellation technique into the IM-OFDM framework. Via careful designs, the proposed scheme not only provides a solution to ICI problems of IM-OFDM, but also achieves an attractive tradeoff between the spectral efficiency and ICI cancellation performance of the system. Simulations validate that in the presence of carrier frequency offset (CFO), the proposed scheme significantly outperforms existing IM-OFDM and more importantly shows better performance than the conventional OFDM with ICI self-cancellation.
Active Disturbance Rejection control (ADRC) as a standalone motion solution has been adopted by companies such as Texas Instruments and Danfoss and made available on various proprietary industrial platforms. The idea ...
详细信息
ISBN:
(纸本)9781479983988
Active Disturbance Rejection control (ADRC) as a standalone motion solution has been adopted by companies such as Texas Instruments and Danfoss and made available on various proprietary industrial platforms. The idea of ADRC, however, can be integrated with the existing control technologies seamlessly, as shown in this paper. It is shown a modularized ADRC design for set-point tracking of motion control such that better uncertainty rejection can be implemented without any change in the existing proportional-derivative (PD) control with linear observer. We prove that certain integration of the observer's error can serve as an estimation for the “total disturbance” in low frequency range. This enables the estimation and cancellation of the “total disturbance” to be incorporated into the existing control loop. Also, a comparison between the methods with and without such “module” is discussed. The proposed ADRC is implemented and validated with experimental results for a 1-degree of freedom robotic manipulator, where desired set-point tracking performance in position control is achieved under unknown mass variations and sudden external disturbances.
暂无评论