This paper proposes a 1D residual convolutional neural network (CNN) for classifying arrhythmias based on electrocardiogram (ECG) signals. The additional residual blocks and skip connections effectively alleviate the ...
详细信息
作者：
Wang, GeFan, Feng-LeiDepartment of Biomedical Engineering
Department of Electrical Computer and Systems Engineering Department of Computer Science Center for Computational Innovations Biomedical Imaging Center Center for Biotechnology and Interdisciplinary Studies Rensselaer Polytechnic Institute TroyNY United States Department of Data Science
City University of Hong Kong Kowloon Hong Kong
The recent awarding of the Nobel Prize in Physics to Geoffrey E. Hinton and John J. Hopfield highlights their profound impact on artificial neural networks. In this perspective, we explore how their foundational insig...
详细信息
This paper presents a 12b hybrid ADC consisting of a 8b successive approximation register (SAR) analog-to-digital converter (ADC) with top-plate sampling, a variable-slope voltage-to-time converter with built-in absol...
This paper presents a 12b hybrid ADC consisting of a 8b successive approximation register (SAR) analog-to-digital converter (ADC) with top-plate sampling, a variable-slope voltage-to-time converter with built-in absolute-value function, and a 3-stage cyclic gated Vernier TDC. The proposed ADC features the ability to accommodate a large input without sacrificing linearity, low power consumption, and technology compatibility. A number of techniques are proposed to improve performance and power efficiency of the ADC. These techniques include a sub-threshold delay-locked loop with pulsed control signals for clock generation, clock kickback reduction to minimize kickback and mismatch induced sensitivity loss of the comparator of the SAR ADC, a gated cyclic Vernier TDC to minimize the resolution loss caused by the improper injection of time inputs and a virtually unlimited dynamic range, a self-resetting arbiter with zero metastability window to improve the resolution of Vernier TDC, and a significantly simplified synchronous SAR for a better power/area efficiency. The ADC is designed in a TSMC 130 nm 1.2 V CMOS technology with a reduced supply voltage of 0.8 V and analyzed using Spectre with BSIM3.3 device models. Simulation results show that at 500 kS/s, the ADC offers a SNDR of 69.04 dB and ENOB of 11.17 while consuming 3.70 µW, yielding a Walden FOM of 3.21 fJ/conv.
Mixture-regularized bidirectional gated recurrent unit with attention (BiGAR) boosts the efficiency of decoding brain signals into hand movement trajectories. The novel neural decoder achieves an R-squared of over 0.8...
详细信息
ISBN:
(数字)9798350359312
ISBN:
(纸本)9798350359329
Mixture-regularized bidirectional gated recurrent unit with attention (BiGAR) boosts the efficiency of decoding brain signals into hand movement trajectories. The novel neural decoder achieves an R-squared of over 0.8 in less than 0.2 ms of computation time on the MC_Maze dataset using fewer than 500 training trials. The expectation maximization (EM) algorithm used to extract neural hidden states improves R-squared and retains relatively low computation time for BiGAR. We further research on how different mixture regularizers impact the model performance. We generate mixture regularizers through pairwise weighted sum mixing of five individual regularizers associated with the Gaussian, Cauchy, Laplace, Sinc-squared, and Sin-fourth probability density functions. Experiments indicate that the improvement of model R-squared with mixture regularizers exceeds that of traditional individual regularizers and no regularizer.
This special issue of the IEEE Transactions On Circuits and Systems— PART II: EXPRESS BRIEFS (TCAS-II) continues the successful tradition of the co-publication initiative started few years ago by the IEEE Circuits an...
详细信息
This special issue of the IEEE Transactions On Circuits and Systems— PART II: EXPRESS BRIEFS (TCAS-II) continues the successful tradition of the co-publication initiative started few years ago by the IEEE Circuits and Systems Society (CASS) to publish a selection of the best papers accepted for presentation at the IEEE International Symposium on Circuits and Systems (ISCAS). This year ISCAS is held in Monterey, California, United States of America, on May 21st – May 25th, and the process for this Special Issue was carried out as soon as the paper selection was done. As TCAS-II only publishes 5-page briefs, and hence both conference and journal versions of selected works would be largely overlapped, the papers included in this Issue will not appear in the Proceedings of the IEEE ISCAS. Also, similar to what is done by other IEEE Societies, IEEE CASS intends to shift the role of IEEE conferences towards more networking events as well as opportunities for discussions on ongoing research efforts.
