Search

Neural style transfer aims at transforming the artistic style of a painting into a photograph. The classic methods are based on the one-to-one principle, where stylization is performed based on a single style image. In the paper, an approach is proposed that allows transferring the style based on multiple images of the same reference painter multiplied, in addition, by data augmentation. Comparing to the original method, the proposed approach produces visibly better results, and allows for greater flexibility in the transferring process. The proposed many-to-one method is illustrated by an example of transferring the style of two Polish painters using the same content image

Clay tablets are the oldest handwritten documents using the cuneiform script named after the wedge shaped imprints of a rectangular stylus left in their surface. Therefore the most suitable documentation technique is 3D acquisition using e.g. structured light. The tablets are heterogeneously shaped and may contain damage in varying degrees. Convolutional neural networks enabled large advances in the analysis of historical script written with ink on paper. However, cuneiform tablets remain inaccessible to common raster-image based recognition methods for handwriting, because the characters are actually represented by 3D shapes. With the release of the Heidelberg Cuneiform Benchmark Dataset (HeiCuBeDa) a plethora of machine learning opportunities for the underlaying representation, 3D surface meshes, was made accessible. Additionally HeiCuBeDa contains transliterations and meta-data such as language-type for approx. 1/3 of the 1.977 tablets. In this work, we combine and adapt the convolution operation of SplineNet with the pooling from PointNet++ to predict the time-period of a tablet directly on basis of its mesh representation. The classification tasks were performed on 336 tablets for training and 158 tablets for testing. Our proposed approach reaches a classification accuracy of 84 % on 4 time-period classes.

This paper discusses the design of a high-speed Static Random Access Memory (SRAM) which is tailored for convolutional neural network (CNN). Training process of CNNs requires memory to be accessed with different data widths, of which the most popular data widths are from 1 byte to 4 bytes. Traditional SRAMs have a set read/write data width and thus can only read/write one byte at a time. This slows down the training process of CNNs. SRAMs have become one of the bottlenecks of CNNs training speed. We proposed an SRAM with a new architecture that can read/write at flexible data widths. It can read/write any data from 1 byte to 4 bytes. This allows multiple bytes of data to be accessed each clock cycle and increases the memory access speed up to 4 times compared to traditional SRAMs. This can greatly improve the CNN training speed. A 1 KB traditional SRAM and another 1 KB SRAM based on the proposed architecture are also designed and simulated to further verify our design concepts. Power overhead and layout area overhead are also analyzed between the proposed SRAM architecture and the traditional SRAM.

The number of known cuneiform tablets is assumed to be in the hundreds of thousands. The Hilprecht Archive Online contains 1977 high-resolution 3D scans of tablets. The online cuneiform database CDLI catalogs metadata for more than 100.000 tablets. While both are accessible publicly, large-scale machine learning and pattern recognition on cuneiform tablets remain elusive. The data is only accessible by searching web pages, the tablet identifiers between collections are inconsistent, and the 3D data is unprepared and challenging for automated processing. We pave the way for large-scale analyses of cuneiform tablets by assembling a cross-referenced benchmark dataset of processed cuneiform tablets: (i) frontally aligned 3D tablets with pre-computed high-dimensional surface features, (ii) six-views raster images for off-the-shelf image processing, and (iii) metadata, transcriptions, and transliterations, for a subset of 707 tablets, for learning alignment between 3D data, image and linguistic expression. This is the first dataset of its kind and of its size in cuneiform research. This benchmark dataset is prepared for ease-of-use and immediate availability for computational researches, lowering the barrier to experiment and apply standard methods of analysis, at https://doi.org/10.11588/data/IE8CCN.

