26,649 Hits in 5.3 sec

Model Selection for Non-Negative Tensor Factorization with Minimum Description Length

Yunhui Fu, Shin Matsushima, Kenji Yamanishi
2019 Entropy  
This paper proposes a novel rank selection criterion for NTF on the basis of the minimum description length (MDL) principle.  ...  Non-negative tensor factorization (NTF) is a widely used multi-way analysis approach that factorizes a high-order non-negative data tensor into several non-negative factor matrices.  ...  In this paper, we propose a method to select an appropriate value of rank for NTF based on the minimum description length (MDL) principle [7] .  ... 
doi:10.3390/e21070632 pmid:33267345 fatcat:374cxtwqtjbq3pxqt6cklx3rpa

Walk 'n' Merge: A Scalable Algorithm for Boolean Tensor Factorization

Dora Erdos, Pauli Miettinen
2013 2013 IEEE 13th International Conference on Data Mining  
Unfortunately the algorithms for computing Boolean tensor factorizations do not usually scale well.  ...  Tensors are becoming increasingly common in data mining, and consequently, tensor factorizations are becoming more important tools for data miners.  ...  To compute the encoding length, we modify the Typed XOR Data-to-Model encoding for encoding Boolean matrix factorizations [7] to work with tensor factorizations.  ... 
doi:10.1109/icdm.2013.141 dblp:conf/icdm/ErdosM13 fatcat:qfij7jqu35gtvhum5midqo2n7e

Scalable Boolean Tensor Factorizations using Random Walks [article]

Dóra Erdős, Pauli Miettinen
2013 arXiv   pre-print
Unfortunately the algorithms for computing Boolean tensor factorizations do not usually scale well.  ...  Tensors are becoming increasingly common in data mining, and consequently, tensor factorizations are becoming more and more important tools for data miners.  ...  Minimum Description Length Principle. The intuition behind the MDL principle is that the best model is the one that allows us to compress the data best.  ... 
arXiv:1310.4843v1 fatcat:g4r5owe4yjd4ral5n4cu3axaam

A Comparison of Non-negative Tucker Decomposition and Parallel Factor Analysis for Identification and Measurement of Human EEG Rhythms

Zuzana Rošt'áková, Roman Rosipal, Saman Seifpour, Leonardo Jose Trejo
2020 Measurement Science Review  
Parallel factor analysis (PARAFAC) is a powerful approach for this purpose.  ...  In this article we discuss the potential benefits of an alternative tensor decomposition method – the Tucker model.  ...  The total length of the selected EEG recordings for the second subject was 40.9 minutes.  ... 
doi:10.2478/msr-2020-0015 fatcat:ra4mzmkszffjjm2x7g5c7vpimi

J-Recs: Principled and Scalable Recommendation Justification [article]

Namyong Park, Andrey Kan, Christos Faloutsos, Xin Luna Dong
2020 arXiv   pre-print
J-Recs is a recommendation model-agnostic method that generates diverse justifications based on various types of product and user data (e.g., purchase history and product attributes).  ...  We address these limitations of earlier approaches by developing J-Recs, a method for producing concise and diverse justifications.  ...  Papers on the minimum description length principle are highlighted in blue font.  ... 
arXiv:2011.05928v1 fatcat:3dqm6aqazzbc5fcmwiapv7ewza

Tensor Factorization of Brain Structural Graph for Unsupervised Classification in Multiple Sclerosis

Beradino Barile, Aldo Marzullo, Claudio Stamile, Francoise Durand-Dubief, Dominique Sappey-Marinier
2021 2020 25th International Conference on Pattern Recognition (ICPR)  
Using non-negative tensor factorization (NTF), we first focused on the detection of pathological patterns of the brain WM network affected by significant longitudinal variations.  ...  In this work, we propose a fully automated tensor-based algorithm for the classification of MS clinical forms based on the structural connectivity graph of the white matter (WM) network.  ...  and Non-negative Tensor Factorization (NTF) [24] .  ... 
doi:10.1109/icpr48806.2021.9412491 fatcat:qolgpjr6b5hb7iv74xbdpu3ovm

Context-aware recommendations from implicit data via scalable tensor factorization [article]

Balázs Hidasi, Domonkos Tikk
2013 arXiv   pre-print
number of non-zero elements in the tensor.  ...  In this paper, we propose a generic context-aware implicit feedback recommender algorithm, coined iTALS. iTALS applies a fast, ALS-based tensor factorization learning method that scales linearly with the  ...  ALS based fast tensor factorization In this section we present iTALS, a general ALS-based tensor factorization algorithm that scales linearly with the non-zero element of a dense tensor (when appropriate  ... 
arXiv:1309.7611v1 fatcat:lpyd6ktghfh7dfe2cmmqd33adm

A Sub-grid Scale Energy Dissipation Rate Model for Large-eddy Spray Simulations [article]

Hongjiang Li, Christopher J. Rutland, Francisco E. Hernandez Perez, Hong G. Im
2020 arXiv   pre-print
In large eddy simulation (LES) with one-equation turbulence models, the energy dissipation process is modeled by a rate term in the transport equation of the subgrid-scale (SGS) kinetic energy.  ...  energy dissipation rate with the SGS kinetic energy and LES fitter width.  ...  Figure 8 . 8 Contour plots of the fuel mixture fraction at various selected times ASOI for three meshes with minimum cell sizes of 0.5, 0.375, and 0.25 mm.  ... 
arXiv:2001.07654v1 fatcat:ppucrfi7ufavpphlir6vlezhx4

