Filters








12,200 Hits in 2.4 sec

Residual Matrix Product State for Machine Learning [article]

Ye-Ming Meng, Jing Zhang, Peng Zhang, Chao Gao, Shi-Ju Ran
2021 arXiv   pre-print
Tensor network, which originates from quantum physics, is emerging as an efficient tool for classical and quantum machine learning.  ...  Nevertheless, there still exists a considerable accuracy gap between tensor network and the sophisticated neural network models for classical machine learning.  ...  Hilbert space.  ... 
arXiv:2012.11841v2 fatcat:fi3fntnsove6zedber2mn4xejy

Molding CNNs for text: non-linear, non-consecutive convolutions [article]

Tao Lei, Regina Barzilay, Tommi Jaakkola
2015 arXiv   pre-print
For instance, we obtain 51.2% accuracy on the fine-grained sentiment classification task.  ...  Instead of concatenating word representations, we appeal to tensor algebra and use low-rank n-gram tensors to directly exploit interactions between words already at the convolution stage.  ...  Acknowledgments We thank Kai Sheng Tai, Mohit Iyyer and Jordan Boyd-Graber for answering questions about their paper. We also thank Yoon Kim, the MIT NLP group and the reviewers for their comments.  ... 
arXiv:1508.04112v2 fatcat:t472virw2jceblpdv5zr7wwa6y

Molding CNNs for text: non-linear, non-consecutive convolutions

Tao Lei, Regina Barzilay, Tommi Jaakkola
2015 Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing  
For instance, we obtain 51.2% accuracy on the fine-grained sentiment classification task. 1  ...  Instead of concatenating word representations, we appeal to tensor algebra and use low-rank n-gram tensors to directly exploit interactions between words already at the convolution stage.  ...  Acknowledgments We thank Kai Sheng Tai, Mohit Iyyer and Jordan Boyd-Graber for answering questions about their paper. We also thank Yoon Kim, the MIT NLP group and the reviewers for their comments.  ... 
doi:10.18653/v1/d15-1180 dblp:conf/emnlp/LeiBJ15 fatcat:hdjayrzlzzcl3gv4wit6kbt36a

Shot classification for action movies based on motion characteristics

Shuhui Wang, Shuqiang Jiang, Qingming Huang, Wen Gao
2008 2008 15th IEEE International Conference on Image Processing  
Considering that motion characteristic is very important for semantic movie analysis, and it contains abundant information in action movies, the structure tensor analysis is used for feature extraction  ...  In this paper, we propose a shot classification method for action movies.  ...  Meanwhile, we will be dedicated to improving the shot classification algorithm by better modeling of feature space and semantic space.  ... 
doi:10.1109/icip.2008.4712303 dblp:conf/icip/WangJHG08 fatcat:qsdmefutynb2ddrxzrghd7mnwq

Guest Editorial: Spatio-temporal Feature Learning for Unconstrained Video Analysis

Yahong Han, Liqiang Nie, Fei Wu
2018 Multimedia tools and applications  
With the development of mobile Internet and personal devices, we are witnessing an explosive growth of video data on the Web.  ...  How to develop a robust feature or representation is the key problem in unconstrained video analysis.  ...  Though deep learning methods have dominated the area of video feature learning, tensor learning is valuable to be explored, as tensor is a natural structure to represent the space-time structural information  ... 
doi:10.1007/s11042-018-6341-6 fatcat:wsp2hi2gyfgkra6yr2pvo2wbgy

Features of blast-induced vibration source and identification of geostructures

H. Ding, R. Labbas, Z.M. Zheng
2005 Journal of Sound and Vibration  
Based on this finding, two promising methods are proposed for the identification of geo-structures and parameters. r  ...  An analysis of the behavior of spherical elastic wave radiation from an impulsively loaded cavity embedded in an infinite space has been carried out.  ...  blasting vibration source could be modeled as a distributed moment tensor enclosed in a certain region around the explosive, and the time variation of the moment tensor depends mainly on its impulse.  ... 
doi:10.1016/j.jsv.2004.12.017 fatcat:djy3ngexuje4phwb3s7d4iyhhy

Disease classification: A probabilistic approach

Yogesh Rathi, J. Malcolm, S. Bouix, R. McCarley, L. Seidman, J. Goldstein, C-F Westin, M. E. Shenton
2010 2010 IEEE International Symposium on Biomedical Imaging: From Nano to Macro  
Three orthogonal measures that capture different aspects of the local tissue are derived from the tensor representation to form a feature vector.  ...  From these feature vectors, we form a probabilistic representation of each patient. This representation is affine invariant, thus obviating the need for registration of the images.  ...  The proposed feature set f lives in a 3-dimensional space. Computing the joint pdf in such a high-dimensional space is computationally intensive.  ... 
doi:10.1109/isbi.2010.5490246 dblp:conf/isbi/RathiMBMSGWS10 fatcat:zk5riccinzb4fovcuffzdr4qu4

