Showing 1–14 of 14 results for author: Shetty, N

GANs Settle Scores!

Authors: Siddarth Asokan, Nishanth Shetty, Aadithya Srikanth, Chandra Sekhar Seelamantula

Abstract: Generative adversarial networks (GANs) comprise a generator, trained to learn the underlying distribution of the desired data, and a discriminator, trained to distinguish real samples from those output by the generator. A majority of GAN literature focuses on understanding the optimality of the discriminator through integral probability metric (IPM) or divergence based analysis. In this paper, we… ▽ More Generative adversarial networks (GANs) comprise a generator, trained to learn the underlying distribution of the desired data, and a discriminator, trained to distinguish real samples from those output by the generator. A majority of GAN literature focuses on understanding the optimality of the discriminator through integral probability metric (IPM) or divergence based analysis. In this paper, we propose a unified approach to analyzing the generator optimization through variational approach. In $f$-divergence-minimizing GANs, we show that the optimal generator is the one that matches the score of its output distribution with that of the data distribution, while in IPM GANs, we show that this optimal generator matches score-like functions, involving the flow-field of the kernel associated with a chosen IPM constraint space. Further, the IPM-GAN optimization can be seen as one of smoothed score-matching, where the scores of the data and the generator distributions are convolved with the kernel associated with the constraint. The proposed approach serves to unify score-based training and existing GAN flavors, leveraging results from normalizing flows, while also providing explanations for empirical phenomena such as the stability of non-saturating GAN losses. Based on these results, we propose novel alternatives to $f$-GAN and IPM-GAN training based on score and flow matching, and discriminator-guided Langevin sampling. △ Less

Submitted 2 June, 2023; originally announced June 2023.

Scalable fabrication of edge contacts to 2D materials

Authors: Naveen Shetty, Hans He, Richa Mitra, Johanna Huhtasaari, Konstantina Iordanau, Julia Wiktor, Sergey Kubatkin, Saroj Dash, Rositsa Yakimova, Lunjie Zeng, Eva Olsson, Samuel Lara-Avila

Abstract: We present a fabrication method for reliably and reproducibly forming electrical contacts to 2D materials, based on the tri-layer resist system. We demonstrate the applicability of this method for epitaxial graphene on silicon carbide (epigraphene) and the transition metal dichalcogenides (TMDCs) molybdenum disulfide ($MoS_2$). For epigraphene, the specific contact resistances are of the order of… ▽ More We present a fabrication method for reliably and reproducibly forming electrical contacts to 2D materials, based on the tri-layer resist system. We demonstrate the applicability of this method for epitaxial graphene on silicon carbide (epigraphene) and the transition metal dichalcogenides (TMDCs) molybdenum disulfide ($MoS_2$). For epigraphene, the specific contact resistances are of the order of $ρ_c$ ~ $50$ $Ωμm$, and follow the Landauer quantum limit, $ρ_c \propto n^{-1/2}$, with $n$ being the carrier density of graphene. For $MoS_2$ flakes, our edge contacts enable field effect transistors (FET) with ON/OFF ratio of $> 10^6$ at room temperature ( $> 10^9$ at cryogenic temperatures). The fabrication route here demonstrated allows for contact metallization using thermal evaporation and also by sputtering, giving an additional flexibility when designing electrical interfaces, which is key in practical devices and when exploring the electrical properties of emerging materials. △ Less

Submitted 6 March, 2023; v1 submitted 8 June, 2022; originally announced June 2022.