Increased use of 3D digitization methods leads to large numbers of 3D models with script like cuneiform tablets, temple wall inscriptions from the Mayan empire and Minoan and Mycenaean seal imprints. Analyzing and in particular comparing those objects in our projects is a core challenge for robust symbol-spotting and transcription of ancient scripts. The discrete manifolds using irregular grids describe the objects' surface and contain complex shapes that cannot be wholly captured by a single rendering to a two dimensional plane in an image format. As the plethora of standard image descriptor extraction methods is not available on those 3D meshes we propose a novel approach adapting relevant image descriptor methods to process irregular grids. We show preliminary evaluations of our approach by performing a symbol spotting task on cuneiform tablets. Our approach proceeds in three stages: (i) We compute a local surface feature with Multi-scale Integral Invariants (MSII), (ii) for each vertex we extract radial geodesic patches which are embedded locally into a polar coordinate system, (iii) a rectangular sampling scheme converts these patches into raster image patches that are accessible to 2D image descriptors. For the symbol spotting task we compute Histogram of Oriented Gradients (HOG) and Oriented FAST and Rotated BRIEF (ORB) descriptors on the query mesh and document mesh for each vertex. Similar symbols are retrieved by sliding a geodesic patch along the document mesh. Like in many other Cultural Heritage applications we have to consider the lack of any ground-truth. Therefore we provide a qualitative analysis of the spotting results showing that a query mesh is reliably found on the document mesh.

Motor symptoms in Parkinson's disease (PD) correlate with an excess in synchrony in the beta frequency band (13-30Hz) of local field potentials recorded from basal ganglia circuits. Recent results have suggested that this abnormal activity arises as a result of changes in specific dynamical features of the underlying neural signatures. In particular, patterns of activity in the beta band have been shown to be structured in bursts of longer durations and higher amplitudes in untreated patients with PD. Closed-loop deep brain stimulation (DBS) paradigms that specifically target these pathological bursts of activity hold promises to help trim, and thus normalize, their abnormal behavior in real-time. Here, we developed classification algorithms that predict pathological beta bursts based on ongoing changes in LFP frequency dynamics. We then compared simulations of prediction-based DBS profiles with existing 'adaptive DBS' alternatives. We show that model-driven stimulation profiles are more precise in restricting the delivery of stimulation to bursts that are considered pathological, while preserving physiological ones. The overall stimulation time required is also diminished, thus supporting longer battery life. These results represent a conceptual and algorithmic framework for the development of more precise DBS strategies that are selectively tailored to the electrophysiological profile of each patient.

Cuneiform tablets appertain to the oldest textual artifacts used for more than three millennia and are comparable in amount and relevance to texts written in Latin or ancient Greek. We present a complete digital analysis workflow enabling modern text processing on the complex and non-linear script. Our tools encompass the whole pipeline, from digitization and wedge extraction to word-spotting and frequent pattern mining facilities. Tablets are being acquired from different sources requiring different methods for digitalization. Each representation is typically processed with its own tool-set. To homogenize these data sources, we introduce an unifying minimal wedge constellation description. For this representation, we develop similarity metrics based on the optimal assignment of wedge configurations. We combine our wedge features with work on segmentation-free word spotting using part-structured models. The presented search and similarity facilities enable the development of advanced linguistic tools for cuneiform sign indexing and spatial n-gram mining of signs.

Cuneiform tablets appertain to the oldest textual artifacts and are in extent comparable to texts written in Latin or ancient Greek. The Cuneiform Commentaries Project (CPP) from Yale University provides tracings of cuneiform tablets with annotated transliterations and translations. As a part of our work analyzing cuneiform script computationally with 3D-acquisition and word-spotting, we present a first approach for automatized learning of transliterations of cuneiform tablets based on a corpus of parallel lines. These consist of manually drawn cuneiform characters and their transliteration into an alphanumeric code. Since the Cuneiform script is only available as raster-data, we segment lines with a projection profile, extract Histogram of oriented Gradients (HoG) features, detect outliers caused by tablet damage, and align those features with the transliteration. We apply methods from part-of-speech tagging to learn a correspondence between features and transliteration tokens. We evaluate point-wise classification with K-Nearest Neighbors (KNN) and a Support Vector Machine (SVM); sequence classification with a Hidden Markov Model (HMM) and a Structured Support Vector Machine (SVM-HMM). Analyzing our findings, we reach the conclusion that the sparsity of data, inconsistent labeling and the variety of tracing styles do currently not allow for fully automatized transliterations with the presented approach. However, the pursuit of automated learning of transliterations is of great relevance as manual annotation in larger quantities is not viable, given the few experts capable of transcribing cuneiform tablets.

