Showing 1–16 of 16 results for author: Popescu, V

Hands-Free VR

Authors: Jorge Askur Vazquez Fernandez, Jae Joong Lee, Santiago Andrés Serrano Vacca, Alejandra Magana, Bedrich Benes, Voicu Popescu

Abstract: The paper introduces Hands-Free VR, a voice-based natural-language interface for VR. The user gives a command using their voice, the speech audio data is converted to text using a speech-to-text deep learning model that is fine-tuned for robustness to word phonetic similarity and to spoken English accents, and the text is mapped to an executable VR command using a large language model that is robu… ▽ More The paper introduces Hands-Free VR, a voice-based natural-language interface for VR. The user gives a command using their voice, the speech audio data is converted to text using a speech-to-text deep learning model that is fine-tuned for robustness to word phonetic similarity and to spoken English accents, and the text is mapped to an executable VR command using a large language model that is robust to natural language diversity. Hands-Free VR was evaluated in a controlled within-subjects study (N = 22) that asked participants to find specific objects and to place them in various configurations. In the control condition participants used a conventional VR user interface to grab, carry, and position the objects using the handheld controllers. In the experimental condition participants used Hands-Free VR. The results confirm that: (1) Hands-Free VR is robust to spoken English accents, as for 20 of our participants English was not their first language, and to word phonetic similarity, correctly transcribing the voice command 96.71% of the time; (2) Hands-Free VR is robust to natural language diversity, correctly map** the transcribed command to an executable command in 97.83% of the time; (3) Hands-Free VR had a significant efficiency advantage over the conventional VR interface in terms of task completion time, total viewpoint translation, total view direction rotation, and total left and right hand translations; (4) Hands-Free VR received high user preference ratings in terms of ease of use, intuitiveness, ergonomics, reliability, and desirability. △ Less

Submitted 22 February, 2024; originally announced February 2024.

An Evaluation of OCR on Egocentric Data

Authors: Valentin Popescu, Dima Damen, Toby Perrett

Abstract: In this paper, we evaluate state-of-the-art OCR methods on Egocentric data. We annotate text in EPIC-KITCHENS images, and demonstrate that existing OCR methods struggle with rotated text, which is frequently observed on objects being handled. We introduce a simple rotate-and-merge procedure which can be applied to pre-trained OCR models that halves the normalized edit distance error. This suggests… ▽ More In this paper, we evaluate state-of-the-art OCR methods on Egocentric data. We annotate text in EPIC-KITCHENS images, and demonstrate that existing OCR methods struggle with rotated text, which is frequently observed on objects being handled. We introduce a simple rotate-and-merge procedure which can be applied to pre-trained OCR models that halves the normalized edit distance error. This suggests that future OCR attempts should incorporate rotation into model design and training procedures. △ Less

Submitted 11 June, 2022; originally announced June 2022.

Comments: Extended Abstract, EPIC workshop at CVPR 22

Network analytics for drug repurposing in COVID-19

Authors: Nicoleta Siminea, Victor Popescu, Jose Angel Sanchez Martin, Daniela Florea, Georgiana Gavril, Ana-Maria Gheorghe, Corina Itcus, Krishna Kanhaiya, Octavian Pacioglu, Laura Ioana Popa, Romica Trandafir, Maria Iris Tusa, Manuela Sidoroff, Mihaela Paun, Eugen Czeizler, Andrei Paun, Ion Petre

Abstract: To better understand the potential of drug repurposing in COVID-19, we analyzed control strategies over essential host factors for SARS-CoV-2 infection. We constructed comprehensive directed protein-protein interaction networks integrating the top ranked host factors, drug target proteins, and directed protein-protein interaction data. We analyzed the networks to identify drug targets and combinat… ▽ More To better understand the potential of drug repurposing in COVID-19, we analyzed control strategies over essential host factors for SARS-CoV-2 infection. We constructed comprehensive directed protein-protein interaction networks integrating the top ranked host factors, drug target proteins, and directed protein-protein interaction data. We analyzed the networks to identify drug targets and combinations thereof that offer efficient control over the host factors. We validated our findings against clinical studies data and bioinformatics studies. Our method offers a new insight into the molecular details of the disease and into potentially new therapy targets for it. Our approach for drug repurposing is significant beyond COVID-19 and may be applied also to other diseases. △ Less

Submitted 23 June, 2021; originally announced June 2021.