Highly efficient UV detection in a metal-semiconductor-metal detector with epigraphene

Authors: Hans He, Naveen Shetty, Sergey Kubatkin, Pascal Stadler, Tomas Löfwander, Mikael Fogelström, José Carlos Miranda-Valenzuela, Rositsa Yakimova, Thilo Bauch, Samuel Lara-Avila

Abstract: We show that epitaxial graphene on silicon carbide (epigraphene) grown at high temperatures (T > 1850 °C) readily acts as material for implementing solar-blind ultraviolet (UV) detectors with outstanding performance. We present centimeter-sized epigraphene metal-semiconductor-metal (MSM) detectors with peak external quantum efficiency of ~ 85% for wavelengths 250-280 nm, corresponding to nearly 10… ▽ More We show that epitaxial graphene on silicon carbide (epigraphene) grown at high temperatures (T > 1850 °C) readily acts as material for implementing solar-blind ultraviolet (UV) detectors with outstanding performance. We present centimeter-sized epigraphene metal-semiconductor-metal (MSM) detectors with peak external quantum efficiency of ~ 85% for wavelengths 250-280 nm, corresponding to nearly 100% internal quantum efficiency when accounting for reflection losses. Zero bias operation is possible in asymmetric devices, with the responsivity to UV remaining as high as R = 134 mA/W, making this a self-powered detector. The low dark currents Io ~50 fA translate into an estimated record high specific detectivity D = 3.5 x 10^15 Jones. The performance that we demonstrate, together with material reproducibility, renders epigraphene technologically attractive to implement high-performance planar MSM devices with a low processing effort, including multi-pixel UV sensor arrays, suitable for a number of practical applications. △ Less

Submitted 3 March, 2022; originally announced March 2022.

Exceptionally accurate large graphene quantum Hall arrays for the new SI

Authors: Hans He, Karin Cedergren, Naveen Shetty, Samuel Lara-Avila, Sergey Kubatkin, Tobias Bergsten, Gunnar Eklund

Abstract: The quantum Hall effect (QHE) is a cornerstone in the new International System of Units (SI), wherein the base units are derived from seven fundamental constants such as Planck's constant h and elementary charge e. Graphene has revolutionized practical resistance metrology by enabling the realization of quantized resistance h/2e^2 = 12.9... kOhm under relaxed experimental conditions. Looking ahead… ▽ More The quantum Hall effect (QHE) is a cornerstone in the new International System of Units (SI), wherein the base units are derived from seven fundamental constants such as Planck's constant h and elementary charge e. Graphene has revolutionized practical resistance metrology by enabling the realization of quantized resistance h/2e^2 = 12.9... kOhm under relaxed experimental conditions. Looking ahead, graphene also has the potential to improve realizations of the electronic kilogram using the Kibble balance, and the quantum Ampere in wide current ranges. However, these prospects require different resistance values than practically achievable in single QHE devices, while also imposing stringent demands on energy dissipation in single QHE devices, ultimately requiring currents almost two orders of magnitude higher than the typical QHE breakdown currents IC ~ 100 uA achievable in graphene. Here we present unprecedented accuracy in the quantization of a record sized quantum Hall array (QHA), demonstrating RK/236 ~ 109 Ohm with 0.2 part-per-billion (nOhm/Ohm) accuracy with IC over 5 mA (~ 1 nOhm/Ohm accuracy for IC = 8.5 mA), using epitaxial graphene on silicon carbide (epigraphene). The array quantization accuracy, comparable to the most precise universality tests of QHE in single Hall bar devices, together with the scalability and reliability of this approach pave the road for superior realizations of three key units in the modern SI: the ohm, the ampere, and the kilogram. △ Less

Submitted 1 December, 2022; v1 submitted 16 November, 2021; originally announced November 2021.

Comments: Main text 12 pages, 4 figures.Supplementary information 4 pages, 4 figures

Journal ref: Nat Commun 13, 6933 (2022)

Electron-phonon coupling of epigraphene at millikelvin temperatures

Authors: Bayan Karimi, Hans He, Yu-Cheng Chang, Libin Wang, Jukka P. Pekola, Rositsa Yakimova, Naveen Shetty, Joonas T. Peltonen, Samuel Lara-Avila, Sergey Kubatkin