Cuneiform scripts constitute an immense source of information about ancient history, dating back almost four thousand years. Documents were written by imprinting wedge-shaped impressions into wet clay tablets, and current scholarly practice typically transcribes the resulting markings by hand with ink on paper. This work develops algorithmic methods for cuneiform script, combining feature extraction for cuneiform wedges with prior work on segmentation-free word spotting using part-structured models. We adapt the inkball model used for word spotting to treat wedge features as individual parts arranged in a tree structure. The geometric relationship between query and target is measured by the energy necessary to deform the tree structure. We also introduce an optimizing method for wedge feature extraction based on optimally assigning tablet structuring elements to hypothesized wedge models. Finally, we evaluate the method on a real-world dataset, and show that it outperforms the state of the art in cuneiform character spotting.

Motivated by the increasing demand for computerized analysis of documents within the Digital Humanities we present an approach to automating handwritten cuneiform character recognition on vectorized cuneiform tablets. Cuneiform is one of the oldest handwritten scripts used for more than three millennia. In previous work we have shown how to extract vector drawings from 3D-models of cuneiform tablets similar to those manually drawn over digital photographs. We approach the problem of recognizing these characters by applying pattern matching against the basic structural features of cuneiform, the wedge-shaped impressions. Then, we find an optimal assignment between the wedge configuration of two characters w.r.t. wedge shape and position. The similarity of two characters is measured by the quality of the assignment. We compare our method against well known methods for handwritten character recognition with favorable results for our method.

Using a 2Mb embedded Flash cell array as a demonstrator, we reported previously that a 3 V reduction in programming voltage was possible by replacing the ONO inter- poly dielectric (IPD) with an IPD comprising the high-k material, AI2O3 by Kakoschke, R., et al, (2007). Adding a thin protective top-oxide to the high-k IPD was later shown to significantly improve reliability in Power, J. R., et al, (2007). In this paper, we show that for integration schemes more suited to the material properties of AI2O3, reliably functioning 2 Mb demonstrators with a high-k IPD but without top-oxide protection are also feasible.

Experimental studies of fundamental structure of nucleons require an electron-ion collider of a center-of- mass energy up to 90 GeV at luminosity up to 1035 cm-2 s-1 with both beams polarized. A CEBAF-based collider of 9 GeV electrons/positrons and 225 GeV ions is envisioned to meet this science need and as a next step for CEBAF after the planned 12 GeV energy upgrade of the fixed target program. A ring-ring scheme of this collider developed recently takes advantage of the existing polarized electron CW beam from the CEBAF and a green-field design of an ion complex with electron cooling. We present a conceptual design and report design studies of this high-luminosity collider.

Parametric-resonance ionization cooling (PIC) is a muon-cooling technique that is useful for low-emittance muon colliders. This method requires a well-tuned focusing channel that is free of chromatic and spherical aberrations. In order to be of practical use in a muon collider, it also necessary that the focusing channel be as short as possible to minimize muon loss due to decay. G4Beamline numerical simulations are presented of a compact PIC focusing channel in which spherical aberrations are minimized by using design symmetry.

The idea of coalescing multiple muon bunches at high energy to enhance the luminosity of a muon collider provides many advantages. It circumvents space-charge, beam loading, and wakefield problems of intense low- energy bunches while restoring the synergy between muon colliders and neutrino factories based on muon storage rings. A sampling of initial conceptual design work for a coalescing ring is presented here.

