-
Neural Optimization: Understanding Trade-offs with Pareto Theory
Authors:
Fabian Pallasdies,
Philipp Norton,
Jan-Hendrik Schleimer,
Susanne Schreiber
Abstract:
Nervous systems, like any organismal structure, have been shaped by evolutionary processes to increase fitness. The resulting neural 'bauplan' has to account for multiple objectives simultaneously, including computational function as well as additional factors like robustness to environmental changes and energetic limitations. Oftentimes these objectives compete and a quantification of the relativ…
▽ More
Nervous systems, like any organismal structure, have been shaped by evolutionary processes to increase fitness. The resulting neural 'bauplan' has to account for multiple objectives simultaneously, including computational function as well as additional factors like robustness to environmental changes and energetic limitations. Oftentimes these objectives compete and a quantification of the relative impact of individual optimization targets is non-trivial. Pareto optimality offers a theoretical framework to decipher objectives and trade-offs between them. We, therefore, highlight Pareto theory as a useful tool for the analysis of neurobiological systems, from biophysically-detailed cells to large-scale network structures and behavior. The Pareto approach can help to assess optimality, identify relevant objectives and their respective impact, and formulate testable hypotheses.
△ Less
Submitted 4 May, 2021;
originally announced May 2021.
-
Trustworthy AI Inference Systems: An Industry Research View
Authors:
Rosario Cammarota,
Matthias Schunter,
Anand Rajan,
Fabian Boemer,
Ágnes Kiss,
Amos Treiber,
Christian Weinert,
Thomas Schneider,
Emmanuel Stapf,
Ahmad-Reza Sadeghi,
Daniel Demmler,
Joshua Stock,
Huili Chen,
Siam Umar Hussain,
Sadegh Riazi,
Farinaz Koushanfar,
Saransh Gupta,
Tajan Simunic Rosing,
Kamalika Chaudhuri,
Hamid Nejatollahi,
Nikil Dutt,
Mohsen Imani,
Kim Laine,
Anuj Dubey,
Aydin Aysu
, et al. (4 additional authors not shown)
Abstract:
In this work, we provide an industry research view for approaching the design, deployment, and operation of trustworthy Artificial Intelligence (AI) inference systems. Such systems provide customers with timely, informed, and customized inferences to aid their decision, while at the same time utilizing appropriate security protection mechanisms for AI models. Additionally, such systems should also…
▽ More
In this work, we provide an industry research view for approaching the design, deployment, and operation of trustworthy Artificial Intelligence (AI) inference systems. Such systems provide customers with timely, informed, and customized inferences to aid their decision, while at the same time utilizing appropriate security protection mechanisms for AI models. Additionally, such systems should also use Privacy-Enhancing Technologies (PETs) to protect customers' data at any time. To approach the subject, we start by introducing current trends in AI inference systems. We continue by elaborating on the relationship between Intellectual Property (IP) and private data protection in such systems. Regarding the protection mechanisms, we survey the security and privacy building blocks instrumental in designing, building, deploying, and operating private AI inference systems. For example, we highlight opportunities and challenges in AI systems using trusted execution environments combined with more recent advances in cryptographic techniques to protect data in use. Finally, we outline areas of further development that require the global collective attention of industry, academia, and government researchers to sustain the operation of trustworthy AI inference systems.
△ Less
Submitted 10 February, 2023; v1 submitted 10 August, 2020;
originally announced August 2020.
-
Characterization of lemniscate atmospheric aberrations in Gemini Planet Imager data
Authors:
Alexander Madurowicz,
Bruce A. Macintosh,
Jean-Baptiste Ruffio,
Jeffery Chilcote,
Vanessa P. Bailey,
Lisa Poyneer,
Eric Nielsen,
Andrew P. Norton
Abstract:
A semi analytic framework for simulating the effects of atmospheric seeing in Adaptive Optics systems on an 8-m telescope is developed with the intention of understanding the origin of the wind-butterfly, a characteristic two-lobed halo in the PSF of AO imaging. Simulations show that errors in the compensated phase on the aperture due to servo-lag have preferential direction orthogonal to the dire…
▽ More
A semi analytic framework for simulating the effects of atmospheric seeing in Adaptive Optics systems on an 8-m telescope is developed with the intention of understanding the origin of the wind-butterfly, a characteristic two-lobed halo in the PSF of AO imaging. Simulations show that errors in the compensated phase on the aperture due to servo-lag have preferential direction orthogonal to the direction of wind propagation which, when Fourier Transformed into the image plane, appear with their characteristic lemniscate shape along the wind direction. We develop a metric to quantify the effect of this aberration with the fractional standard deviation in an annulus centered around the PSF, and use telescope pointing to correlate this effect with data from an atmospheric models, the NOAA GFS. Our results show that the jet stream at altitudes of 100-200 hPa (equivalently 10-15 km above sea level) is highly correlated (13.2$σ$) with the strong butterfly, while the ground wind and other layers are more or less uncorrelated.