EEG-based Classification of Epileptic and Non-Epileptic Events using Multi-Array Decomposition

Evangelia Pippa, Vasileios G. Kanas, Evangelia I. Zacharaki, Vasiliki Tsirka, Michael Koutroumanidis, Vasileios Megalooikonomou
2016 International Journal of Monitoring and Surveillance Technologies Research  
Initially, the incoming EEG signals were frame blocked to epochs with a 19 sliding Hamming window of length and without time-overlap between successive epochs.  ...  The core tensor 11consists of signature features of projected onto the factor subspace spammed by and .  ... 
doi:10.4018/ijmstr.2016040101 fatcat:ercixprfkfhqfeivoykyw6rmza

A molecular viewer for the analysis of TLS rigid-body motion in macromolecules

Jay Painter, Ethan A. Merritt
2005 Acta Crystallographica Section D: Biological Crystallography  
TLSView may also be used to prepare, analyze and validate TLS models for crystallographic refinement.  ...  The use of TLS models for protein motion has been relatively limited, partly because the physical meaning of the refined TLS parameters is not intuitive.  ...  Animation of screw motion The three TLS non-intersecting orthogonal screw axes are drawn as rods with a length proportional to their The same view and TLS model for 1r75 as in Fig. 2 is shown here in  ... 
doi:10.1107/s0907444905001897 pmid:15809496 fatcat:opz7cztzu5elxkavk6rfeb7laa


2014 International Journal of Neural Systems  
Non-negative tensor factorization (NTF) has been successfully applied to analyze event-related potentials (ERPs), and shown superiority in terms of capturing multi-domain features.  ...  However, the time-frequency representation of ERPs by higher-order tensors are usually large-scale, which prevents the popularity of most tensor factorization algorithms.  ...  It should be noted that: (1) tensor decomposition with the non-negative constraint for a non-negative tensor has been named as NTF (non-negative tensor factorization), 3, 4 and that (2) tensor decomposition  ... 
doi:10.1142/s012906571440005x pmid:25164246 fatcat:36l3wmqy25cnbmj53vzhjprd2u

Unsupervised Discovery of Demixed, Low-Dimensional Neural Dynamics across Multiple Timescales through Tensor Component Analysis

Alex H. Williams, Tony Hyun Kim, Forea Wang, Saurabh Vyas, Stephen I. Ryu, Krishna V. Shenoy, Mark Schnitzer, Tamara G. Kolda, Surya Ganguli
2018 Neuron  
We demonstrate a simple tensor component analysis (TCA) can meet this challenge by extracting three interconnected, low-dimensional descriptions of neural data: neuron factors, reflecting cell assemblies  ...  While recent experimental advances enable large-scale and long-term neural recordings with high temporal fidelity, it remains a formidable challenge to extract unbiased and interpretable descriptions of  ...  ACKNOWLEDGMENTS The authors thank Subhaneil Lahiri (Stanford University), Jeff Seely (Cognes-cent Corporation), and Casey Battaglino (Georgia Tech) for discussions pertaining to this work.  ... 
doi:10.1016/j.neuron.2018.05.015 pmid:29887338 pmcid:PMC6907734 fatcat:7wfwookp2nbttesy4bxnjnhmwa

User-Aware Image Tag Refinement via Ternary Semantic Analysis

Jitao Sang, Changsheng Xu, Jing Liu
2012 IEEE transactions on multimedia  
To tackle the problem of tag refinement, we propose a method of Ranking based Multi-correlation Tensor Factorization (RMTF), to jointly model the ternary relations among user, image, and tag, and further  ...  During the model estimation, we employ a ranking based optimization scheme to interpret the tagging data, in which the pair-wise qualitative difference between positive and negative examples is used, instead  ...  The minus signs indicate the filtered negative triplets. Any tag is considered a better description for image than all the tags .  ... 
doi:10.1109/tmm.2012.2188782 fatcat:lbllgvb4fncnzk7zuwsmwkozwi

Measurement ofT20in Elastic Electron-Deuteron Scattering

M. Bouwhuis, R. Alarcon, T. Botto, J. F. J. van den Brand, H. J. Bulten, S. Dolfini, R. Ent, M. Ferro-Luzzi, D. W. Higinbotham, C. W. de Jager, J. Lang, D. J. J. de Lange (+8 others)
1999 Physical Review Letters  
The tensor polarization of the deuterium nuclei was determined with an ion-extraction system, allowing an absolute measurement of T20.  ...  The data are described well by a non-relativistic calculation that includes the effects of meson-exchange currents.  ...  To correct for the fact that the direction of the holding field-and thus the spin orientation-varies over the length of the cell with respect to q the uncorrected tensor asymmetry A T d is weighted with  ... 
doi:10.1103/physrevlett.82.3755 fatcat:c6rk64d4ojbzboxwww7o6xeb3e

Online NON-negative Tensor Deconvolution for source detection in 3DTV audio

Yuki Mitsufuji, Marco Liuni, Alex Baker, Axel Roebel
2014 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)  
An algorithm based on online Non-negative Tensor Deconvolution is realized, to deal with sound sources with time dependent positions in the channel matrix.  ...  shown, compared with other decomposition techniques.  ...  This approach has been extended to tensors in [15] , with a different formalism, leading to Non-negative Tensor Deconvolution (NTD).  ... 
doi:10.1109/icassp.2014.6854167 dblp:conf/icassp/MitsufujiLBR14 fatcat:emkc5tcxgnaahblb3uj7m6zuja
« Previous Showing results 1 — 15 out of 26,649 results