Comments: 21 pages, 3 tables, 9 figures, supplementary information included at the end, 4 files as supplementary material

Network modeling methods for precision medicine

Authors: Elio Nushi, Victor-Bogdan Popescu, Jose Angel Sanchez Martin, Sergiu Ivanov, Eugen Czeizler, Ion Petre

Abstract: We discuss in this survey several network modeling methods and their applicability to precision medicine. We review several network centrality methods (degree centrality, closeness centrality, eccentricity centrality, betweenness centrality, and eigenvector-based prestige) and two systems controllability methods (minimum dominating sets and network structural controllability). We demonstrate their… ▽ More We discuss in this survey several network modeling methods and their applicability to precision medicine. We review several network centrality methods (degree centrality, closeness centrality, eccentricity centrality, betweenness centrality, and eigenvector-based prestige) and two systems controllability methods (minimum dominating sets and network structural controllability). We demonstrate their applicability to precision medicine on three multiple myeloma patient disease networks. Each network consists of protein-protein interactions built around a specific patient's mutated genes, around the targets of the drugs used in the standard of care in multiple myeloma, and around multiple myeloma-specific essential genes. For each network we demonstrate how the network methods we discuss can be used to identify personalized, targeted drug combinations uniquely suited to that patient. △ Less

Submitted 19 April, 2021; originally announced April 2021.

Comments: 55 pages, 3 figures

Representation range needs for 16-bit neural network training

Authors: Valentina Popescu, Abhinav Venigalla, Di Wu, Robert Schreiber

Abstract: Deep learning has grown rapidly thanks to its state-of-the-art performance across a wide range of real-world applications. While neural networks have been trained using IEEE-754 binary32 arithmetic, the rapid growth of computational demands in deep learning has boosted interest in faster, low precision training. Mixed-precision training that combines IEEE-754 binary16 with IEEE-754 binary32 has be… ▽ More Deep learning has grown rapidly thanks to its state-of-the-art performance across a wide range of real-world applications. While neural networks have been trained using IEEE-754 binary32 arithmetic, the rapid growth of computational demands in deep learning has boosted interest in faster, low precision training. Mixed-precision training that combines IEEE-754 binary16 with IEEE-754 binary32 has been tried, and other $16$-bit formats, for example Google's bfloat16, have become popular. In floating-point arithmetic there is a tradeoff between precision and representation range as the number of exponent bits changes; denormal numbers extend the representation range. This raises questions of how much exponent range is needed, of whether there is a format between binary16 (5 exponent bits) and bfloat16 (8 exponent bits) that works better than either of them, and whether or not denormals are necessary. In the current paper we study the need for denormal numbers for mixed-precision training, and we propose a 1/6/9 format, i.e., 6-bit exponent and 9-bit explicit mantissa, that offers a better range-precision tradeoff. We show that 1/6/9 mixed-precision training is able to speed up training on hardware that incurs a performance slowdown on denormal operations or eliminates the need for denormal numbers altogether. And, for a number of fully connected and convolutional neural networks in computer vision and natural language processing, 1/6/9 achieves numerical parity to standard mixed-precision. △ Less

Submitted 6 April, 2021; v1 submitted 29 March, 2021; originally announced March 2021.

Comments: 7 pages, 3 figures, 4 tables; conference

Identifying efficient controls of complex interaction networks using genetic algorithms

Authors: Victor-Bogdan Popescu, Krishna Kanhaiya, Iulian Năstac, Eugen Czeizler, Ion Petre

Abstract: Control theory has seen recently impactful applications in network science, especially in connections with applications in network medicine. A key topic of research is that of finding minimal external interventions that offer control over the dynamics of a given network, a problem known as network controllability. We propose in this article a new solution for this problem based on genetic algorith… ▽ More Control theory has seen recently impactful applications in network science, especially in connections with applications in network medicine. A key topic of research is that of finding minimal external interventions that offer control over the dynamics of a given network, a problem known as network controllability. We propose in this article a new solution for this problem based on genetic algorithms. We tailor our solution for applications in computational drug repurposing, seeking to maximise its use of FDA-approved drug targets in a given disease-specific protein-protein interaction network. We show how our algorithm identifies a number of potentially efficient drugs for breast, ovarian, and pancreatic cancer. We demonstrate our algorithm on several benchmark networks from cancer medicine, social networks, electronic circuits, and several random networks with their edges distributed according to the Erdős-Rényi, the small-world, and the scale-free properties. Overall, we show that our new algorithm is more efficient in identifying relevant drug targets in a disease network, advancing the computational solutions needed for new therapeutic and drug repurposing approaches. △ Less

Submitted 9 July, 2020; originally announced July 2020.

