{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,11]],"date-time":"2026-05-11T13:46:29Z","timestamp":1778507189764,"version":"3.51.4"},"reference-count":34,"publisher":"Springer Science and Business Media LLC","issue":"6","license":[{"start":{"date-parts":[[2024,9,1]],"date-time":"2024-09-01T00:00:00Z","timestamp":1725148800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2024,9,2]],"date-time":"2024-09-02T00:00:00Z","timestamp":1725235200000},"content-version":"vor","delay-in-days":1,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"DOI":"10.13039\/501100005362","name":"Universita degli Studi di Bari Aldo Moro","doi-asserted-by":"publisher","award":["H93C23000720001"],"award-info":[{"award-number":["H93C23000720001"]}],"id":[{"id":"10.13039\/501100005362","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100021856","name":"Ministero dell\u2019Universit\u00e0e della Ricerca","doi-asserted-by":"publisher","award":["H53D23008870001"],"award-info":[{"award-number":["H53D23008870001"]}],"id":[{"id":"10.13039\/501100021856","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Universit\u00e0 degli Studi di Bari Aldo Moro"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Comp. Appl. Math."],"published-print":{"date-parts":[[2024,9]]},"abstract":"<jats:title>Abstract<\/jats:title><jats:p>Nonnegative matrix factorization (NMF) is a popular dimensionality reduction technique. NMF is typically cast as a non-convex optimization problem solved via standard iterative schemes, such as coordinate descent methods. Hence the choice of the initialization for the variables is crucial as it will influence the factorization quality and the convergence speed. Different strategies have been proposed in the literature, the most popular ones rely on singular value decomposition (SVD). In particular, Atif et al. (Pattern Recognit Lett 122:53\u201359, 2019) have introduced a very efficient SVD-based initialization, namely NNSVD-LRC, that overcomes the drawbacks of previous methods, namely, it guarantees that (i)\u00a0the error decreases as the factorization rank increases, (ii)\u00a0the initial factors are sparse, and (iii)\u00a0the computational cost is low. In this paper, we improve upon NNSVD-LRC by using the low-rank structure of the residual matrix; this allows us to obtain NMF initializations with similar quality to NNSVD-LRC (in terms of error and sparsity) while reducing the computational load. We evaluate our proposed solution over other NMF initializations on several real dense and sparse datasets.<\/jats:p>","DOI":"10.1007\/s40314-024-02905-1","type":"journal-article","created":{"date-parts":[[2024,9,2]],"date-time":"2024-09-02T19:02:23Z","timestamp":1725303743000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":4,"title":["Accelerated SVD-based initialization for nonnegative matrix factorization"],"prefix":"10.1007","volume":"43","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-2791-9610","authenticated-orcid":false,"given":"Flavia","family":"Esposito","sequence":"first","affiliation":[]},{"given":"Syed Muhammad","family":"Atif","sequence":"additional","affiliation":[]},{"given":"Nicolas","family":"Gillis","sequence":"additional","affiliation":[]}],"member":"297","published-online":{"date-parts":[[2024,9,2]]},"reference":[{"key":"2905_CR1","doi-asserted-by":"crossref","unstructured":"Ara\u00fajo MCU, Saldanha TCB, Galvao RKH, Yoneyama T, Chame HC, Visani V (2001) The successive projections algorithm for variable selection in spectroscopic multicomponent analysis. Chemometr Intell Lab Syst 57(2):65\u201373","DOI":"10.1016\/S0169-7439(01)00119-8"},{"key":"2905_CR2","doi-asserted-by":"publisher","first-page":"53","DOI":"10.1016\/j.patrec.2019.02.018","volume":"122","author":"SM Atif","year":"2019","unstructured":"Atif SM, Qazi S, Gillis N (2019) Improved svd-based initialization for nonnegative matrix factorization using low-rank correction. Pattern Recognit Lett 122:53\u201359","journal-title":"Pattern