Hierarchical Low-Rank Tensors for Multilingual Transfer Parsing

Yuan Zhang, Regina Barzilay
2015 Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing  
Accurate multilingual transfer parsing typically relies on careful feature engineering. In this paper, we propose a hierarchical tensor-based approach for this task.  ...  However, unlike traditional tensor models, it enables us to incorporate prior knowledge about desired feature interactions, eliminating invalid feature combinations.  ...  We thank the MIT NLP group and the EMNLP reviewers for their comments.  ... 
doi:10.18653/v1/d15-1213 dblp:conf/emnlp/ZhangB15 fatcat:ostoupo74ngmpnr6a62h3m7mx4

Page 361 of Astronomy and Astrophysics Vol. 268, Issue 1 [page]

1993 Astronomy and Astrophysics  
However, different from Marti et al. (1990), we use an up-to-date tensor form for the artificial viscosity. Section 2 shortly addresses the essential features of both codes.  ...  The discretization is first order in space and time, and becomes second order in both variables for equidistant grids.  ... 

Violent Scene Detection Using a Super Descriptor Tensor Decomposition

Muhammad Rizwan Khokher, Abdesselam Bouzerdoum, Son Lam Phung
2015 2015 International Conference on Digital Image Computing: Techniques and Applications (DICTA)  
To obtain a compact set of features for classification, the TUCKER-3 decomposition is applied to the super descriptor tensors, followed by feature selection using Fisher feature ranking.  ...  To obtain a compact set of features for classification, the TUCKER-3 decomposition is applied to the super descriptor tensors, followed by feature selection using Fisher feature ranking.  ...  feature core tensor of the data tensor X (i) .  ... 
doi:10.1109/dicta.2015.7371320 dblp:conf/dicta/KhokherBP15 fatcat:qyolt2pamjh5tlkva6asyzr3ge

Multi-aspect-streaming tensor analysis

Hadi Fanaee-T, João Gama
2015 Knowledge-Based Systems  
Abstract Tensor analysis is a powerful tool for multiway problems in data mining, signal processing, pattern recognition and many other areas.  ...  The new approach, which is developed for analysis-only purposes, instead of relying on expensive linear algebra techniques is founded on the histogram approximation concept.  ...  We extend the application of histograms to tensor analysis problem We propose the first approach for multi-aspect-streaming tensor analysis (MASTA) MASTA is space-efficient, fast and constant-time for  ... 
doi:10.1016/j.knosys.2015.07.013 fatcat:2564unfdi5el5hd2hopdo2z3je

Particle production during inflation and gravitational waves detectable by ground-based interferometers

Jessica L. Cook, Lorenzo Sorbo
2012 Physical Review D  
tensor spectrum, production of vectors can.  ...  We also discuss the prospects of detection by a space interferometer like LISA.  ...  , Eugene Lim and especially Laura Cadonati for useful discussions.  ... 
doi:10.1103/physrevd.85.023534 fatcat:m4nrydr6r5b4jnn5h63ky3om2e

XML Documents Clustering Using Tensor Space Model -- A Preliminary Study

Sangetha Kutty, Richi Nayak, Yuefeng Li
2010 2010 IEEE International Conference on Data Mining Workshops  
Hence in this paper, we introduce a novel method of representing the XML documents in Tensor Space Model (TSM) and then utilize it for clustering.  ...  The traditional Vector Space Model (VSM) is not sufficient to represent both the structure and the content of such web documents.  ...  In this paper we propose a novel method that represents the XML documents in a Tensor Space Model (TSM) and uses the TSM for clustering.  ... 
doi:10.1109/icdmw.2010.106 dblp:conf/icdm/KuttyNL10 fatcat:xtmoxebrrfcfrb6d7ajuqmdc3u

High-Order Low-Rank Tensors for Semantic Role Labeling

Tao Lei, Yuan Zhang, Lluís Màrquez, Alessandro Moschitti, Regina Barzilay
2015 Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies  
The motivation behind the approach is to automatically induce a compact feature representation for words and their relations, tailoring them to the task.  ...  The tensor parameters are optimized for the SRL performance using standard online algorithms. Our tensor-based approach rivals the best performing system on the CoNLL-2009 shared task.  ...  We are grateful to Anders Bjökelund and Michael Roth for providing the outputs of their systems. We thank Yu Xin, Tommi Jaakkola, the MIT NLP group and the ACL reviewers for their comments.  ... 
doi:10.3115/v1/n15-1121 dblp:conf/naacl/LeiZVMB15 fatcat:ymq2edw6gvc5lhcz7o6s23txqa

Embedding Lexical Features via Low-Rank Tensors

Mo Yu, Mark Dredze, Raman Arora, Matthew R. Gormley
2016 Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies  
We apply lowrank tensor approximations to the corresponding parameter tensors to reduce the parameter space and improve prediction speed.  ...  We present a new model that represents complex lexical features-comprised of parts for words, contextual information and labels-in a tensor that captures conjunction information among these parts.  ...  As a result, they still suffer from large parameter spaces when the feature space is very huge. 5 Another line of research studies the inner structures of lexical features: e.g.  ... 
doi:10.18653/v1/n16-1117 dblp:conf/naacl/YuDAG16 fatcat:2ron3mbuefhohje6hpwljnpicm
« Previous Showing results 1 — 15 out of 12,200 results