Filters








2,272 Hits in 1.8 sec

Towards Flexible Sparsity-Aware Modeling: Automatic Tensor Rank Learning Using The Generalized Hyperbolic Prior [article]

Lei Cheng, Zhongtao Chen, Qingjiang Shi, Yik-Chung Wu, Sergios Theodoridis
2022 arXiv   pre-print
Tensor rank learning for canonical polyadic decomposition (CPD) has long been deemed as an essential yet challenging problem.  ...  advanced generalized hyperbolic (GH) prior to the probabilistic CPD model, which not only includes the Gaussian-gamma model as a special case, but also is more flexible to adapt to different levels of sparsity  ...  Among all the tensor decompositions, canonical polyadic decomposition (CPD) is the most fundamental format.  ... 
arXiv:2009.02472v2 fatcat:f5xhbv7warhi7fofltmhgwkaw4

Quantized Sparse Weight Decomposition for Neural Network Compression [article]

Andrey Kuzmin, Mart van Baalen, Markus Nagel, Arash Behboodi
2022 arXiv   pre-print
We use projected gradient descent methods to find quantized and sparse factorization of the weight tensors.  ...  In our method, we store weight tensors as sparse, quantized matrix factors, whose product is computed on the fly during inference to generate the target model's weights.  ...  Related work SVD-based methods and tensor decompositions SVD decomposition was first used to demonstrate redundacy in weight parameters in neural networks in (Denil et al., 2013) .  ... 
arXiv:2207.11048v1 fatcat:n3v4kjogufaevg43etmgtrcqmm

Fast Time-Aware Sparse Trajectories Prediction with Tensor Factorization

Lei ZHANG, Qingfu FAN, Guoxing ZHANG, Zhizheng LIANG
2018 IEICE transactions on information and systems  
Existing trajectory prediction methods suffer from the "data sparsity" and neglect "time awareness", which leads to low accuracy.  ...  Due to the scale of the tensor, we design a divide and conquer tensor factorization model to reduce memory consumption and speed up decomposition.  ...  We design a divide and conquer tensor factorization model to reduce memory consumption and speed up decomposition.  ... 
doi:10.1587/transinf.2018edl8017 fatcat:4jqfgjshz5ezndobw3f6yrev3q

Tensor Decomposition for Spatial—Temporal Traffic Flow Prediction with Sparse Data

Funing Yang, Guoliang Liu, Liping Huang, Cheng Siong Chin
2020 Sensors  
Based on the constructed traffic flow tensor, we either propose a model to estimate the correlation in each dimension of the tensor.  ...  The main challenge of traffic flow prediction is the data sparsity problem, meaning that traffic flow on some roads or of certain periods cannot be monitored.  ...  Among the effective methods to solve data sparsity are generative models, such as the tensor decomposition method [28] .  ... 
doi:10.3390/s20216046 pmid:33114275 fatcat:rprapznq5vcg3dpgnucvfmyds4

Multi-Context-Aware Location Recommendation using Tensor Decomposition

Jing Lu, Martin A. Indeche
2020 IEEE Access  
In this paper, we propose a Multi-Context-aware Location Recommendation using Tensor Decomposition (MCLR-TD) approach that incorporates multiple context information at different granularity scales in modeling  ...  INDEX TERMS Context information, contextual preference, location recommendation, tensor decomposition.  ...  CONTEXT-AWARE COLLABORATIVE TENSOR DECOMPOSITION The ultimate goal of tensor decomposition is to supplement missing values of a tensor [39] .  ... 
doi:10.1109/access.2020.2983555 fatcat:ywxf3hjwgngmjc25tnbsl5p4ie

Matrix Factorization Techniques for Context-Aware Collaborative Filtering Recommender Systems: A Survey

Mohamed Hussein Abdi, George Onyango Okeyo, Ronald Waweru Mwangi
2018 Computer and Information Science  
The incorporation of context information and Matrix and Tensor Factorization techniques have proved to be a promising solution to some of these challenges.  ...  This approach encounters data sparsity and scalability limitations as the volume of accessible information and the active users continue to grow leading to performance degradation, poor quality recommendations  ...  The extension of standard Matrix Factorization has incorporated context information such as Time-Aware Matrix Factorization (Liu, Cao, Zhao, & Yang, 2010) , Context-aware Matrix Factorization (Baltrunas  ... 
doi:10.5539/cis.v11n2p1 fatcat:vyyrbt7exba2bhufdwoyrad3fa

Personalized Web Service Recommendation based on QoS Prediction and Hierarchical Tensor Decomposition

Tian Cheng, Junhao Wen, Qingyu Xiong, Jun Zeng, Wei Zhou, Xueyuan Cai
2019 IEEE Access  
INDEX TERMS QoS, service computing, tensor decomposition.  ...  The hierarchical tensor decomposition is then performed on the local and global triadic tensors.  ...  GLOBAL TENSOR DECOMPOSITION The tensor decomposition model has been extensively used in context-aware web service recommendation systems.  ... 
doi:10.1109/access.2019.2909548 fatcat:tsajo2gmprbktnh77tspn6vaim

Towards Compact CNNs via Collaborative Compression [article]

Yuchao Li, Shaohui Lin, Jianzhuang Liu, Qixiang Ye, Mengdi Wang, Fei Chao, Fan Yang, Jincheng Ma, Qi Tian, Rongrong Ji
2021 arXiv   pre-print
In this paper, we propose a Collaborative Compression (CC) scheme, which joints channel pruning and tensor decomposition to compress CNN models by simultaneously learning the model sparsity and low-rankness  ...  Channel pruning and tensor decomposition have received extensive attention in convolutional neural network compression.  ...  To leverage the benefits of both compression operations, training-aware methods [43, 29, 20] use two regularizations to separately handle the sparsity on channel pruning and the low-rankness on tensor  ... 
arXiv:2105.11228v1 fatcat:ndcnrfzekrfk7gknhntf5y35z4