Recognit Lett"},{"key":"2905_CR3","doi-asserted-by":"crossref","unstructured":"Berman A, Plemmons RJ (1994) Nonnegative matrices in the mathematical sciences. SIAM","DOI":"10.1137\/1.9781611971262"},{"issue":"4","key":"2905_CR4","doi-asserted-by":"publisher","first-page":"1350","DOI":"10.1016\/j.patcog.2007.09.010","volume":"41","author":"C Boutsidis","year":"2008","unstructured":"Boutsidis C, Gallopoulos E (2008) Svd based initialization: a head start for nonnegative matrix factorization. Pattern Recognit 41(4):1350\u20131362","journal-title":"Pattern Recognit"},{"issue":"2","key":"2905_CR5","doi-asserted-by":"publisher","first-page":"135","DOI":"10.1002\/cem.1122","volume":"22","author":"R Bro","year":"2008","unstructured":"Bro R, Acar E, Kolda TG (2008) Resolving the sign ambiguity in the singular value decomposition. J Chemometr 22(2):135\u2013140","journal-title":"J Chemometr"},{"key":"2905_CR6","doi-asserted-by":"publisher","DOI":"10.1002\/9780470747278","volume-title":"Nonnegative matrix and tensor factorizations: applications to exploratory multi-way data analysis and blind source separation","author":"A Cichocki","year":"2009","unstructured":"Cichocki A, Zdunek R, Phan AH, Amari SI (2009) Nonnegative matrix and tensor factorizations: applications to exploratory multi-way data analysis and blind source separation. Wiley, New York"},{"key":"2905_CR7","doi-asserted-by":"crossref","unstructured":"Cichocki A, Zdunek R, Amari SI (2007) Hierarchical ALS algorithms for nonnegative matrix and 3d tensor factorization. In: international conference on independent component analysis and signal separation. Springer, Berlin, pp 169\u2013176","DOI":"10.1007\/978-3-540-74494-8_22"},{"issue":"4","key":"2905_CR8","doi-asserted-by":"publisher","first-page":"861","DOI":"10.1007\/s10898-017-0578-x","volume":"74","author":"R Du","year":"2019","unstructured":"Du R, Drake B, Park H (2019) Hybrid clustering based on content and connection structure using joint nonnegative matrix factorization. J Glob Optim 74(4):861\u2013877","journal-title":"J Glob Optim"},{"key":"2905_CR9","unstructured":"Ensari T (2016) Character recognition analysis with nonnegative matrix factorization. Int J Comput 1"},{"issue":"9","key":"2905_CR10","doi-asserted-by":"publisher","first-page":"1006","DOI":"10.3390\/math9091006","volume":"9","author":"F Esposito","year":"2021","unstructured":"Esposito F (2021) A review on initialization methods for nonnegative matrix factorization: towards omics data experiments. Mathematics 9(9):1006","journal-title":"Mathematics"},{"key":"2905_CR11","doi-asserted-by":"crossref","unstructured":"Geng X, Ji L, Sun K (2016) Non-negative matrix factorization based unmixing for principal component transformed hyperspectral data. Front Inf Technol Electron Eng 403\u2013412","DOI":"10.1631\/FITEE.1600028"},{"key":"2905_CR12","doi-asserted-by":"publisher","DOI":"10.1137\/1.9781611976410","volume-title":"Nonnegative matrix factorization","author":"N Gillis","year":"2020","unstructured":"Gillis N (2020) Nonnegative matrix factorization. SIAM, Philadelphia"},{"issue":"4","key":"2905_CR13","doi-asserted-by":"publisher","first-page":"1085","DOI":"10.1162\/NECO_a_00256","volume":"24","author":"N Gillis","year":"2012","unstructured":"Gillis N, Glineur F (2012) Accelerated multiplicative updates and hierarchical als algorithms for nonnegative matrix factorization. Neural Comput 24(4):1085\u20131105","journal-title":"Neural Comput"},{"key":"2905_CR14","unstructured":"Gillis N, Glineur F (2008) Nonnegative factorization and the maximum edge biclique problem. arXiv:0810.4225"},{"key":"2905_CR15","doi-asserted-by":"crossref","unstructured":"Janecek A, Tan Y (2011) Using population based algorithms for initializing nonnegative matrix factorization. In: International conference in swarm intelligence. Springer, Berlin, pp 307\u2013316","DOI":"10.1007\/978-3-642-21524-7_37"},{"key":"2905_CR16","unstructured":"Langville AN, Meyer CD, Albright R, Cox J, Duling D (2006) Initializations for the nonnegative matrix factorization. In: Proceedings of the twelfth ACM SIGKDD international conference on knowledge discovery and data mining, Citeseer, pp 23\u201326"},{"issue":"18","key":"2905_CR17","doi-asserted-by":"publisher","first-page":"4717","DOI":"10.1109\/TSP.2017.2713761","volume":"65","author":"Z Liu","year":"2017","unstructured":"Liu Z, Tan VY (2017) Rank-one nmf-based initialization for nmf and relative error bounds under a geometric assumption. IEEE Trans Signal Process 65(18):4717\u20134731","journal-title":"IEEE Trans Signal Process"},{"key":"2905_CR18","doi-asserted-by":"crossref","unstructured":"Liu Z, Tan V (2018) Rank-one NMF-based initialization for NMF and relative error bounds under a geometric assumption. Information Theory And Applications Workshop (ITA), pp. 1-15","DOI":"10.1109\/ITA.2018.8503169"},{"issue":"9","key":"2905_CR19","doi-asserted-by":"publisher","first-page":"1464","DOI":"10.1177\/0003702816662600","volume":"70","author":"R Luce","year":"2016","unstructured":"Luce R, Hildebrandt P, Kuhlmann U, Liesen J (2016) Using separable nonnegative matrix factorization techniques for the analysis of time-resolved raman spectra. Appl Spectrosc 70(9):1464\u20131475","journal-title":"Appl Spectrosc"},{"key":"2905_CR20","doi-asserted-by":"publisher","first-page":"11","DOI":"10.1016\/j.neunet.2014.03.007","volume":"55","author":"R Maruyama","year":"2014","unstructured":"Maruyama R, Maeda K, Moroda H, Kato I, Inoue M, Miyakawa H, Aonishi T (2014) Detecting cells using non-negative matrix factorization on calcium imaging data. Neural Netw 55:11\u201319","journal-title":"Neural Netw"},{"key":"2905_CR21","doi-asserted-by":"crossref","unstructured":"Nascimento J, Dias J (2005) Vertex component analysis: a fast algorithm to unmix hyperspectral data. IEEE Trans Geosci Remote Sens 898\u2013910","DOI":"10.1109\/TGRS.2005.844293"},{"key":"2905_CR22","unstructured":"Nesterov Y (1983) A method of solving a convex programming problem with convergence rate $$O(1\/k^2)$$. In: Sov. Math. Dokl, vol 27, pp 372\u2013376"},{"key":"2905_CR23","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-319-91578-4","volume-title":"Lectures on convex optimization","author":"Y Nesterov","year":"2018","unstructured":"Nesterov Y et al (2018) Lectures on convex optimization, vol 137. Springer, Berlin"},{"key":"2905_CR24","doi-asserted-by":"crossref","unstructured":"O\u2019donoghue B, Candes E (2015) Adaptive restart for accelerated gradient schemes. Found Comput Math 15(3):715\u2013732","DOI":"10.1007\/s10208-013-9150-3"},{"issue":"3","key":"2905_CR25","first-page":"600","volume":"4","author":"SJ Prajapati","year":"2015","unstructured":"Prajapati SJ, Jadhav KR (2015) Brain tumor detection by various image segmentation techniques with introduction to non negative matrix factorization. Brain 4(3):600\u20133","journal-title":"Brain"},{"key":"2905_CR26","doi-asserted-by":"publisher","first-page":"71","DOI":"10.1016\/j.patrec.2015.05.019","volume":"63","author":"H Qiao","year":"2015","unstructured":"Qiao H (2015) New svd based initialization strategy for non-negative matrix factorization. Pattern Recognit Lett 63:71\u201377","journal-title":"Pattern Recognit Lett"},{"issue":"2","key":"2905_CR27","doi-asserted-by":"publisher","first-page":"354","DOI":"10.3923\/jas.2011.354.359","volume":"11","author":"M Rezaei","year":"2011","unstructured":"Rezaei M, Boostani R, Rezaei M (2011) An efficient initialization method for nonnegative matrix factorization. J Appl Sci 11(2):354\u2013359","journal-title":"J Appl Sci"},{"issue":"8","key":"2905_CR28","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pone.0180268","volume":"12","author":"N Sauwen","year":"2017","unstructured":"Sauwen N, Acou M, Bharath HN, Sima DM, Veraart J, Maes F, Himmelreich U, Achten E, Van Huffel S (2017) The successive projection algorithm as an initialization method for brain tumor segmentation using non-negative matrix factorization. Plos