Naresuan University Journal
Not a member yet
39887 research outputs found
Sort by
Gradient-descent-based learning in memristive crossbar arrays
No full textThis paper describes techniques to implement gradient-descent-based machine learning algorithms on crossbar arrays made of memristors or other analog memory devices. We introduce the Unregulated Step Descent (USD) algorithm, which is an approximation of the steepest descent algorithm, and discuss how it addresses various hardware implementation issues. We discuss the effect of device parameters and their variability on performance of the algorithm by using artificially generated and real-world datasets. In addition to providing insights on the effect of device parameters on learning, we illustrate how the USD algorithm partially offsets the effect of device variability. Finally, we discuss how the USD algorithm can be implemented in crossbar arrays using a simple 4-phase training scheme. The method allows parallel update of crossbar memory elements and reduces the hardware cost and complexity of the training architecture significantly
Resin-based composites show similar kinetic profiles for dimensional change and recovery with solvent storage.
No full textObjectives. To investigate the sorption, solubility, mass change and hygroscopic expansion (solvent swelling) of resin-composites after long term storage in different solvents. Methods. Eight materials were studied: two bulk-fill flowable materials (SDR and Venus bulk fill, V-BF), a packable bulk-fill material (Tetric Evoceram bulk-fill, TET-BF), a fiber reinforced material (Ever X posterior, EVX), a nano-hybrid conventional material (Tetric Evoceram, TET) and micro-hybrid conventional materials (G-aenial anterior, GA-P and posterior, GAA). Three groups of disk shaped specimens were prepared using split stainless steel molds. Each group was stored, respectively, in: water, 75% ethanol/water and methyl ethyl ketone (MEK). The total storage time was 180 d plus a reconditioning time of 120 d. A non-contact laser scanning micrometer was used to measure the diametral changes. Results. Significant differences were found in the sorption and solubility of the materials.Generally, MEK stored specimens had the highest values followed by 75% ethanol/water then water. A similar trend was found with the mass and volume changes (except for EVX). V-BF showed the highest sorption (98.1g/mm3) and solubility (10g/mm3) after MEK storage. Mass and volume changes showed near-linear correlation, with high Pearson coefficients (0.86–0.99). Significance. Generally the materials were most greatly affected by MEK storage compared to the other two solvents. The glass-fiber-reinforced EVX, however, was most affected by water immersion. The pattern of change/recovery behavior of the materials, during solvent challenge, was similar to the pattern of viscoelastic creep/recovery behavior of resin-composite materials
We are Imazighen: The Development of Algerian Berber Identity in Twentieth-Century Literature and Culture
No full textChronic mould exposure as a risk factor for severe community acquired pneumonia in a patient requiring extra corporeal membrane oxygenation.
No full textA previously fit and well man developed acute respiratory failure due to environmental mould exposure from living in damp rental accommodation. Despite aggressive intensive care management he rapidly deteriorated and required respiratory and cardiac Extracorporeal Membrane Oxygenation. We hypothesize that poor domiciliary conditions may make an underestimated contribution to community respiratory disease. These conditions may present as acute and severe illness with non-typical pathogens identified
XCT Image-based homogenisation of the elastic properties of ultra high performance fibre reinforced concrete
No full textMicro-scale X-ray computed tomography images of UHPFRC are used to construct finite element models for prediction of elastic properties using a two-step homogenization approach. In the first step, the mortar matrix with a large number of small voids is homogenized, and in the second the fewer large voids and steel fibres are modelled with the homogenised matrix. This leads to significant reduction in computational costs over one-step approaches. The results are compared favourably with experimental data. The effects of RVE size are also studied
Revisiting a framework for researching multilingually:Contributions from critical theory, intercultural relations, ethics, and the creative arts
No full textHirshfeld surface investigation of structure-directing interactions within dipicolinic acid derivatives
No full text
Meta-analysis of absolute mean differences from randomised trials with treatment-related clustering associated with care providers
No full textNesting of patients within care providers in trials of physical and talking therapies creates an additional level within the design. The statistical implications of this are analogous to those of cluster randomised trials, except that the clustering effect may interact with treatment and can be restricted to one or more of the arms. The statistical model that is recommended at the trial level includes a random effect for the care provider but allows the provider and patient level variances to differ across arms. Evidence suggests that, while potentially important, such within-trial clustering effects have rarely been taken into account in trials and do not appear to have been considered in meta-analyses of these trials. This paper describes summary measures and individual-patient-data methods for meta-analysing absolute mean differences from randomised trials with two-level nested clustering effects, contrasting fixed and random effects meta-analysis models. It extends methods for incorporating trials with unequal variances and homogeneous clustering to allow for between-arm and between-trial heterogeneity in intra-class correlation coefficient estimates. The work is motivated by a meta-analysis of trials of counselling in primary care, where the control is no counselling and the outcome is the Beck Depression Inventory. Assuming equal counsellor intra-class correlation coefficients across trials, the recommended random-effects heteroscedastic model gave a pooled absolute mean difference of -2.53 (95% CI -5.33 to 0.27) using summary measures and -2.51 (95% CI -5.35 to 0.33) with the individual-patient-data. Pooled estimates were consistently below a minimally important clinical difference of four to five points on the Beck Depression Inventory