△ Less
Submitted 18 July, 2018;
originally announced July 2018.
-
Adversarial-Playground: A Visualization Suite Showing How Adversarial Examples Fool Deep Learning
Authors:
Andrew P. Norton,
Yanjun Qi
Abstract:
Recent studies have shown that attackers can force deep learning models to misclassify so-called "adversarial examples": maliciously generated images formed by making imperceptible modifications to pixel values. With growing interest in deep learning for security applications, it is important for security experts and users of machine learning to recognize how learning systems may be attacked. Due…
▽ More
Recent studies have shown that attackers can force deep learning models to misclassify so-called "adversarial examples": maliciously generated images formed by making imperceptible modifications to pixel values. With growing interest in deep learning for security applications, it is important for security experts and users of machine learning to recognize how learning systems may be attacked. Due to the complex nature of deep learning, it is challenging to understand how deep models can be fooled by adversarial examples. Thus, we present a web-based visualization tool, Adversarial-Playground, to demonstrate the efficacy of common adversarial methods against a convolutional neural network (CNN) system. Adversarial-Playground is educational, modular and interactive. (1) It enables non-experts to compare examples visually and to understand why an adversarial example can fool a CNN-based image classifier. (2) It can help security experts explore more vulnerability of deep learning as a software module. (3) Building an interactive visualization is challenging in this domain due to the large feature space of image classification (generating adversarial examples is slow in general and visualizing images are costly). Through multiple novel design choices, our tool can provide fast and accurate responses to user requests. Empirically, we find that our client-server division strategy reduced the response time by an average of 1.5 seconds per sample. Our other innovation, a faster variant of JSMA evasion algorithm, empirically performed twice as fast as JSMA and yet maintains a comparable evasion rate.
Project source code and data from our experiments available at: https://github.com/QData/AdversarialDNN-Playground
△ Less
Submitted 1 August, 2017;
originally announced August 2017.
-
MEMS practice, from the lab to the telescope
Authors:
Katie M. Morzinski,
Andrew P. Norton,
Julia Wilhelmson Evans,
Layra Reza,
Scott A. Severson,
Daren Dillon,
Marc Reinig,
Donald T. Gavel,
Steven Cornelissen,
Bruce A. Macintosh,
Claire E. Max
Abstract:
Micro-electro-mechanical systems (MEMS) technology can provide for deformable mirrors (DMs) with excellent performance within a favorable economy of scale. Large MEMS-based astronomical adaptive optics (AO) systems such as the Gemini Planet Imager are coming on-line soon. As MEMS DM end-users, we discuss our decade of practice with the micromirrors, from inspecting and characterizing devices to ev…
▽ More
Micro-electro-mechanical systems (MEMS) technology can provide for deformable mirrors (DMs) with excellent performance within a favorable economy of scale. Large MEMS-based astronomical adaptive optics (AO) systems such as the Gemini Planet Imager are coming on-line soon. As MEMS DM end-users, we discuss our decade of practice with the micromirrors, from inspecting and characterizing devices to evaluating their performance in the lab. We also show MEMS wavefront correction on-sky with the "Villages" AO system on a 1-m telescope, including open-loop control and visible-light imaging. Our work demonstrates the maturity of MEMS technology for astronomical adaptive optics.
△ Less
Submitted 7 February, 2012;
originally announced February 2012.
-
MICE: the Muon Ionization Cooling Experiment. Step I: First Measurement of Emittance with Particle Physics Detectors
Authors:
U. Bravar,
M. Bogomilov,
Y. Karadzhov,
D. Kolev,
I. Russinov,
R. Tsenov,
L. Wang,
F. Y. Xu,
S. X. Zheng,
R. Bertoni,
M. Bonesini,
R. Mazza,
V. Palladino,
G. Cecchet,
A. de Bari,
M. Capponi,
A. Iaciofano,
D. Orestano,
F. Pastore,
L. Tortora,
S. Ishimoto,
S. Suzuki,
K. Yoshimura,
Y. Mori,
Y. Kuno
, et al. (123 additional authors not shown)
Abstract:
The Muon Ionization Cooling Experiment (MICE) is a strategic R&D project intended to demonstrate the only practical solution to providing high brilliance beams necessary for a neutrino factory or muon collider. MICE is under development at the Rutherford Appleton Laboratory (RAL) in the United Kingdom. It comprises a dedicated beamline to generate a range of input muon emittances and momenta, with…
▽ More
The Muon Ionization Cooling Experiment (MICE) is a strategic R&D project intended to demonstrate the only practical solution to providing high brilliance beams necessary for a neutrino factory or muon collider. MICE is under development at the Rutherford Appleton Laboratory (RAL) in the United Kingdom. It comprises a dedicated beamline to generate a range of input muon emittances and momenta, with time-of-flight and Cherenkov detectors to ensure a pure muon beam. The emittance of the incoming beam will be measured in the upstream magnetic spectrometer with a scintillating fiber tracker. A cooling cell will then follow, alternating energy loss in Liquid Hydrogen (LH2) absorbers to RF cavity acceleration. A second spectrometer, identical to the first, and a second muon identification system will measure the outgoing emittance. In the 2010 run at RAL the muon beamline and most detectors were fully commissioned and a first measurement of the emittance of the muon beam with particle physics (time-of-flight) detectors was performed. The analysis of these data was recently completed and is discussed in this paper. Future steps for MICE, where beam emittance and emittance reduction (cooling) are to be measured with greater accuracy, are also presented.