Abstract: We investigate the basic charge and heat transport properties of charge neutral epigraphene at sub-kelvin temperatures, demonstrating nearly logarithmic dependence of electrical conductivity over more than two decades in temperature. Using graphene's sheet conductance as in-situ thermometer, we present a measurement of electron-phonon heat transport at mK temperatures and show that it obeys the… ▽ More We investigate the basic charge and heat transport properties of charge neutral epigraphene at sub-kelvin temperatures, demonstrating nearly logarithmic dependence of electrical conductivity over more than two decades in temperature. Using graphene's sheet conductance as in-situ thermometer, we present a measurement of electron-phonon heat transport at mK temperatures and show that it obeys the $T^4$ dependence characteristic for clean two-dimensional conductor. Based on our measurement we predict the noise-equivalent power of $\sim 10^{-22}~{\rm W}/\sqrt{\rm Hz}$ of epigraphene bolometer at the low end of achievable temperatures. △ Less

Submitted 29 September, 2020; originally announced September 2020.

The performance limits of epigraphene Hall sensors

Authors: Hans He, Naveen Shetty, Thilo Bauch, Sergey Kubatkin, Timo Kaufmann, Martin Cornills, Rositsa Yakimova, Samuel Lara-Avila

Abstract: Epitaxial graphene on silicon carbide, or epigraphene, provides an excellent platform for Hall sensing devices in terms of both high electrical quality and scalability. However, the challenge in controlling its carrier density has thus far prevented systematic studies of epigraphene Hall sensor performance. In this work we investigate epigraphene Hall sensors where epigraphene is doped across the… ▽ More Epitaxial graphene on silicon carbide, or epigraphene, provides an excellent platform for Hall sensing devices in terms of both high electrical quality and scalability. However, the challenge in controlling its carrier density has thus far prevented systematic studies of epigraphene Hall sensor performance. In this work we investigate epigraphene Hall sensors where epigraphene is doped across the Dirac point using molecular do**. Depending on the carrier density, molecular-doped epigraphene Hall sensors reach room temperature sensitivities $S_V=0.23 V/VT$,$S_I=1440 V/AT$ and magnetic field detection limits down to $B_{MIN}=27$ $nT/\sqrt{Hz}$ at 20 kHz. Thermally stabilized devices demonstrate operation up to $T=150$ $^oC$ with $S_V=0.12 V/VT$, $S_I=300 V/AT$ and $B_{MIN}\approx 100$ $nT/\sqrt{Hz}$ at 20 kHz. △ Less

Submitted 18 March, 2020; originally announced March 2020.

Kernel Bi-Linear Modeling for Reconstructing Data on Manifolds: The Dynamic-MRI Case

Authors: Gaurav N. Shetty, Konstantinos Slavakis, Ukash Nakarmi, Gesualdo Scutari, Leslie Ying

Abstract: This paper establishes a kernel-based framework for reconstructing data on manifolds, tailored to fit the dynamic-(d)MRI-data recovery problem. The proposed methodology exploits simple tangent-space geometries of manifolds in reproducing kernel Hilbert spaces and follows classical kernel-approximation arguments to form the data-recovery task as a bi-linear inverse problem. Departing from mainstrea… ▽ More This paper establishes a kernel-based framework for reconstructing data on manifolds, tailored to fit the dynamic-(d)MRI-data recovery problem. The proposed methodology exploits simple tangent-space geometries of manifolds in reproducing kernel Hilbert spaces and follows classical kernel-approximation arguments to form the data-recovery task as a bi-linear inverse problem. Departing from mainstream approaches, the proposed methodology uses no training data, employs no graph Laplacian matrix to penalize the optimization task, uses no costly (kernel) pre-imaging step to map feature points back to the input space, and utilizes complex-valued kernel functions to account for k-space data. The framework is validated on synthetically generated dMRI data, where comparisons against state-of-the-art schemes highlight the rich potential of the proposed approach in data-recovery problems. △ Less

Submitted 26 February, 2020; originally announced February 2020.

