Word-embedding acts as one of the backbones of modern natural language processing(NLP).Recently,with the need for deploying NLP models to low-resource devices,there has been a surge of interest to compress word embedd...
详细信息
Word-embedding acts as one of the backbones of modern natural language processing(NLP).Recently,with the need for deploying NLP models to low-resource devices,there has been a surge of interest to compress word embeddings into hash codes or binary vectors so as to save the storage and memory ***,existing work learns to encode an embedding into a compressed representation from which the original embedding can be *** these methods aim to preserve most information of every individual word,they often fail to retain the relation between words,thus can yield large loss on certain *** this end,this paper presents Relation Reconstructive Binarization(R2B)to transform word embeddings into binary codes that can preserve the relation between *** its heart,R2B trains an auto-encoder to generate binary codes that allow reconstructing the wordby-word relations in the original embedding *** showed that our method achieved significant improvements over previous methods on a number of tasks along with a space-saving of up to 98.4%.Specifically,our method reached even better results on word similarity evaluation than the uncompressed pre-trained embeddings,and was significantly better than previous compression methods that do not consider word relations.
A new dual-band low-profile quadrifilar-helical antenna (QHA) was presented. The spiral radiation arm of QHA is printed to three circular dielectric plates to reduce the antenna height. In this case, the radiation arm...
详细信息
A control circuit for active metasurface based on varactor diode is designed in this paper. By varying the reverse bias voltage on the varactor diode, the designed circuit can regulate the additional phase of the acti...
详细信息
The paper proposes a noise reduction algorithm based on symplectic Geometric Mode Decomposition (SGMD) and Savitzky-Golay (SG)filtering to address the issue of noise interference during signal transmission. Firstly, t...
详细信息
ISBN:
(数字)9798350384437
ISBN:
(纸本)9798350384444
The paper proposes a noise reduction algorithm based on symplectic Geometric Mode Decomposition (SGMD) and Savitzky-Golay (SG)filtering to address the issue of noise interference during signal transmission. Firstly, the noisy signal is decomposed using symplectic geometric mode decomposition to obtain several symplectic geometric components. The component with less noise is then selected and superimposed based on its energy entropy. Finally, the reconstructed signal undergoes denoising through SG filtering. The proposed method is applied to conduct comparison experiments on noise reduction for linear frequency modulation signals and multicomponent periodic signals. The results demonstrate a significant noise reduction effect of the proposed method, which holds great significance for signal analysis and research on noise reduction.
The paper presents a novel circular polarization(CP) antenna loading with a parasitic ring metal strip, which is designed for global positioning system (GPS) L1 band applications. The antenna consists of a defected gr...
详细信息
A novel millimeter wave sensing antenna array is proposed for application in miniature detection and sensing devices suitable for operating in multiple scenarios, e.g., collecting in the face of video and infrared, an...
详细信息
Reversible data hiding in encrypted domain(RDH-ED) fortifies data security and privacy safeguards while upholding the original data's integrity and *** research on RDH-ED focuses on 2D images,while research on 3D ...
详细信息
Reversible data hiding in encrypted domain(RDH-ED) fortifies data security and privacy safeguards while upholding the original data's integrity and *** research on RDH-ED focuses on 2D images,while research on 3D mesh models is still *** paper introduces an RDH-ED method using block modulus encryption and multi-MSB ***,the original mesh model is partitioned into non-overlapping subblocks of equal size,and then the vertices in each subblock are encrypted with the same key for additive modulus encryption,ensuring that the spatial correlation present in the original mesh subblocks remains preserved within the encrypted ***,the subblocks are disrupted one by one using the 3D Arnold Transform to enhance *** vertices in each embeddable subblock are divided into a reference vertex and several embeddable vertices,where the multi-MSB prediction strategy is employed to allocate embedding room for each embeddable ***,the secret information is embedded into the allocated *** the proposed method almost completely preserves the spatial correlation within each subblock,the achieved embedding rate surpasses that of all previous outstanding methods that rely on vacating room after encryption(VRAE).The experimental findings demonstrate that the proposed approach achieves an average embedding rate of 45.55 bits per vertex(bpv),surpassing the state-of-the-art method that achieves 25.63 bpv.
The currently constructed millimeter wave imaging system has the problems of long sampling time and more sampling points of antenna units, and the use of compressed perception algorithm can improve the imaging quality...
详细信息
The Affine-Projection Maximum Asymmetric Correntropy Criterion (APMACC) is constructed, drawing upon the fundamental principles of the maximum asymmetric correntropy criterion and an affine-projection scheme. The APMA...
详细信息
作者:
Tian, ChanglinZhi, HuiTie, KuiAnhui University
School of Electronic and Information Engineering Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education Hefei China CO.
Ltd Bengbu China
In this paper, a RIS-assisted multiuser MIMO communication method based on deep reinforcement learning (RMMC-DRL) is proposed for multiuser scenarios. Our objective is to find the optimal transmit beamforming matrix o...
详细信息
暂无评论