△ Less
Submitted 30 July, 2013; v1 submitted 9 October, 2011;
originally announced October 2011.
-
Search for neutral Higgs bosons decaying into four taus at LEP2
Authors:
ALEPH Collaboration,
S. Schael,
R. Barate,
R. Brunelière,
I. De Bonis,
D. Decamp,
C. Goy,
S. Jézéquel,
J. -P. Lees,
F. Martin,
E. Merle,
M. -N. Minard,
B. Pietrzyk,
B. Trocmé S. Bravo,
M. P. Casado,
M. Chmeissani,
J. M. Crespo,
E. Fernandez,
M. Fernandez-Bosman,
Ll. Garrido,
M. Martinez,
A. Pacheco,
H. Ruiz,
A. Colaleo,
D. Creanza
, et al. (236 additional authors not shown)
Abstract:
A search for the production and non-standard decay of a Higgs boson, h, into four taus through intermediate pseudoscalars, a, is conducted on 683 pb-1 of data collected by the ALEPH experiment at centre-of-mass energies from 183 to 209 GeV. No excess of events above background is observed, and exclusion limits are placed on the combined production cross section times branching ratio, ξ^2 = σ(e+e…
▽ More
A search for the production and non-standard decay of a Higgs boson, h, into four taus through intermediate pseudoscalars, a, is conducted on 683 pb-1 of data collected by the ALEPH experiment at centre-of-mass energies from 183 to 209 GeV. No excess of events above background is observed, and exclusion limits are placed on the combined production cross section times branching ratio, ξ^2 = σ(e+e- --> Zh)/σ_{SM}(e+e- --> Zh) x B(h --> aa)x B(a --> τ^+τ^-)^2. For mh < 107 GeV/c2 and 4 < ma < 10 GeV/c2, ξ^2 > 1 is excluded at the 95% confidence level.
△ Less
Submitted 19 April, 2010; v1 submitted 2 March, 2010;
originally announced March 2010.
-
The scattering of muons in low Z materials
Authors:
MuScat Collaboration,
D. Attwood,
P. Bell,
S. Bull,
T. McMahon,
J. Wilson,
R. Fernow,
P. Gruber,
A. Jamdagni,
K. Long,
E. McKigney,
P. Savage,
M. Curtis-Rouse,
T. R. Edgecock,
M. Ellis,
J. Lidbury,
W. J. Murray,
P. Norton,
K. Peach,
K. Ishida,
Y. Matsuda,
K. Nagamine,
S. Nakamura,
G. M. Marshall,
S. Benveniste
, et al. (6 additional authors not shown)
Abstract:
This paper presents the measurement of the scattering of 172 MeV/c muons in assorted materials, including liquid hydrogen, motivated by the need to understand ionisation cooling for muon acceleration.
Data are compared with predictions from the Geant 4 simulation code and this simulation is used to deconvolute detector effects. The scattering distributions obtained are compared with the Molier…
▽ More
This paper presents the measurement of the scattering of 172 MeV/c muons in assorted materials, including liquid hydrogen, motivated by the need to understand ionisation cooling for muon acceleration.
Data are compared with predictions from the Geant 4 simulation code and this simulation is used to deconvolute detector effects. The scattering distributions obtained are compared with the Moliere theory of multiple scattering and, in the case of liquid hydrogen, with ELMS. With the exception of ELMS, none of the models are found to provide a good description of the data. The results suggest that ionisation cooling will work better than would be predicted by Geant 4.7.0p01.
△ Less
Submitted 2 December, 2005;
originally announced December 2005.