Augmenting Variational Autoencoders with Sparse Labels: A Unified Framework for Unsupervised, Semi-(un)supervised, and Supervised Learning

Authors: Felix Berkhahn, Richard Keys, Wajih Ouertani, Nikhil Shetty, Dominik Geißler

Abstract: We present a new flavor of Variational Autoencoder (VAE) that interpolates seamlessly between unsupervised, semi-supervised and fully supervised learning domains. We show that unlabeled datapoints not only boost unsupervised tasks, but also the classification performance. Vice versa, every label not only improves classification, but also unsupervised tasks. The proposed architecture is simple: A c… ▽ More We present a new flavor of Variational Autoencoder (VAE) that interpolates seamlessly between unsupervised, semi-supervised and fully supervised learning domains. We show that unlabeled datapoints not only boost unsupervised tasks, but also the classification performance. Vice versa, every label not only improves classification, but also unsupervised tasks. The proposed architecture is simple: A classification layer is connected to the topmost encoder layer, and then combined with the resampled latent layer for the decoder. The usual evidence lower bound (ELBO) loss is supplemented with a supervised loss target on this classification layer that is only applied for labeled datapoints. This simplicity allows for extending any existing VAE model to our proposed semi-supervised framework with minimal effort. In the context of classification, we found that this approach even outperforms a direct supervised setup. △ Less

Submitted 14 November, 2019; v1 submitted 8 August, 2019; originally announced August 2019.

Bi-Linear Modeling of Data Manifolds for Dynamic-MRI Recovery

Authors: Gaurav N. Shetty, Konstantinos Slavakis, Abhishek Bose, Ukash Nakarmi, Gesualdo Scutari, Leslie Ying

Abstract: This paper puts forth a novel bi-linear modeling framework for data recovery via manifold-learning and sparse-approximation arguments and considers its application to dynamic magnetic-resonance imaging (dMRI). Each temporal-domain MR image is viewed as a point that lies onto or close to a smooth manifold, and landmark points are identified to describe the point cloud concisely. To facilitate compu… ▽ More This paper puts forth a novel bi-linear modeling framework for data recovery via manifold-learning and sparse-approximation arguments and considers its application to dynamic magnetic-resonance imaging (dMRI). Each temporal-domain MR image is viewed as a point that lies onto or close to a smooth manifold, and landmark points are identified to describe the point cloud concisely. To facilitate computations, a dimensionality reduction module generates low-dimensional/compressed renditions of the landmark points. Recovery of the high-fidelity MRI data is realized by solving a non-convex minimization task for the linear decompression operator and those affine combinations of landmark points which locally approximate the latent manifold geometry. An algorithm with guaranteed convergence to stationary solutions of the non-convex minimization task is also provided. The aforementioned framework exploits the underlying spatio-temporal patterns and geometry of the acquired data without any prior training on external data or information. Extensive numerical results on simulated as well as real cardiac-cine and perfusion MRI data illustrate noteworthy improvements of the advocated machine-learning framework over state-of-the-art reconstruction techniques. △ Less

Submitted 11 June, 2019; v1 submitted 26 December, 2018; originally announced December 2018.

UV-Vis-IR Spectral Complex Refractive Indices and Optical Properties of Brown Carbon Aerosol from Biomass Burning

Authors: Benjamin J. Sumlin, Yuli W. Heinson, Nishit Shetty, Apoorva Pandey, Robert S. Pattison, Stephen Baker, Wei Min Hao, Rajan K. Chakrabarty

Abstract: Constraining the complex refractive indices, optical properties and size of brown carbon (BrC) aerosols is a vital endeavor for improving climate models and satellite retrieval algorithms. Smoldering wildfires are the largest source of primary BrC, and fuel parameters such as moisture content, source depth, geographic origin, and fuel packing density could influence the properties of the emitted a… ▽ More Constraining the complex refractive indices, optical properties and size of brown carbon (BrC) aerosols is a vital endeavor for improving climate models and satellite retrieval algorithms. Smoldering wildfires are the largest source of primary BrC, and fuel parameters such as moisture content, source depth, geographic origin, and fuel packing density could influence the properties of the emitted aerosol. We measured in situ spectral (375-1047 nm) optical properties of BrC aerosols emitted from smoldering combustion of Boreal and Indonesian peatlands across a range of these fuel parameters. Inverse Lorenz-Mie algorithms used these optical measurements along with simultaneously measured particle size distributions to retrieve the aerosol complex refractive indices (m=n+iκ). Our results show that the real part n is constrained between 1.5 and 1.7 with no obvious functionality in wavelength (λ), moisture content, source depth, or geographic origin. With increasing λ from 375 to 532 nm, κ decreased from 0.014 to 0.003, with corresponding increase in single scattering albedo from 0.93 to 0.99. For λ = 532 nm, both κ and single scattering albedo showed no spectral dependency. We discuss differences between this study and previous work. The imaginary part κ was sensitive to changes in FPD, and we hypothesize mechanisms that might help explain this observation. △ Less

Submitted 14 December, 2017; v1 submitted 13 December, 2017; originally announced December 2017.

