Constrained low-rank matrix approximations have been known for decades as powerful linear dimensi... more Constrained low-rank matrix approximations have been known for decades as powerful linear dimensionality reduction techniques to be able to extract the information contained in large data sets in a relevant way. However, such low-rank approaches are unable to mine complex, interleaved features that underlie hierarchical semantics. Recently, deep matrix factorization (deep MF) was introduced to deal with the extraction of several layers of features and has been shown to reach outstanding performances on unsupervised tasks. Deep MF was motivated by the success of deep learning, as it is conceptually close to some neural networks paradigms. In this paper, we present the main models, algorithms, and applications of deep MF through a comprehensive literature review. We also discuss theoretical questions and perspectives of research.
Nonnegative matrix factorization (NMF) is a widely used linear dimensionality reduction technique... more Nonnegative matrix factorization (NMF) is a widely used linear dimensionality reduction technique for nonnegative data. NMF requires that each data point is approximated by a convex combination of basis elements. Archetypal analysis (AA), also referred to as convex NMF, is a well-known NMF variant imposing that the basis elements are themselves convex combinations of the data points. AA has the advantage to be more interpretable than NMF because the basis elements are directly constructed from the data points. However, it usually suffers from a high data fitting error because the basis elements are constrained to be contained in the convex cone of the data points. In this letter, we introduce near-convex ar-chetypal analysis (NCAA) which combines the advantages of both AA and NMF. As for AA, the basis vectors are required to be linear combinations of the data points and hence are easily interpretable. As for NMF, the additional flexibility in choosing the basis elements allows NCAA to have a low data fitting error. We show that NCAA compares favorably with a state-of-the-art minimum-volume NMF method on synthetic datasets and on a real-world hyperspectral image.
This paper reports on the LaughterCycle project, held during a three month period between April a... more This paper reports on the LaughterCycle project, held during a three month period between April and June 2009, within the numediart research programme. In this project, we have been developing technological building blocks for an application allowing to record ...
DI-fusion, le Dépôt institutionnel numérique de l'ULB, est l... more DI-fusion, le Dépôt institutionnel numérique de l'ULB, est l'outil de référencementde la production scientifique de l'ULB.L'interface de recherche DI-fusion permet de consulter les publications des chercheurs de l'ULB et les thèses qui y ont été défendues.
Even though video compression has become a mature field, a lot of research is still ongoing. Inde... more Even though video compression has become a mature field, a lot of research is still ongoing. Indeed, as the quality of the compressed video for a given size or bit rate increases, so does users’ level of expectations and their intolerance to artefacts. The development of compression technology has enabled number of applications; key applications in television broadcast field. Compression technology is the basis for digital television. The “Video Compression” book was written for scientists and development engineers. The aim of the book is to showcase the state of the art in the wider field of compression beyond encoder centric approach and to appreciate the need for video quality assurance. It covers compressive video coding, distributed video coding, motion estimation and video quality.
Many bioactive peptides require amidation of their carboxy terminus to exhibit full biological ac... more Many bioactive peptides require amidation of their carboxy terminus to exhibit full biological activity. Peptidylglycine alpha-hydroxylating monooxygenase (PHM; EC 1.14.17.3), the enzyme that catalyzes the first of the two steps of this reaction, is composed of two domains, each of which binds one copper atom (CuH and CuM). The CuM site includes Met(314) and two His residues as ligands. Mutation of Met(314) to Ile inactivates PHM, but has only a minimal effect on the EXAFS spectrum of the oxidized enzyme, implying that it contributes only marginally to stabilization of the CuM site. To characterize the role of Met(314) as a CuM ligand, we determined the structure of the Met(314)Ile-PHM mutant. Since the mutant protein failed to crystallize in the conditions of the original wild-type protein, this structure determination required finding a new crystal form. The Met(314)Ile-PHM mutant structure confirms that the mutation does not abolish CuM binding to the enzyme, but causes other str...
Constrained low-rank matrix approximations have been known for decades as powerful linear dimensi... more Constrained low-rank matrix approximations have been known for decades as powerful linear dimensionality reduction techniques to be able to extract the information contained in large data sets in a relevant way. However, such low-rank approaches are unable to mine complex, interleaved features that underlie hierarchical semantics. Recently, deep matrix factorization (deep MF) was introduced to deal with the extraction of several layers of features and has been shown to reach outstanding performances on unsupervised tasks. Deep MF was motivated by the success of deep learning, as it is conceptually close to some neural networks paradigms. In this paper, we present the main models, algorithms, and applications of deep MF through a comprehensive literature review. We also discuss theoretical questions and perspectives of research.
