131,078 research outputs found
Neural Networks and Machine Learning in Bioinformatics - Theory and Applications
No full textSeiffert U, Hammer B, Kaski S, Villmann T. Neural Networks and Machine Learning in Bioinformatics - Theory and Applications. In: Verleysen M, ed. Proc. Of European Symposium on Artificial Neural Networks. Brussels, Belgium: d-side publications; 2006: 521-532
A Universal Sequence of Tensors for the Asymptotic Rank Conjecture
No full textThe exponent σ(T) of a tensor T ∈ ^d⊗^d⊗^d over a field captures the base of the exponential growth rate of the tensor rank of T under Kronecker powers. Tensor exponents are fundamental from the standpoint of algorithms and computational complexity theory; for example, the exponent ω of square matrix multiplication can be characterized as ω = 2σ(MM₂), where MM₂ ∈ ⁴⊗⁴⊗⁴ is the tensor that represents 2×2 matrix multiplication.
Strassen [FOCS 1986] initiated a duality theory for spaces of tensors that enables one to characterize the exponent of a tensor via objects in a dual space, called the asymptotic spectrum of the primal (tensor) space. While Strassen’s theory has considerable generality beyond the setting of tensors - Wigderson and Zuiddam [Asymptotic Spectra: Theory, Applications, and Extensions, preprint, 2023] give a recent exposition - progress in characterizing the dual space in the tensor setting has been slow, with the first universal points in the dual identified by Christandl, Vrana, and Zuiddam [J. Amer. Math. Soc. 36 (2023)]. In parallel to Strassen’s theory, the algebraic geometry community has developed a geometric theory of tensors aimed at characterizing the structure of the primal space and tensor exponents therein; the latter study was motivated in particular by an observation of Strassen (implicit in [J. Reine Angew. Math. 384 (1988)]) that matrix-multiplication tensors have limited universality in the sense that σ(^d⊗^d⊗^d) ≤ 2ω/3 = 4/3σ(MM₂) holds for all d ≥ 1. In particular, this limited universality of the tensor MM₂ puts forth the question whether one could construct explicit universal tensors that exactly characterize the worst-case tensor exponent in the primal space. Such explicit universal objects would, among others, give means towards a proof or a disproof of Strassen’s asymptotic rank conjecture [Progr. Math. 120 (1994)]; the former would immediately imply ω = 2 and, among others, refute the Set Cover Conjecture (cf. Björklund and Kaski [STOC 2024] and Pratt [STOC 2024]).
Our main result is an explicit construction of a sequence _d of zero-one-valued tensors that is universal for the worst-case tensor exponent; more precisely, we show that σ(_d) = σ(d) where σ(d) = sup_{T ∈ ^d⊗^d⊗^d}σ(T). We also supply an explicit universal sequence _Δ localised to capture the worst-case exponent σ(Δ) of tensors with support contained in Δ ⊆ [d]×[d]×[d]; by combining such sequences, we obtain a universal sequence _d such that σ(_d) = 1 holds if and only if Strassen’s asymptotic rank conjecture holds for d. Finally, we show that the limit lim_{d → ∞}σ(d) exists and can be captured as lim_{d → ∞} σ(D_d) for an explicit sequence (D_d)_{d = 1}^∞ of tensors obtained by diagonalisation of the sequences _d.
As our second result we relate the absence of polynomials of fixed degree vanishing on tensors of low rank, or more generally asymptotic rank, with upper bounds on the exponent σ(d). Using this technique, one may bound asymptotic rank for all tensors of a given format, knowing enough specific tensors of low asymptotic rank
Characterisation of the structure, deuterium quadrupolar tensors, and orientational order of acenaphthene, a rigid, prochiral molecule, from the NMR spectra of samples dissolved in nematic and chiral nematic liquid crystalline solvents
No full textMolecules like acenaphthene which have a point group symmetry of C-2v behave as though their symmetry is C-2 when dissolved in chiral nematic liquid crystalline solvents. To quantify this effect a sample of perdeuterated acenaphthene dissolved in the chiral nematic solvent formed by dissolving poly-(gamma -benzyl-L-glutamate), PBLG, in CHCl3 has been studied by deuterium NMR spectroscopy. The quadrupolar splittings obtained were used to determine the orientational order parameters of acenaphthene-d(10) when dissolved in PBLG/CHCl3. To do this it was necessary to also record and analyse the proton and deuterium spectra given by a sample containing both acenaphthene and acenaphthene-d(10) in a non-chiral liquid crystalline solvent. The proton spectrum is very complex, and was analysed only after first recording and analysing the simpler H-1-{H-2} spectrum given by a sample of acenaphthene-d(6). This procedure finally yielded a set of dipolar couplings for the fully protonated molecule, which after correction for vibrational motion, were used to determine both the relative positions of the protons and the orientational order of the molecules. This information was then used to derive the quadrupolar coupling constants from the measured quadrupolar splittings. The lowering of the symmetry of the orientational distribution function is quantified by the angle alpha by which the principal axes of the molecular orientational order matrix of acenaphthene dissolved in PBLG/CHCl3 are rotated out of the plane defined by the aromatic ring. The values of alpha are in the range 1.5 +/- 0.1 degrees to 1.7 +/- 0.1 degrees for the temperature range 295-330 K
Learning to Rank Images from Eye Movements
No full textCombining multiple information sources can improve the accuracy of search in information retrieval. This paper presents a new image search strategy which combines image features together with implicit feedback from users' eye movements, using them to rank images. In order to better deal with larger data sets, we present a perceptron formulation of the Ranking Support Vector Machine algorithm. We present initial results on inferring the rank of images presented in a page based on simple image features and implicit feedback of users. The results show that the perceptron algorithm improves the results, and that fusing eye movements and image histograms gives better rankings to images than either of these features alone
Impact of Vitamin D Supplementation on Arterial Vasomotion, Stiffness and Endothelial Biomarkers in Chronic Kidney Disease Patients
No full textBackground: Cardiovascular events are frequent and vascular endothelial function is abnormal in patients with chronic
kidney disease (CKD). We demonstrated endothelial dysfunction with vitamin D deficiency in CKD patients; however the impact of cholecalciferol supplementation on vascular stiffness and vasomotor function, endothelial and bone biomarkers in CKD patients with low 25-hydroxy vitamin D [25(OH)D] is unknown, which this study investigated.