Towards Flexible Sparsity-Aware Modeling: Automatic Tensor Rank Learning Using The Generalized Hyperbolic Prior

Lei Cheng, Zhongtao Chen, Qingjiang Shi, Yik-chung Wu, Sergios Theodoridis
2022 IEEE Transactions on Signal Processing  
Tensor rank learning for canonical polyadic decomposition (CPD) has long been deemed as an essential yet challenging problem.  ...  advanced generalized hyperbolic (GH) prior to the probabilistic CPD model, which not only includes the Gaussian-gamma model as a special case, but also is more flexible to adapt to different levels of sparsity  ...  Among all the tensor decompositions, canonical polyadic decomposition (CPD) is the most fundamental format.  ... 
doi:10.1109/tsp.2022.3164200 fatcat:ipe7z33hungsjjqdz6eclbfmcu

Compression-aware Training of Neural Networks using Frank-Wolfe [article]

Max Zimmer and Christoph Spiegel and Sebastian Pokutta
2022 arXiv   pre-print
We also extend these ideas to the structured pruning domain and propose novel approaches to both ensure robustness to the pruning of convolutional filters as well as to low-rank tensor decompositions of  ...  Further, we propose to enforce robustness to low-rank tensor decomposition by interpreting convolutional tensors as matrices and constraining their spectral-k-support norm, resulting in low-rank updates  ...  Low-Rank Matrix decomposition To evaluate the capabilities of SFW for tensor decomposition as proposed in Section 2.2.3, we compare the Spectral-k-support norm approach to two natural baselines: nuclear  ... 
arXiv:2205.11921v1 fatcat:3widp6r2lba6ri2j3smd3ohvxi

Context-aware tensor decomposition for relation prediction in social networks

Achim Rettinger, Hendrik Wermser, Yi Huang, Volker Tresp
2012 Social Network Analysis and Mining  
While the first approach, the Context-Aware Recommendation Tensor Decomposition (CARTD), proposes an efficient optimization criterion and decomposition  ...  Unfortunately, the straightforward application of higher-order tensor models becomes problematic, due to the sparsity of the data and due to the complexity of the computations.  ...  While the first approach, the Context-Aware Recommender Tensor Decomposition (CARTD), proposes an efficient ranking and optimization criterion, the second approach, the Context-aware Regularized Singular  ... 
doi:10.1007/s13278-012-0069-5 fatcat:ysogtuva65eovhcvvjbuescgsq

Sturm: Sparse Tubal-Regularized Multilinear Regression for fMRI [article]

Wenwen Li, Jian Lou, Shuo Zhou, Haiping Lu
2018 arXiv   pre-print
The newly proposed tensor singular value decomposition (t-SVD) sheds light on new directions.  ...  Recent sparse multilinear regression methods based on tensor are emerging as promising solutions for fMRI, yet existing works rely on unfolding/folding operations and a tensor rank relaxation with limited  ...  Related Work Tensor decomposition and rank.  ... 
arXiv:1812.01496v1 fatcat:llyqce27jrdgba3j7dkk6nofgm

Diagnosing New York city's noises with ubiquitous data

Yu Zheng, Tong Liu, Yilun Wang, Yanmin Zhu, Yanchi Liu, Eric Chang
2014 Proceedings of the 2014 ACM International Joint Conference on Pervasive and Ubiquitous Computing - UbiComp '14 Adjunct  
Supplementing the missing entries of the tensor through a context-aware tensor decomposition approach, we recover the noise situation throughout NYC.  ...  We model the noise situation of NYC with a three dimension tensor, where the three dimensions stand for regions, noise categories, and time slots, respectively.  ...  Algorithm 1: Context-Aware Tensor Decomposition Input: tensor , matrix , matrix , matrix , an error threshold Output: , , , 1.  ... 
doi:10.1145/2632048.2632102 dblp:conf/huc/ZhengLWZLC14 fatcat:7tsfdgschnghzdy6tcsolvi76m

Expert Recommendation via Tensor Factorization with Regularizing Hierarchical Topical Relationships [article]

Chaoran Huang, Lina Yao, Xianzhi Wang, Boualem Benatallah, Shuai Zhang, Manqing Dong
2018 arXiv   pre-print
Tensor decomposition enables to leverage the latent expertise of users, and the posts and related tags help identify the related areas.  ...  We employ the recent techniques of tree-guided learning (via tensor decomposition), and matrix factorization to explore user expertise from past voted posts.  ...  [14] offer a context-aware tensor decomposition model to integrate context information with collaborative filtering tightly. Hidas et al.  ... 
arXiv:1808.01092v2 fatcat:swxrw5vrkjezrhjkcp6lnrfi5i

Expert Recommendation via Tensor Factorization with Regularizing Hierarchical Topical Relationships [chapter]

Chaoran Huang, Lina Yao, Xianzhi Wang, Boualem Benatallah, Shuai Zhang, Manqing Dong
2018 Lecture Notes in Computer Science  
Tensor decomposition enables to leverage the latent expertise of users, and the posts and related tags help identify the related areas.  ...  We employ the recent techniques of tree-guided learning (via tensor decomposition), and matrix factorization to explore user expertise from past voted posts.  ...  [14] offer a context-aware tensor decomposition model to integrate context information with collaborative filtering tightly. Hidas et al.  ... 
doi:10.1007/978-3-030-03596-9_27 fatcat:2zpzlbj56jezbpfbvpvqyivc3m
« Previous Showing results 1 — 15 out of 2,272 results