【摘 要】
：
Nonnegative tensor decomposition has become increasingly important for multiway data analysis in recent years.The alternating proximal gradient (APG) is a popular optimization method for nonnegative tensor decomposition in the block coordinate descent fra
【机 构】
：
School of Biomedical Engineering,Faculty of Electronic Information and Electrical Engineering,Dalian
论文部分内容阅读
Nonnegative tensor decomposition has become increasingly important for multiway data analysis in recent years.The alternating proximal gradient (APG) is a popular optimization method for nonnegative tensor decomposition in the block coordinate descent framework.In this study,we propose an inexact version of the APG algorithm for nonnegative CANDECOMP/PARAFAC decomposition,wherein each factor matrix is updated by only finite inner iterations.We also propose a parameter warm-start method that can avoid the frequent parameter resetting of conventional APG methods and improve convergence performance.By experimental tests,we find that when the number of inner iterations is limited to around 10 to 20,the convergence speed is accelerated significantly without losing its low relative error.We evaluate our method on both synthetic and real-world tensors.The results demonstrate that the proposed inexact APG algorithm exhibits outstanding performance on both convergence speed and computational precision compared with existing popular algorithms.
其他文献
Dear editor,rnComponents such as gyros,turbines,and bearings are vital in aircraft and spacecraft,and must run safely during their entire life cycle.However,gradual damage inevitably occurs resulting in component failure.Therefore,necessary main-tenance s
Plenoptic imaging(PI)systems can flexibly record both spatial and angular information on flame radiation,enabling volumetric reconstruction of complex flames.The accuracy and efficiency of the reconstruction are significantly affected by the orientation p
High-order tensor data are prevalent in real-world applications,and multiway clustering is one of the most important techniques for exploratory data mining and compression of multiway data.However,existing multiway clustering is based on the K-means proce
The tool path smoothness of the industrial robot directly influences the dynamic performance of its motion control.Therefore,it is important to smooth the path that is composed of small linear segments,in order to achieve continuous and smooth movement.Th
Calcium carbonate is promising thermochemical heat storage material for next-generation solar power systems due to its high energy storage density,low cost,and high operation temperature.Researchers have tried to improve energy storage performances of cal
Dear editor,rnIn recent years,novel MOSFET structures such as fully-depleted silicon-on-insulator(FDSOI)MOSFET and Fin-FET are adopted by the industry for their excellent elec-trostatic behavior and higher on-state current per footprint.Nevertheless,these
Dear editor,rnCompressed sensing (CS)[1],as an efficient data acquisi-tion paradigm,has attracted much attentions since it came up.The fundamental principle of CS is that a signal,which is sparse under some sparsity basis,can be efficiently ac-quired and
A novel Al0.5V0.1FeCrMnNi0.9 high-entropy alloy was designed to achieve enhanced strength and plasticity.The as-cast Al0.5V0.1FeCrMnNi0.9 alloy shows a typical dendritic microstructure consisting of a body-centered cubic-structured matrix phase and a B2-s
Dear editor,rnTo meet the requirement of rapid growth in the number of mobile terminals and demand for network service qual-ity from users,wireless networks must implement faster and more intelligent network operation and maintenance func-tions.One feasib
Dear editor,rnNon-orthogonal multiple access(NOMA),regarded as a promising technique for substantially increasing the spec-tral efficiency of wireless networks that comprise massive devices,has exhibited a number of significant performance advantages,incl