Nonnegative matrix factorization (NMF) is a widely used linear dimensionality reduction technique... more Nonnegative matrix factorization (NMF) is a widely used linear dimensionality reduction technique for nonnegative data. NMF requires that each data point is approximated by a convex combination of basis elements. Archetypal analysis (AA), also referred to as convex NMF, is a well-known NMF variant imposing that the basis elements are themselves convex combinations of the data points. AA has the advantage to be more interpretable than NMF because the basis elements are directly constructed from the data points. However, it usually suffers from a high data fitting error because the basis elements are constrained to be contained in the convex cone of the data points. In this letter, we introduce near-convex ar-chetypal analysis (NCAA) which combines the advantages of both AA and NMF. As for AA, the basis vectors are required to be linear combinations of the data points and hence are easily interpretable. As for NMF, the additional flexibility in choosing the basis elements allows NCAA to have a low data fitting error. We show that NCAA compares favorably with a state-of-the-art minimum-volume NMF method on synthetic datasets and on a real-world hyperspectral image.
This paper reports on the LaughterCycle project, held during a three month period between April a... more This paper reports on the LaughterCycle project, held during a three month period between April and June 2009, within the numediart research programme. In this project, we have been developing technological building blocks for an application allowing to record ...
DI-fusion, le Dépôt institutionnel numérique de l'ULB, est l... more DI-fusion, le Dépôt institutionnel numérique de l'ULB, est l'outil de référencementde la production scientifique de l'ULB.L'interface de recherche DI-fusion permet de consulter les publications des chercheurs de l'ULB et les thèses qui y ont été défendues.
Even though video compression has become a mature field, a lot of research is still ongoing. Inde... more Even though video compression has become a mature field, a lot of research is still ongoing. Indeed, as the quality of the compressed video for a given size or bit rate increases, so does users’ level of expectations and their intolerance to artefacts. The development of compression technology has enabled number of applications; key applications in television broadcast field. Compression technology is the basis for digital television. The “Video Compression” book was written for scientists and development engineers. The aim of the book is to showcase the state of the art in the wider field of compression beyond encoder centric approach and to appreciate the need for video quality assurance. It covers compressive video coding, distributed video coding, motion estimation and video quality.
Many bioactive peptides require amidation of their carboxy terminus to exhibit full biological ac... more Many bioactive peptides require amidation of their carboxy terminus to exhibit full biological activity. Peptidylglycine alpha-hydroxylating monooxygenase (PHM; EC 1.14.17.3), the enzyme that catalyzes the first of the two steps of this reaction, is composed of two domains, each of which binds one copper atom (CuH and CuM). The CuM site includes Met(314) and two His residues as ligands. Mutation of Met(314) to Ile inactivates PHM, but has only a minimal effect on the EXAFS spectrum of the oxidized enzyme, implying that it contributes only marginally to stabilization of the CuM site. To characterize the role of Met(314) as a CuM ligand, we determined the structure of the Met(314)Ile-PHM mutant. Since the mutant protein failed to crystallize in the conditions of the original wild-type protein, this structure determination required finding a new crystal form. The Met(314)Ile-PHM mutant structure confirms that the mutation does not abolish CuM binding to the enzyme, but causes other str...
Constrained low-rank matrix approximations have been known for decades as powerful linear dimensi... more Constrained low-rank matrix approximations have been known for decades as powerful linear dimensionality reduction techniques to be able to extract the information contained in large data sets in a relevant way. However, such low-rank approaches are unable to mine complex, interleaved features that underlie hierarchical semantics. Recently, deep matrix factorization (deep MF) was introduced to deal with the extraction of several layers of features and has been shown to reach outstanding performances on unsupervised tasks. Deep MF was motivated by the success of deep learning, as it is conceptually close to some neural networks paradigms. In this paper, we present the main models, algorithms, and applications of deep MF through a comprehensive literature review. We also discuss theoretical questions and perspectives of research.
Uploads
Papers by xavier siebert