-
RHEED Studies of Epitaxial Oxide Seed-Layer Growth on RABiTS Ni(001): The Role of Surface Structure and Chemistry
Authors:
C. Cantoni,
D. K. Christen,
R. Feenstra,
A. Goyal,
G. W. Ownby,
D. M. Zehner,
D. P. Norton
Abstract:
The epitaxial deposition of the first oxide buffer layer (seed layer) on biaxially textured Ni tape for coated conductors is a critical step that is dependent on the atomistic surface condition of the metal. We present a study of the {100}<100> biaxially textured Ni (001) surface and seed-layer growth using in situ reflection high-energy electron diffraction (RHEED) and Auger electron spectrosco…
▽ More
The epitaxial deposition of the first oxide buffer layer (seed layer) on biaxially textured Ni tape for coated conductors is a critical step that is dependent on the atomistic surface condition of the metal. We present a study of the {100}<100> biaxially textured Ni (001) surface and seed-layer growth using in situ reflection high-energy electron diffraction (RHEED) and Auger electron spectroscopy (AES). Our observations are consistent with formation of a c(2 x 2) 2-D superstructure due to surface segregation of sulfur contained in the metal. We show that this superstructure can have a dramatic effect on the heteroepitaxial growth of oxide seed layers. In particular, the surface superstructure promotes the (200) epitaxial oxide growth of Y2O3-stabilized ZrO2 (YSZ), which is necessary for the development of high-Jc superconducting films for coated conductors.
△ Less
Submitted 13 June, 2001;
originally announced June 2001.
-
Superconducting magnesium diboride films with Tc \approx 24K grown by pulsed laser deposition with in-situ anneal
Authors:
H. M. Christen,
H. Y. Zhai,
C. Cantoni,
M. Paranthaman,
B. C. Sales,
C. Rouleau,
D. P. Norton,
D. K. Christen,
D. H. Lowndes,
.
Abstract:
Thin superconducting films of magnesium diboride (MgB2) with Tc \approx 24K were prepared on various oxide substrates by pulsed laser deposition (PLD) followed by an in-situ anneal. A systematic study of the influence of various in-situ annealing parameters shows an optimum temperature of about 600C in a background of 0.7 atm. of Ar/4%H2 for layers consisting of a mixture of magnesium and boron.…
▽ More
Thin superconducting films of magnesium diboride (MgB2) with Tc \approx 24K were prepared on various oxide substrates by pulsed laser deposition (PLD) followed by an in-situ anneal. A systematic study of the influence of various in-situ annealing parameters shows an optimum temperature of about 600C in a background of 0.7 atm. of Ar/4%H2 for layers consisting of a mixture of magnesium and boron. Contrary to ex-situ approaches (e.g. reacting boron films with magnesium vapor at 900C), these films are processed below the decomposition temperature of MgB2. This may prove enabling in the formation of multilayers, junctions, and epitaxial films in future work. Issues related to the improvement of these films and to the possible in-situ growth of MgB2 at elevated temperature are discussed.
△ Less
Submitted 22 March, 2001;
originally announced March 2001.
-
Conductive Buffer Layers and Overlayers for the Thermal Stability of Coated Conductors
Authors:
Claudia Cantoni,
Tolga Aytug,
Darren T. Verebelyi,
Mariappan Paranthaman,
Eliot D. Specht,
David P. Norton,
David K. Christen
Abstract:
We analyze fundamental issues related to the thermal and electrical stability of a coated conductor during its operation. We address the role of conductive buffer layers in the stability of Ni-based coated conductors, and the effect of a metallic cap layer on the electrical properties of Ni alloy-based superconducting tapes. For the first case we report on the fabrication of a fully conductive R…
▽ More
We analyze fundamental issues related to the thermal and electrical stability of a coated conductor during its operation. We address the role of conductive buffer layers in the stability of Ni-based coated conductors, and the effect of a metallic cap layer on the electrical properties of Ni alloy-based superconducting tapes. For the first case we report on the fabrication of a fully conductive RABiTS architecture formed of bilayers of conductive oxides SrRuO3 and LaNiO3 on textured Ni tapes. For the second case we discuss measurements of current-voltage relations on Ag/YBa2Cu3O7-d and Cu/Ag/ YBa2Cu3O7-d prototype multilayers on insulating substrates. Limitations on the overall tape structure and properties that are posed by the stability requirement are presented.
△ Less
Submitted 25 January, 2001;
originally announced January 2001.
-
Associative subalgebras of the Griess algebra and related topics
Authors:
C. Dong,
H. Li,
G. Mason,
S. P. Norton
Abstract:
It is shown how certain idempotents in the Griess algebra generate the discrete series representations for the Virasoro algebra inside the Frenkel-Lepowsky-Meurman's moonshine module vertex operator algebra. It is also shown that each Niemeier lattice determines (in many ways) certain maximal associative subalgebras of the Griess algebra.
It is shown how certain idempotents in the Griess algebra generate the discrete series representations for the Virasoro algebra inside the Frenkel-Lepowsky-Meurman's moonshine module vertex operator algebra. It is also shown that each Niemeier lattice determines (in many ways) certain maximal associative subalgebras of the Griess algebra.
△ Less
Submitted 9 July, 1996;
originally announced July 1996.
