Determinantal Point Processes for Image Processing

Determinantal point processes (DPPs) are probabilistic models of configurations that favour diversity or repulsion. They have recently gained influence in the machine learning community, mainly because of their ability to elegantly and efficiently subsample large sets of data. In this paper, we consider DPPs from an image processing perspective, meaning that the data we want to subsample are pixels or patches of a given image. To this end, our framework is discrete and finite. First, we adapt their basic definition and properties to DPPs defined on the pixels of an image, that we call determinantal pixel processes (DPixPs). We are mainly interested in the repulsion properties of such a process and we apply DPixPs to texture synthesis using shot noise models. Finally, we study DPPs on the set of patches of an image. Because of their repulsive property, DPPs provide a strong tool to subsample discrete distributions such as that of image patches.

[1]  Victor-Emmanuel Brunel,et al.  Learning Signed Determinantal Point Processes through the Principal Minor Assignment Problem , 2018, NeurIPS.

[2]  Josiane Zerubia,et al.  Automatic Flamingo detection using a multiple birth and death process , 2008, 2008 IEEE International Conference on Acoustics, Speech and Signal Processing.

[3]  J. Zerubia,et al.  Point processes in forestry : an application to tree crown detection , 2006 .

[4]  Krzysztof Choromanski,et al.  Notes on using Determinantal Point Processes for Clustering with Applications to Text Clustering , 2014, ArXiv.

[5]  Alexander Soshnikov.  Gaussian limit for determinantal random point fields , 2000 .

[6]  Ares Lagae,et al.  Procedural noise using sparse Gabor convolution , 2009, SIGGRAPH '09.

[7]  B. Delyon,et al.  Mixing properties and central limit theorem for associated point processes , 2017, Bernoulli.

[8]  Bruno Galerne,et al.  Texton Noise , 2017, Comput. Graph. Forum.

[9]  Ulrich Paquet,et al.  Low-Rank Factorization of Determinantal Point Processes , 2017, AAAI.

[10]  Steven W. Shaw,et al.  Circulant Matrices and Their Application to Vibration Analysis , 2014 .

[11]  T. Shirai,et al.  Random point fields associated with certain Fredholm determinants I: fermion, Poisson and boson point processes , 2003 .

[12]  T. Shirai,et al.  Fermion Process and Fredholm Determinant , 2000 .

[13]  Rémi Bardenet,et al.  Inference for determinantal point processes without spectral knowledge , 2015, NIPS.

[14]  Fr'ed'eric Lavancier,et al.  Quantifying repulsiveness of determinantal point processes , 2014, 1406.2796.

[15]  Nisheeth K. Vishnoi,et al.  Fair and Diverse DPP-based Data Summarization , 2018, ICML.

[16]  J. van Wijk,et al.  Spot noise texture synthesis for data visualization , 1991, SIGGRAPH.

[17]  Hans Schneider,et al.  Matrices diagonally similar to a symmetric matrix , 1980 .

[18]  Michael Lam,et al.  Unsupervised Video Summarization with Adversarial LSTM Networks , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[19]  Zhijing Yang,et al.  Dimensionality reduction based on determinantal point process and singular spectrum analysis for hyperspectral images , 2019, IET Image Process..

[20]  B. David Saunders,et al.  Flows on graphs applied to diagonal similarity and diagonal equivalence for matrices , 1978, Discret. Math..

[21]  M. Combescure,et al.  Block-circulant matrices with circulant blocks, Weil sums, and mutually unbiased bases. II. The prime power case , 2007, 0710.5643.

[22]  T. Shirai,et al.  Random point fields associated with certain Fredholm determinants II: Fermion shifts and their ergodic and Gibbs properties , 2003 .

[23]  Ke Zhang,et al.  Video Summarization with Long Short-Term Memory , 2016, ECCV.

[24]  Bruno Galerne,et al.  Gabor noise by example , 2012, ACM Trans. Graph..

[25]  A. Soshnikov.  Determinantal random point fields , 2000, math/0002099.

[26]  Charu C. Aggarwal,et al.  Outlier Analysis , 2013, Springer New York.

[27]  Ben Taskar,et al.  Determinantal Point Processes for Machine Learning , 2012, Found. Trends Mach. Learn..