Comments: The submission contains 34 pages, 9 figures and 6 tables

Ab initio typical medium theory of substitutional disorder

Authors: A. Östlin, Y. Zhang, H. Terletska, F. Beiuseanu, V. Popescu, K. Byczuk, L. Vitos, M. Jarrell, D. Vollhardt, L. Chioncel

Abstract: By merging single-site typical medium theory with density functional theory we introduce a self-consistent framework for electronic structure calculations of materials with substitutional disorder which takes into account Anderson localization. The scheme and details of the implementation are presented and applied to the hypothetical alloy Li$_{c}$Be$_{1-c}$, and the results are compared with thos… ▽ More By merging single-site typical medium theory with density functional theory we introduce a self-consistent framework for electronic structure calculations of materials with substitutional disorder which takes into account Anderson localization. The scheme and details of the implementation are presented and applied to the hypothetical alloy Li$_{c}$Be$_{1-c}$, and the results are compared with those obtained with the coherent potential approximation. Furthermore we demonstrate that Anderson localization suppresses ferromagnetic order for a very low concentration of (i) carbon impurities substituting oxygen in MgO$_{1-c}$C$_{c}$, and (ii) manganese impurities substituting magnesium in Mg$_{1-c}$Mn$_c$O for the low-spin magnetic configuration. △ Less

Submitted 30 January, 2020; v1 submitted 6 November, 2019; originally announced November 2019.

Comments: Minor revisions

Journal ref: Phys. Rev. B 101, 014210 (2020)

Native defects in the Co$_2$Ti$Z$ ($Z=$ Si, Ge, Sn) full Heusler alloys: formation and influence on the thermoelectric properties

Authors: Voicu Popescu, Peter Kratzer, Sebastian Wimmer, Hubert Ebert

Abstract: We have performed first-principles investigations on the native defects in the full Heusler alloys Co$_2$Ti$Z$ ($Z$ one of the group IV elements Si, Ge, Sn), determining their formation energies and how they influence the transport properties. We find that Co vacancies (Vc) in all compounds and the Ti$_\text{Sn}$ anti-site exhibit negative formation energies. The smallest positive values occur for… ▽ More We have performed first-principles investigations on the native defects in the full Heusler alloys Co$_2$Ti$Z$ ($Z$ one of the group IV elements Si, Ge, Sn), determining their formation energies and how they influence the transport properties. We find that Co vacancies (Vc) in all compounds and the Ti$_\text{Sn}$ anti-site exhibit negative formation energies. The smallest positive values occur for Co in excess on anti-sites (Co$_Z$ or Co$_\text{Ti}$) and for Ti$_Z$. The most abundant native defects were modeled as dilute alloys, treated with the coherent potential approximation in combination with the multiple-scattering theory Green function approach. The self-consistent potentials determined this way were used to calculate the residual resistivity via the Kubo-Greenwood formula and, based on its energy dependence, the Seebeck coefficient of the systems. The latter is shown to depend significantly on the type of defect, leading to variations that are related to subtle, spin-orbit coupling induced, changes in the electronic structure above the half-metallic gap. Two of the systems, Vc$_\text{Co}$ and Co$_Z$, are found to exhibit a negative Seebeck coefficient. This observation, together with their low formation energy, offers an explanation for the experimentally observed negative Seebeck coefficient of the Co$_2$Ti$Z$ compounds as being due to unintentionally created native defects. △ Less

Submitted 5 July, 2017; v1 submitted 1 April, 2017; originally announced April 2017.

