A recommender system(RS)relying on latent factor analysis usually adopts stochastic gradient descent(SGD)as its learning algorithm.However,owing to its serial mechanism,an SGD algorithm suffers from low efficiency and...A recommender system(RS)relying on latent factor analysis usually adopts stochastic gradient descent(SGD)as its learning algorithm.However,owing to its serial mechanism,an SGD algorithm suffers from low efficiency and scalability when handling large-scale industrial problems.Aiming at addressing this issue,this study proposes a momentum-incorporated parallel stochastic gradient descent(MPSGD)algorithm,whose main idea is two-fold:a)implementing parallelization via a novel datasplitting strategy,and b)accelerating convergence rate by integrating momentum effects into its training process.With it,an MPSGD-based latent factor(MLF)model is achieved,which is capable of performing efficient and high-quality recommendations.Experimental results on four high-dimensional and sparse matrices generated by industrial RS indicate that owing to an MPSGD algorithm,an MLF model outperforms the existing state-of-the-art ones in both computational efficiency and scalability.展开更多
A novel algorithm is presented in this paper to jointly estimate frequency, two-dimensional (2-D) direc- tion-of-arrival (DOA), and range of near-field narrowband sources. The proposed method extends the parallel fact...A novel algorithm is presented in this paper to jointly estimate frequency, two-dimensional (2-D) direc- tion-of-arrival (DOA), and range of near-field narrowband sources. The proposed method extends the parallel factor (PARAFAC) analysis model from the common data- and subspace-domain to the cumu- lant one, and forms three-way arrays by using the five cumulant matrices obtained from the array out- puts, and analyzes the uniqueness of low-rank decomposition of the three-way arrays, then jointly es- timates source parameters via the low-rank decomposition. In comparison with the conventional methods, the proposed method alleviates the loss of the array aperture, and avoids pairing parameters. What is more important, this algorithm can deal with mixed far-field and near-field sources. Finally, the simulation results validate the performance of the proposed method.展开更多
基金supported in part by the National Natural Science Foundation of China(61772493)the Deanship of Scientific Research(DSR)at King Abdulaziz University(RG-48-135-40)+1 种基金Guangdong Province Universities and College Pearl River Scholar Funded Scheme(2019)the Natural Science Foundation of Chongqing(cstc2019jcyjjqX0013)。
文摘A recommender system(RS)relying on latent factor analysis usually adopts stochastic gradient descent(SGD)as its learning algorithm.However,owing to its serial mechanism,an SGD algorithm suffers from low efficiency and scalability when handling large-scale industrial problems.Aiming at addressing this issue,this study proposes a momentum-incorporated parallel stochastic gradient descent(MPSGD)algorithm,whose main idea is two-fold:a)implementing parallelization via a novel datasplitting strategy,and b)accelerating convergence rate by integrating momentum effects into its training process.With it,an MPSGD-based latent factor(MLF)model is achieved,which is capable of performing efficient and high-quality recommendations.Experimental results on four high-dimensional and sparse matrices generated by industrial RS indicate that owing to an MPSGD algorithm,an MLF model outperforms the existing state-of-the-art ones in both computational efficiency and scalability.
文摘A novel algorithm is presented in this paper to jointly estimate frequency, two-dimensional (2-D) direc- tion-of-arrival (DOA), and range of near-field narrowband sources. The proposed method extends the parallel factor (PARAFAC) analysis model from the common data- and subspace-domain to the cumu- lant one, and forms three-way arrays by using the five cumulant matrices obtained from the array out- puts, and analyzes the uniqueness of low-rank decomposition of the three-way arrays, then jointly es- timates source parameters via the low-rank decomposition. In comparison with the conventional methods, the proposed method alleviates the loss of the array aperture, and avoids pairing parameters. What is more important, this algorithm can deal with mixed far-field and near-field sources. Finally, the simulation results validate the performance of the proposed method.