Comments: Benjamin J. Sumlin and Yuli W. Heinson contributed equally to this work

Testing Beam-Induced Quench Levels of LHC Superconducting Magnets

Authors: B. Auchmann, T. Baer, M. Bednarek, G. Bellodi, C. Bracco, R. Bruce, F. Cerutti, V. Chetvertkova, B. Dehning, P. P. Granieri, W. Hofle, E. B. Holzer, A. Lechner, E. Nebot Del Busto, A. Priebe, S. Redaelli, B. Salvachua, M. Sapinski, R. Schmidt, N. Shetty, E. Skordis, M. Solfaroli, J. Steckert, D. Valuch, A. Verweij , et al. (3 additional authors not shown)

Abstract: In the years 2009-2013 the Large Hadron Collider (LHC) has been operated with the top beam energies of 3.5 TeV and 4 TeV per proton (from 2012) instead of the nominal 7 TeV. The currents in the superconducting magnets were reduced accordingly. To date only seventeen beam-induced quenches have occurred; eight of them during specially designed quench tests, the others during injection. There has not… ▽ More In the years 2009-2013 the Large Hadron Collider (LHC) has been operated with the top beam energies of 3.5 TeV and 4 TeV per proton (from 2012) instead of the nominal 7 TeV. The currents in the superconducting magnets were reduced accordingly. To date only seventeen beam-induced quenches have occurred; eight of them during specially designed quench tests, the others during injection. There has not been a single beam- induced quench during normal collider operation with stored beam. The conditions, however, are expected to become much more challenging after the long LHC shutdown. The magnets will be operating at near nominal currents, and in the presence of high energy and high intensity beams with a stored energy of up to 362 MJ per beam. In this paper we summarize our efforts to understand the quench levels of LHC superconducting magnets. We describe beam-loss events and dedicated experiments with beam, as well as the simulation methods used to reproduce the observable signals. The simulated energy deposition in the coils is compared to the quench levels predicted by electro-thermal models, thus allowing to validate and improve the models which are used to set beam-dump thresholds on beam-loss monitors for Run 2. △ Less

Submitted 26 June, 2015; v1 submitted 18 February, 2015; originally announced February 2015.

Comments: 19 pages

Journal ref: Phys. Rev. ST Accel. Beams 18, 061002 (2015)

On the Shoulders of Giants: The Growing Impact of Older Articles

Authors: Alex Verstak, Anurag Acharya, Helder Suzuki, Sean Henderson, Mikhail Iakhiaev, Cliff Chiung Yu Lin, Namit Shetty

Abstract: In this paper, we examine the evolution of the impact of older scholarly articles. We attempt to answer four questions. First, how often are older articles cited and how has this changed over time. Second, how does the impact of older articles vary across different research fields. Third, is the change in the impact of older articles accelerating or slowing down. Fourth, are these trends different… ▽ More In this paper, we examine the evolution of the impact of older scholarly articles. We attempt to answer four questions. First, how often are older articles cited and how has this changed over time. Second, how does the impact of older articles vary across different research fields. Third, is the change in the impact of older articles accelerating or slowing down. Fourth, are these trends different for much older articles. To answer these questions, we studied citations from articles published in 1990-2013. We computed the fraction of citations to older articles from articles published each year as the measure of impact. We considered articles that were published at least 10 years before the citing article as older articles. We computed these numbers for 261 subject categories and 9 broad areas of research. Finally, we repeated the computation for two other definitions of older articles, 15 years and older and 20 years and older. There are three conclusions from our study. First, the impact of older articles has grown substantially over 1990-2013. In 2013, 36% of citations were to articles that are at least 10 years old; this fraction has grown 28% since 1990. The fraction of older citations increased over 1990-2013 for 7 out of 9 broad areas and 231 out of 261 subject categories. Second, the increase over the second half (2002-2013) was double the increase in the first half (1990-2001). Third, the trend of a growing impact of older articles also holds for even older articles. In 2013, 21% of citations were to articles >= 15 years old with an increase of 30% since 1990 and 13% of citations were to articles >= 20 years old with an increase of 36%. Now that finding and reading relevant older articles is about as easy as finding and reading recently published articles, significant advances aren't getting lost on the shelves and are influencing work worldwide for years after. △ Less