Journal ref: Phys. Rev. B 96, 054443 (2017)

Large morphological sensitivity of the magnetothermopower in Co/Cu multilayered systems

Authors: Voicu Popescu, Peter Kratzer

Abstract: We present results of first-principles calculations on the transport properties, both under an electric field or a temperature gradient, in the Co/Cu multilayered systems. The various effects brought about by the changes in the morphological parameters, such as the number of repeats and the layer thickness, are discussed in a systematic way. Our calculations show that the Seebeck coefficient and t… ▽ More We present results of first-principles calculations on the transport properties, both under an electric field or a temperature gradient, in the Co/Cu multilayered systems. The various effects brought about by the changes in the morphological parameters, such as the number of repeats and the layer thickness, are discussed in a systematic way. Our calculations show that the Seebeck coefficient and the magnetothermopower (MTP) converge rather rapidly with the number of Co repeats. In the range of thin Co layers, we find strong variations in amplitude and sign of both the Seebeck coefficient and the MTP. These large variations, which have no correspondent in the (magneto)conductance, are shown to be the result of quantum well states present in the minority spin channel of thin Co layers. △ Less

Submitted 5 December, 2014; originally announced December 2014.

Interplay of growth mode and thermally induced spin accumulation in epitaxial Al/Co$_2$TiSi/Al and Al/Co$_2$TiGe/Al contacts

Authors: Benjamin Geisler, Peter Kratzer, Voicu Popescu

Abstract: The feasibility of thermally driven spin injectors built from half-metallic Heusler alloys inserted between aluminum leads was investigated by means of {\em ab initio} calculations of the thermodynamic equilibrium and electronic transport. We have focused on two main issues and found that: (i) the interface between Al and the closely lattice-matched Heusler alloys of type Co$_2$Ti$Z$ ($Z=$ Si or G… ▽ More The feasibility of thermally driven spin injectors built from half-metallic Heusler alloys inserted between aluminum leads was investigated by means of {\em ab initio} calculations of the thermodynamic equilibrium and electronic transport. We have focused on two main issues and found that: (i) the interface between Al and the closely lattice-matched Heusler alloys of type Co$_2$Ti$Z$ ($Z=$ Si or Ge) is stable under various growth conditions; and (ii) the conventional and spin-dependent Seebeck coefficients in such heterojunctions exhibit a strong dependence on both the spacer and the atomic composition of the Al/Heusler interface. The latter quantity gives a measure of the spin accumulation and varies between $+8$~$μ$V/K and $-3$~$μ$V/K near $300$~K, depending on whether a Ti-Ge or a Co-Co plane makes the contact between Al and Co$_2$TiGe in the trilayer. Our results show that it is in principle possible to tailor the spin-caloric effects by a targeted growth control of the samples. △ Less

Submitted 15 May, 2014; v1 submitted 18 February, 2014; originally announced February 2014.

Comments: 16 pages, 13 figures

ATC Taskload Inherent to the Geometry of Stochastic 4-D Trajectory Flows with Flight Technical Errors

Authors: Vlad Popescu, John-Paul B. Clarke, Karen M. Feigh, Eric Feron

Abstract: A method to quantify the probabilistic controller taskload inherent to maintaining aircraft adherence to 4-D trajectories within flow corridors is presented. An Ornstein-Uhlenbeck model of the aircraft motion and a Poisson model of the flow scheduling are introduced along with reasonable numerical values of the model parameters. Analytic expressions are derived for the taskload probability densit… ▽ More A method to quantify the probabilistic controller taskload inherent to maintaining aircraft adherence to 4-D trajectories within flow corridors is presented. An Ornstein-Uhlenbeck model of the aircraft motion and a Poisson model of the flow scheduling are introduced along with reasonable numerical values of the model parameters. Analytic expressions are derived for the taskload probability density functions for basic functional elements of the flow structure. Monte Carlo simulations are performed for these basic functional elements and the controller taskload probabilities are exhibited. △ Less

Submitted 7 February, 2011; originally announced February 2011.

Comments: Submitted to the 9th USA/Europe ATM R&D Seminar - Berlin 2011

The effect of the spin-orbit interaction on the band gap of half-metals

Authors: Ph. Mavropoulos, K. Sato, R. Zeller, P. H. Dederichs, V. Popescu, H. Ebert

Abstract: The spin-orbit interaction can cause a nonvanishing density of states (DOS) within the minority-spin band gap of half-metals around the Fermi level. We examine the magnitude of the effect in Heusler alloys, zinc-blende half metals and diluted magnetic semiconductors, using first-principles calculations. We find that the ratio of spin-down to spin-up DOS at the Fermi level can range from below 1%… ▽ More The spin-orbit interaction can cause a nonvanishing density of states (DOS) within the minority-spin band gap of half-metals around the Fermi level. We examine the magnitude of the effect in Heusler alloys, zinc-blende half metals and diluted magnetic semiconductors, using first-principles calculations. We find that the ratio of spin-down to spin-up DOS at the Fermi level can range from below 1% (e.g. 0.5% for NiMnSb) over several percents (4.2% for (Ga,Mn)As) to 13% for MnBi. △ Less

Submitted 7 August, 2003; originally announced August 2003.