[28]  Y. Peres,et al.  Determinantal Processes and Independence , 2005, math/0503110.

[29]  Kristen Grauman,et al.  Diverse Sequential Subset Selection for Supervised Video Summarization , 2014, NIPS.

[30]  M. Stevens.  Equivalent symmetric kernels of determinantal point processes , 2019, Random Matrices: Theory and Applications.

[31]  Standard and robust intensity parameter estimation for stationary determinantal point processes , 2016, 1603.07460.

[32]  Ankur Moitra,et al.  Learning Determinantal Point Processes with Moments and Cycles , 2017, ICML.

[33]  Nicolas Tremblay,et al.  Asymptotic equivalence of fixed-size and varying-size determinantal point processes , 2018, Bernoulli.

[34]  Bruno Galerne,et al.  Random Phase Textures: Theory and Synthesis , 2011, IEEE Transactions on Image Processing.

[35]  Raphael Loewy,et al.  Principal minors and diagonal similarity of matrices , 1986 .

[36]  Gilles Pagès,et al.  Introduction to vector quantization and its applications for numerics , 2015 .

[37]  Russell Lyons,et al.  Stationary determinantal processes: Phase multiplicity, Bernoullicity, entropy, and domination , 2002 .

[38]  Ben Taskar,et al.  k-DPPs: Fixed-Size Determinantal Point Processes , 2011, ICML.

[39]  Yair Weiss,et al.  From learning models of natural image patches to whole image restoration , 2011, 2011 International Conference on Computer Vision.

[40]  Alessandro Foi,et al.  Image Denoising by Sparse 3-D Transform-Domain Collaborative Filtering , 2007, IEEE Transactions on Image Processing.

[41]  Philip Wolfe,et al.  Validation of subgradient optimization , 1974, Math. Program..

[42]  O. Macchi.  The coincidence approach to stochastic point processes , 1975, Advances in Applied Probability.

[43]  Arthur Leclaire,et al.  Determinantal Patch Processes for Texture Synthesis , 2019 .

[44]  Luca Vogt.  Statistics For Spatial Data , 2016 .

[45]  Ankur Moitra,et al.  Rates of estimation for determinantal point processes , 2017, COLT.

[46]  Yuval Peres,et al.  Zeros of Gaussian Analytic Functions and Determinantal Point Processes , 2009, University Lecture Series.

[47]  Pierre Chainais,et al.  A determinantal point process for column subset selection , 2018, J. Mach. Learn. Res..

[48]  Ben Taskar,et al.  Learning the Parameters of Determinantal Point Process Kernels , 2014, ICML.

[49]  Fr'ed'eric Lavancier,et al.  Contrast Estimation for Parametric Stationary Determinantal Point Processes , 2015, 1510.04222.

[50]  François Baccelli,et al.  Stochastic Geometry and Wireless Networks, Volume 1: Theory , 2009, Found. Trends Netw..

[51]  Yann Strozecki,et al.  From patches to pixels in Non-Local methods: Weighted-average reprojection , 2010, 2010 IEEE International Conference on Image Processing.

[52]  Pierre-Olivier Amblard,et al.  Optimized Algorithms to Sample Determinantal Point Processes , 2018, ArXiv.

[53]  Pierre-Olivier Amblard,et al.  Determinantal Point Processes for Coresets , 2018, J. Mach. Learn. Res..

[54]  Ajith Ramanathan,et al.  Practical Diversified Recommendations on YouTube with Determinantal Point Processes , 2018, CIKM.

[55]  R. Waagepetersen,et al.  Adaptive estimating function inference for nonstationary determinantal point processes , 2018, Scandinavian Journal of Statistics.

[56]  S. R. Simanca,et al.  On Circulant Matrices , 2012 .

[57]  J. Møller,et al.  Determinantal point process models and statistical inference , 2012, 1205.4818.

[58]  Ben Taskar,et al.  Learning Determinantal Point Processes , 2011, UAI.

[59]  D. J. Hartfiel,et al.  On matrices having equal corresponding principal minors , 1984 .

[60]  Charles Bouveyron,et al.  High-Dimensional Mixture Models for Unsupervised Image Denoising (HDMI) , 2018, SIAM J. Imaging Sci..

[61]  Ben Taskar,et al.  An efficient algorithm for the symmetric principal minor assignment problem , 2015 .