Methods: We assessed non-diabetic patients with CKD stage 3/4, age 17–80 years and serum 25(OH)D ,75 nmol/L. Brachial
artery Flow Mediated Dilation (FMD), Pulse Wave Velocity (PWV), Augmentation Index (AI) and circulating blood biomarkers were evaluated at baseline and at 16 weeks. Oral 300,000 units cholecalciferol was administered at baseline and 8-weeks.
Results: Clinical characteristics of 26 patients were: age 50614 (mean61SD) years, eGFR 41611 ml/min/1.73 m2, males
73%, dyslipidaemia 36%, smokers 23% and hypertensives 87%. At 16-week serum 25(OH)D and calcium increased (43616
to 84629 nmol/L, p,0.001 and 2.3760.09 to 2.4260.09 mmol/L; p = 0.004, respectively) and parathyroid hormone
decreased (10.868.6 to 7.464.4; p = 0.001). FMD improved from 3.163.3% to 6.163.7%, p = 0.001. Endothelial biomarker
concentrations decreased: E-Selectin from 566662123 to 525662058 pg/mL; p = 0.032, ICAM-1, 3.4560.01 to
3.1061.04 ng/mL; p = 0.038 and VCAM-1, 54633 to 42633 ng/mL; p = 0.006. eGFR, BP, PWV, AI, hsCRP, von Willebrand
factor and Fibroblast Growth Factor-23, remained unchanged.
Conclusion: This study demonstrates for the first time improvement of endothelial vasomotor and secretory functions with vitamin D in CKD patients without significant adverse effects on arterial stiffness, serum calcium or FGF-23.
Trial Registration: ClinicalTrials.gov NCT0200571
MeSH term explosion and author rank improve expert recommendations
No full textInformation overload is an often-cited phenomenon that reduces the productivity, efficiency and efficacy of scientists. One challenge for scientists is to find appropriate collaborators in their research. The literature describes various solutions to the problem of expertise location, but most current approaches do not appear to be very suitable for expert recommendations in biomedical research. In this study, we present the development and initial evaluation of a vector space model-based algorithm to calculate researcher similarity using four inputs: 1) MeSH terms of publications; 2) MeSH terms and author rank; 3) exploded MeSH terms; and 4) exploded MeSH terms and author rank. We developed and evaluated the algorithm using a data set of 17,525 authors and their 22,542 papers. On average, our algorithms correctly predicted 2.5 of the top 5/10 coauthors of individual scientists. Exploded MeSH and author rank outperformed all other algorithms in accuracy, followed closely by MeSH and author rank. Our results show that the accuracy of MeSH term-based matching can be enhanced with other metadata such as author rank
A Connectionist and Multivariate Approach to Science Maps: Som, Clustering and Mds Applied to Library & Information Science Research.
No full textThe visualization of scientific field structures is a classic of scientometric studies. This paper presents a domain analysis of the library and information science discipline based on author co-citation analysis (ACA) and journal cocitation analysis (JCA). The techniques used for map construction are the self-organizing map (SOM) neural
algorithm, Ward’s clustering method and multidimensional
scaling (MDS). The results of this study are compared with
similar research developed by Howard White and Katherine
McCain [1]. The methodologies used allow us to confirm that
the subject domains identified in this paper are, as well,
present in our study for the corresponding period. The appearance of studies pertaining to library science reveals
the relationship of this realm with information science.
Especially significant is the presence of the management on the journal maps. From a methodological standpoint, meanwhile, we would agree with those authors who consider
MDS, the SOM and clustering as complementary methods
that provide representations of the same reality from different analytical points of view. Even so, the MDS representation is the one offering greater possibilities for the structural representation of the clusters in a set of variables
Predicting relevance of parts of an image
No full textThis report studies the task of inferring which parts of an image are relevant for the user viewing the image. The relevance is inferred from gaze trajectory of users viewing the images given a specific task. Novel computational models based on both Bayesian generative modeling and kernel methods are developed for inferring the regions of interest from raw fixation data, as well as from combination of eye movements and image content features
A connectionist and multivariate approach to science maps: the SOM, clustering and MDS applied to library and information science research
No full textThe visualization of scientific field structures is a classic of scientometric studies. This paper presents a domain analysis of the library and information science discipline based on author co-citation analysis (ACA) and journal cocitation analysis (JCA). The techniques used for map construction are the self-organizing map (SOM) neural algorithm, Ward’s clustering method and multidimensional scaling (MDS). The results of this study are compared with similar research developed by Howard White and Katherine McCain [1]. The methodologies used allow us to confirm that the subject domains identified in this paper are, as well,
present in our study for the corresponding period. The appearance of studies pertaining to library science reveals the relationship of this realm with information science. Especially significant is the presence of the management on the journal maps. From a methodological standpoint, meanwhile, we would agree with those authors who consider MDS, the SOM and clustering as complementary methods that provide representations of the same reality from different analytical points of view. Even so, the MDS representation is the one offering greater possibilities for the structural representation of the clusters in a set of variables
Going Beyond Counting First Authors in Author Co-citation Analysis
No full textThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
- …