Submitted 2 November, 2014; originally announced November 2014.

Rise of the Rest: The Growing Impact of Non-Elite Journals

Authors: Anurag Acharya, Alex Verstak, Helder Suzuki, Sean Henderson, Mikhail Iakhiaev, Cliff Chiung Yu Lin, Namit Shetty

Abstract: In this paper, we examine the evolution of the impact of non-elite journals. We attempt to answer two questions. First, what fraction of the top-cited articles are published in non-elite journals and how has this changed over time. Second, what fraction of the total citations are to non-elite journals and how has this changed over time. We studied citations to articles published in 1995-2013. We… ▽ More In this paper, we examine the evolution of the impact of non-elite journals. We attempt to answer two questions. First, what fraction of the top-cited articles are published in non-elite journals and how has this changed over time. Second, what fraction of the total citations are to non-elite journals and how has this changed over time. We studied citations to articles published in 1995-2013. We computed the 10 most-cited journals and the 1000 most-cited articles each year for all 261 subject categories in Scholar Metrics. We marked the 10 most-cited journals in a category as the elite journals for the category and the rest as non-elite. There are two conclusions from our study. First, the fraction of top-cited articles published in non-elite journals increased steadily over 1995-2013. While the elite journals still publish a substantial fraction of high-impact articles, many more authors of well-regarded papers in diverse research fields are choosing other venues. The number of top-1000 papers published in non-elite journals for the representative subject category went from 149 in 1995 to 245 in 2013, a growth of 64%. Looking at broad research areas, 4 out of 9 areas saw at least one-third of the top-cited articles published in non-elite journals in 2013. For 6 out of 9 areas, the fraction of top-cited papers published in non-elite journals for the representative subject category grew by 45% or more. Second, now that finding and reading relevant articles in non-elite journals is about as easy as finding and reading articles in elite journals, researchers are increasingly building on and citing work published everywhere. Considering citations to all articles, the percentage of citations to articles in non-elite journals went from 27% in 1995 to 47% in 2013. Six out of nine broad areas had at least 50% of citations going to articles published in non-elite journals in 2013. △ Less

Submitted 8 October, 2014; originally announced October 2014.

Agent Based Intelligent Alert System for Smart-Phones

Authors: Sandeep Venkatesh, Shreyas Balakuntala, Rajarajeswari S, Nytika N Shetty, Namratha Shetty, Neha Sudhakar

Abstract: The paper deals with the design of an agent which modifies and enhances the various alert systems in the smartphones. The actions of the agent includes sorting the notifications abiding to human thinking, hel** the user to have a safe conversation, assisting in tracking back the reach-ability status of the caller when needed, conveying the user about the notifications in times of situations like… ▽ More The paper deals with the design of an agent which modifies and enhances the various alert systems in the smartphones. The actions of the agent includes sorting the notifications abiding to human thinking, hel** the user to have a safe conversation, assisting in tracking back the reach-ability status of the caller when needed, conveying the user about the notifications in times of situations like drained battery and smartly alerting the user in situations like slee**. The agent uses the information gathered from a survey, to modify the existing methods of alerts and produce alerts which abide by the human cognitive responses. △ Less

Submitted 25 May, 2013; originally announced May 2013.

Comments: 4 pages, 3 figures