A helical cooling channel (HCC) has been proposed to quickly reduce the six dimensional phase space of muon beams for muon colliders, neutrino factories, and intense muon sources. The HCC is composed of solenoidal, helical dipole, and helical quadrupole current coils to provide focusing and dispersion needed for emittance exchange as the beam follows an equilibrium helical orbit. Inside the coils constituting the HCC examined here, a series of RF cavities filled with dense hydrogen gas acts as the energy absorber for ionization cooling and also suppressed RF breakdown. Two Monte Carlo simulation programs have been developed to compare HCC performance with analytic predictions and to begin the process of optimizing practical designs that could be built in the near future. We discuss the programs, the comparisons with the analytical theory, and the prospects for a HCC design with the capability to reduce the six-dimensional phase space emittance of a muon beam by a factor of over five orders of magnitude in a linear channel less than 100 meters long.

Parametric-resonance ionization cooling (PIC) is being developed to create small beams so that high muon collider luminosity can be achieved with fewer muons. In the linear channel that is studied in this effort, a half integer resonance is induced such that the normal elliptical motion of particles in x − x′ phase space becomes hyperbolic, with particles moving to smaller x and larger x′ as they pass down the channel. Thin absorbers placed at the focal points of the channel then cool the angular divergence of the beam by the usual ionization cooling mechanism where each absorber is followed by RF cavities. Thus the phase space of the beam is compressed in transverse position by the dynamics of the resonance and its angular divergence is compressed by the ionization cooling mechanism. We report the first results of simulations of this process, a study of the compensation of chromatic aberration by using synchrotron oscillations.

We report on the progress of the conceptual development of the energy recovering linac (ERL)-based electron-light ion collider (ELIC) at CEBAF that is envisioned to reach luminosity level of 1033-1035/cm2s with both beams polarized to perform a new class of experiments in fundamental nuclear physics. Four interaction points with all light ion species longitudinally or transversally polarized and fast flipping of the spin for all beams are planned. The unusually high luminosity concept is based on the use of the electron cooling and crab crossing colliding beams. Our recent studies focused on the design of low beta interaction points, exploration on raising the polarized electron injector current to the level of 3-30 mA with the use of electron circulator-collider ring, forming a concept of stacking and cooling of the ion beams, and specifications of the electron cooling facility.

The Electron-Light Ion Collider ELIC proposed by Jefferson Lab aims at a luminosity in the 1035cm-2sec-1range for collisions of 150 GeV protons on 7 GeV electrons [1, 2]. To achieve these high luminosities, very strong low-β focusing of low-emittance beams is required. Taking advantage of the unequal design proton beam emittances in the two transverse planes, an interaction region design based on superconducting quadrupole doublets has been developed. Compared with the original design in [2], this scheme provides larger beam apertures at lower magnetic fields, while potentially doubling the luminosity.

Beam physics aspects of the 12 GeV Upgrade of CEBAF are presented. The CEBAF Upgrade to 12 GeV is achieved via 5.5 recirculations through the linacs, and the installation of 10 new high-gradient cryomodules. A new experimental hall, Hall D, is envisioned at the end of the North Linac. Simulation results for a straight-ahead and a recirculated injector are summarized and compared. Beam transport designs are discussed and evaluated with respect to matching and beam breakup (BBU) optimization. Effects of synchrotron radiation excitation on the beam properties are calculated. BBU simulations and derived specifications for the damping of higher order modes of the new 7-cell cavities are presented. The energies that provide longitudinal polarization in multiple experimental halls simultaneously are calculated. Finally, detailed optics of the Hall D transport line has been obtained.

The CEBAF recirculator at Jefferson Lab includes two linear accelerators, each 200 meters in length. Due to varying betatron phase advance for different recirculation passes, misalignment, and other steering effects, orbit correction in the CEBAF linacs presents a complicated problem defying pass-by-pass solutions. Utilization of information from the beam position measurements at all recirculation passes allows us not only to perform multipass steering minimizing beam displacements inside the linacs, but also to determine displacements of linac BPM's and focusing quadrupoles from an ideal axis. This paper describes a steering algorithm and presents the experience in multi-pass orbit correction.