One 12(8):e0180268","journal-title":"Plos One"},{"key":"2905_CR29","doi-asserted-by":"publisher","first-page":"43","DOI":"10.1016\/j.ultramic.2016.08.006","volume":"170","author":"M Shiga","year":"2016","unstructured":"Shiga M, Tatsumi K, Muto S, Tsuda K, Yamamoto Y, Mori T, Tanji T (2016) Sparse modeling of eels and edx spectral imaging data by nonnegative matrix factorization. Ultramicroscopy 170:43\u201359","journal-title":"Ultramicroscopy"},{"key":"2905_CR30","doi-asserted-by":"crossref","unstructured":"Wang X, Xie X, Lu L (2012) An effective initialization for orthogonal nonnegative matrix factorization. J Comput Math 34\u201346","DOI":"10.4208\/jcm.1110-m11si10"},{"issue":"11","key":"2905_CR31","doi-asserted-by":"publisher","first-page":"2217","DOI":"10.1016\/j.patcog.2004.02.013","volume":"37","author":"S Wild","year":"2004","unstructured":"Wild S, Curry J, Dougherty A (2004) Improving non-negative matrix factorizations through structured initialization. Pattern Recognit 37(11):2217\u20132232","journal-title":"Pattern Recognit"},{"key":"2905_CR32","doi-asserted-by":"crossref","unstructured":"Yoshii K, Itoyama K, Goto M (2016) Student\u2019s t nonnegative matrix factorization and positive semidefinite tensor factorization for single-channel audio source separation. In: 2016 IEEE international conference on acoustics, speech and signal processing (ICASSP), IEEE, pp 51\u201355","DOI":"10.1109\/ICASSP.2016.7471635"},{"key":"2905_CR33","doi-asserted-by":"crossref","unstructured":"Zdunek R (2012) Initialization of nonnegative matrix factorization with vertices of convex polytope. In: International conference on artificial intelligence and soft computing, pp 448\u2013455","DOI":"10.1007\/978-3-642-29347-4_52"},{"issue":"3","key":"2905_CR34","doi-asserted-by":"publisher","first-page":"374","DOI":"10.1007\/s10115-004-0194-1","volume":"8","author":"S Zhong","year":"2005","unstructured":"Zhong S, Ghosh J (2005) Generative model-based document clustering: a comparative study. Knowl Inf Syst 8(3):374\u2013384","journal-title":"Knowl Inf Syst"}],"container-title":["Computational and Applied Mathematics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s40314-024-02905-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s40314-024-02905-1\/fulltext.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s40314-024-02905-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,9,13]],"date-time":"2024-09-13T13:12:55Z","timestamp":1726233175000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s40314-024-02905-1"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,9]]},"references-count":34,"journal-issue":{"issue":"6","published-print":{"date-parts":[[2024,9]]}},"alternative-id":["2905"],"URL":"https:\/\/doi.org\/10.1007\/s40314-024-02905-1","relation":{},"ISSN":["2238-3603","1807-0302"],"issn-type":[{"value":"2238-3603","type":"print"},{"value":"1807-0302","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,9]]},"assertion":[{"value":"29 February 2024","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"23 July 2024","order":2,"name":"revised","label":"Revised","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"12 August 2024","order":3,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"2 September 2024","order":4,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declaration"}},{"value":"Not applicable.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Conflict of interest"}},{"value":"Not applicable.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethics approval"}},{"value":"Not applicable.","order":4,"name":"Ethics","group":{"name":"EthicsHeading","label":"Consent to participate"}},{"value":"All authors consent to the publication of the work.","order":5,"name":"Ethics","group":{"name":"EthicsHeading","label":"Consent for publication"}},{"value":"The code of accNNSVD-PRP is available on GitHub at","order":6,"name":"Ethics","group":{"name":"EthicsHeading","label":"Code availability"}}],"article-number":"392"}}