This paper introduces a real-time collaborative sensing scheme for wireless sensor networks in time-varying environments. The objective is to maximize the sensors' performance by effectively allocating communicati...
详细信息
ISBN:
(数字)9798331517786
ISBN:
(纸本)9798331517793
This paper introduces a real-time collaborative sensing scheme for wireless sensor networks in time-varying environments. The objective is to maximize the sensors' performance by effectively allocating communication resources for data sharing. Specifically, we utilize digital twins (DTs) to characterize dynamic collaborative sensing demands for each sensor through data-driven methods. Building on the DT design, we propose a resource allocation scheme to optimize the communication resources allocated at each stage of collaborative sensing and determine the most effective collaborative sensing policy. By profiling sensors using DTs, the network controller can effectively coordinate the sensors without exhaustively exploring all collaborative sensing policies. Numerical results demonstrate the effectiveness of our proposed scheme in optimizing the sensing performance for all sensors.
Monocular 3D human pose estimation involves predicting the 3D pixel coordinates of key body joints from a 2D image or video. Typically, a 2D estimation model is employed to initially determine joint locations in an im...
详细信息
ISBN:
(数字)9798350394948
ISBN:
(纸本)9798350394955
Monocular 3D human pose estimation involves predicting the 3D pixel coordinates of key body joints from a 2D image or video. Typically, a 2D estimation model is employed to initially determine joint locations in an image, followed by training a separate model to lift these positions to 3D coordinates. In this paper, we evaluate the performance of recently proposed 2D human pose estimation models as different inputs for training and evaluation of 2D-3D lifting models. In addition, we propose four simple merging strategies to combine the outputs of these 2D human pose estimators and generate less noisy 2D inputs. To evaluate, four recent 2D pose estimators—ViTPose, PCT, MogaNet, and TransPose—are selected, and their corresponding 2D outputs are generated on the Human3.6M dataset. Subsequently, MotionAGFormer and PoseFormerV2 are trained and evaluated using each created 2D input and its corresponding 3D motion-capture ground truth. ViTPose stands out as the top-performing 2D estimator, and employing all merging strategies proves beneficial in generating a less noisy 2D input. Code and data are available at https://***/TaatiTeam/2DEstimatorEval.
This paper concentrates on pattern learning of Granger causality. In this context, the entities of the Granger causality matrix estimation derived from the state-space model indicate directional dependencies of the ob...
This paper concentrates on pattern learning of Granger causality. In this context, the entities of the Granger causality matrix estimation derived from the state-space model indicate directional dependencies of the observations. Existing methods propose different forms of thresholding to eliminate insignificant entities from this matrix. These approaches do not exploit the difference in statistical characteristics of insignificant entities compared to the Granger causality itself. In this work, Noise Invalidation Soft Thresholding is chosen as the thresholding method to discard null relations in the Granger causality matrix pattern learning. Unlike existing approaches, the proposed method benefits from the statistical properties of insignificant entities. The simulation results demonstrate the advantages and superiority of the proposed method in the sense of accuracy and robustness for both randomly generated datasets with causal footprints as well as simulated electroencephalogram datasets.
The advance of sensory and computing technology has attracted wide attention in the study of intelligent information fusion for multimedia computing and analysis. As a result, information fusion has been taking center...
The advance of sensory and computing technology has attracted wide attention in the study of intelligent information fusion for multimedia computing and analysis. As a result, information fusion has been taking center stage in the intelligent multimedia and machine learning communities. In this paper, a deep discriminant fractional-order canonical correlation analysis (DDFCCA) method is proposed with application to information fusion. Benefiting from the integration of deep cascade neural networks (NNs) with discriminant power of the fractionalorder correlation matrix across multiple data/information sources, the proposed DDFCCA method demonstrates the ability to generate high quality data/information representation. To verify the effectiveness and generic nature of the proposed method, we conduct experiments on three database (MNIST database, RML audio emotional database, and Caltech101 database). Experimental results validate the superiority of the DDFCCA method over stateof-the-art for information fusion.
暂无评论