The use of a longitudinal phase-space tracking code, ESME, to simulate the growth of a coupled-bunch instability in the Fermilab Booster is examined. The calculation of the resonant response is described. Two simulations are presented. In the first, all 84 RF buckets are populated with a bi-Gaussian distribution of emittance of 0.02 eV-s. Each bunch is represented by 100 macro-particles. The entire distribution (the ring) is divided into 3000 bins for the purposes of computing and applying the voltage due to the resonance. The charge corresponding to each bunch is set to 6.6*10/sup 12/ protons. In the second simulation, the parameters are the same as in the first, except for the fact that five buckets are not initially populated. The absence of these bunches enhanced the growth of the mode considerably. The induced voltage in the second case rises to an amplitude an order of magnitude larger than the first, 200 kV vs. 7 kV. The larger growth of the mode in the gapped-beam case may be caused by the enhancement of the harmonic component of the current at mode 56 initially, whereas in the symmetrical case the instability grows out of noise. >

Possible intensity limits due to the coherent betatron motion are outlined for the upgraded Tevatron with electrostatic separators. Numerical simulations show that this new vacuum chamber structure dominates the high-frequency part of the coupling impedance spectrum and is likely to excite a slow head-tail instability. A simple stability analysis yields the characteristic growth time of the unstable modes. >

Here, an idea of using a visible light wave to accelerate relativistic particles via the inverse FEL mechanism is explored. A strain modulated crystal structure the superlattice, plays the role of a microscopic undulator providing very strong ponderomotive coupling between the beam and the light wave. Purely classical treatment of relativistic protons channeling through a superlattice is performed in a self consistent fashion involving the Maxwell wave equation for the accelerating electromagnetic field and the relativistic Boltzmann equation for the protons. It yields the accelerating efficiency in terms of the negative gain coefficient for the amplitude of the electromagnetic wave-the rate the energy is extracted from the light by the beam. Presented analytic formalism allows one to find the acceleration rate in a simple closed form, which is further evaluated for a model beam-optical cavity system to verify feasibility of this scheme. >

During collider studies it was observed that with bunch intensities of >60*10/sup 9/ particles per bunch in the Tevatron, the beam would go unstable if the machine ran close to the coupling resonance. Simple head-tail stability only requires the horizontal and vertical chromaticities to be positive. Since the beam went unstable even when this condition was met, a set of experiments were performed which showed that if there is significant linear coupling the head-tail stability criterion is modified. A formalism for calculating head-tail stability is presented. The predictions of the formalism are compared to data taken with the Tevatron. >

The authors present the results from beam-based longitudinal impedance measurements in the Tevatron. Longitudinal transfer function scanning with high current unbunched beam at injection is the method used for determining the impedance at every revolution harmonic. Self-bunching instabilities and wave-mixing phenomena occurring during these transfer function measurements are also described. Methods to minimize noise and systematic errors are reviewed. >

While neither the matched nor the unmatched transition jump scheme is entirely satisfactory as presented, either one could be improved and made to work well in the main injector. The unmatched scheme has simpler hardware requirements-only one family of perturbation quadrupoles is required. Betatron functions and tunes are negligibly affected (in an ideal lattice without errors). Its disadvantages stem from the large induced dispersion wave, and from the second-order dependence of Delta gamma /sub t/ on perturbation strength. The matched scheme requires two quadrupole families to keep the betatron tunes unchanged. This leads to a large perturbation strength that, in the worst case of a unipolar Delta gamma /sub t/=-1.3 jump, distorts the linear lattice almost to the point of instability. >

A simple model of a beam at transition driven by a storage ring impedance is formulated in the framework of the nonlinear Vlasov equation. This yields a set of coupled equations of motion describing time evolution of a single coherent mode and the overall equilibrium density distribution function. At transition, contour integration in the dispersion relation can be carried out analytically and a simple closed formula for the coherent frequency is obtained. From the resulting stability diagram further conclusions about the growth time of the microwave instability and the longitudinal emittance blowup at transition are derived. >

A bipolar /spl gamma//sub t/-jump design is reported for the Fermilab Main Injector (Lattice MI-17). The total amount of jump is 1.3 units. Both the betatron and dispersion waves are confined, while the betatron tunes remain nearly unchanged.