Comments: 5 pages, 3 figures

Journal ref: Phys. Rev. B 69, 054424 (2004)

Analysis of a Fossil Bone from the Archaeological Settlement Malu Rosu, Romania by Accelerator Mass Spectrometry

Authors: Agata Olariu, Ion V. Popescu, Ragnar Hellborg, Kristina Stenström, Mikko Faarinen, Per Persson, Bengt Erlandsson, Göran Skog, Emilian Alexandrescu

Abstract: A fossil bone from the archaeological site Malu Rosu Giurgiu, in Romania has been analyzed by accelerator mass spectrometry to estimate its age by determining its $^{14}$C content. The radiocarbon age of the bone is in agreement with the date obtained by the method for age determination, based on fluorine content. This is the first radiocarbon dating for the final Neolithic period, for this arch… ▽ More A fossil bone from the archaeological site Malu Rosu Giurgiu, in Romania has been analyzed by accelerator mass spectrometry to estimate its age by determining its $^{14}$C content. The radiocarbon age of the bone is in agreement with the date obtained by the method for age determination, based on fluorine content. This is the first radiocarbon dating for the final Neolithic period, for this archaeological settlement in the Romanian region. △ Less

Submitted 13 March, 2001; originally announced March 2001.

Comments: latex manuscript, 4 pdf figures

Trace elements analysis of aerosol samples from some Romanian urban zones

Authors: I. V. Popescu, T. Badica, Agata Olariu, Olga Guguianu

Abstract: Aerosols deposits on filters from ten Romanian towns with different kinds and levels of industrial development have been studied. The concentration of elements with Z~16 have been measured by particle - induced X-ray emission (PIXE) analysis. It has been determined 15 elements: S, K, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Hg and Pb. Aerosols deposits on filters from ten Romanian towns with different kinds and levels of industrial development have been studied. The concentration of elements with Z~16 have been measured by particle - induced X-ray emission (PIXE) analysis. It has been determined 15 elements: S, K, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Hg and Pb. △ Less

Submitted 19 August, 1999; originally announced August 1999.

Comments: Latex manuscript, 7 pages, 1 eps figure

Trace Analysis of Ancient Gold Objects Using Radiochemical Neutron Activation

Authors: Agata Olariu, Mioara Constantinescu, O. Constantinescu, T. Badica, I. V. Popescu, C. Besliu, Doina Leahu

Abstract: Radiochemical neutron activation analysis has been applied to investigate the microelements in gold samples with archaeological importance. Chemical separation has allowed the determination of traces of Ir, Os, Sb, Zn, Co, Fe, Ni. Instrumental neutron activation analysis has been used for the determination of Cu. Radiochemical neutron activation analysis has been applied to investigate the microelements in gold samples with archaeological importance. Chemical separation has allowed the determination of traces of Ir, Os, Sb, Zn, Co, Fe, Ni. Instrumental neutron activation analysis has been used for the determination of Cu. △ Less

Submitted 19 August, 1999; originally announced August 1999.

Comments: Latex manuscript, 16 pages, 3 eps figures

Journal ref: J. Rad. Nucl. Chem. 240 (1999) 261-267

Compositional Studies of Ancient Copper from Romanian Territories

Authors: Agata Olariu, C. Besliu, M. Belc, I. V. Popescu, T. Badica, Gh. Lazarovici

Abstract: Ancient copper objects from Romanian Territories have been analyzed by neutron activation analysis. A series of elements is determined: Au, Ag, As, Co, Cr, Fe, Hg, Ni, Zn, Sb, Sc, Se, Sn. Using mathematical dendograms some classifications and correlation have been established. Ancient copper objects from Romanian Territories have been analyzed by neutron activation analysis. A series of elements is determined: Au, Ag, As, Co, Cr, Fe, Hg, Ni, Zn, Sb, Sc, Se, Sn. Using mathematical dendograms some classifications and correlation have been established. △ Less

Submitted 21 July, 1999; originally announced July 1999.

Comments: Latex manuscript, 14 pages, 5 eps figures