The current Booster lattice is studied in the context of full chromaticity compensation in the presence of the sextupole fields generated by the combined function magnets. The sextupole excitation at various energies, found from chromaticity measurements and Booster lattice analysis, was compared with magnetostatic multipole calculations. Both results agree very well and they are consistent with the original design specifications. Two families of correcting sextupole magnets are employed to compensate the sextupole excitations and to adjust the chromaticity (in both planes) to a desired value, which is set by head-tail stability consideration. Analysis of the required correcting sextupole strengths is carried out along the momentum ramp with the measured sextupole excitations of the combined function magnets. The results of our calculation give quantitative insight into the requisite performance of the sextupole magnets. It calls for much stronger sextupole strengths-at the level which can no longer be supported by the present correcting sextupole magnet design.

The Fermilab Main Injector (FMI) is a high intensity proton synchrotron which will be used to accelerate protons and antiprotons from 8.9 GeV to 150 GeV. The natural chromaticities of the machine for the horizontal and the vertical planes are -33.6 and -33.9 respectively. The /spl Delta/p/p of the beam at injection energy of 8.9 GeV is about 0.002. The chromaticity requirements of the FMI are primarily decided by the /spl Delta/p/p=0.002 of the beam at injection. This limits the final chromaticity of the FMI to be /spl plusmn/5 units. To correct the chromaticity in the FMI, two families of sextupole magnets will be installed in the lattice, one for each plane. A sextupole magnet suitable for the FMI needs has been designed and are being built. New chromaticity compensation schemes have been worked out in the light of recently proposed faster acceleration ramps. On a R/D sextupole magnet, the low current measurements have been carried out to determine the electrical properties. Also with a Morgan coil, measurements have been performed to determine the higher ordered multipole components up to 18-poles. An overview of these results are presented here.

The extraction of a small flux of primary protons from a collider in a parasitic manner is very attractive for a number of fixed target applications. The key requirements of the extracted beam are small average currents with relatively small temporal variations. At this time a bent crystal is installed in the Tevatron Collider which is designed to channel protons down an instrumented beam abort transfer line. In this paper the results of experiments with this crystal are presented. The impact on the high energy physics collider detectors of the crystal as a collimator and the transverse mechanism necessary to deposit the protons into the crystal for alignment purposes are described.

This paper presents a biologically plausible mechanism of backpropagation network output error to previous layers of processing in a particular multilayer neural network. This mechanism is used in a network that is designed to mimic familiarity discrimination as performed by the perirhinal cortex of the temporal lobe. In the algorithm, the error of the network during an initial classification period regulates the frequency of neuronal activity in a succeeding memorising period via an inhibitory circuit, such that the frequency in this memorising period is proportional to the error. Synaptic weight modifications are made according to activity-dependent Hebbian rules, such as may be used in the brain. The magnitude of the modification depends on the frequency of the activity. Hence, the magnitude of weight modification is proportional to the network error.

Measurements of differential orbits excited by two pairs of horizontal and vertical correctors, and by a change of the energy gain for the first one of two superconducting linacs allowed us to perform on-line control of the machine optics, which provided valuable information for studying optics discrepancies. Off-line analysis of the data has uncovered a number of malfunctioning hardware pieces, e.g. improperly functioning BPMs and incorrectly focusing quadrupoles. It also indicated that our dipoles have significant focusing terms, which have to be taken into account to build a predictable optics. The analysis resulted in a significantly improved optics model for the CEBAF recirculator. The new optics preserved desired lattice architecture and orthogonal tunability. The presented approach was proven very successful in minimizing required tuning time and in building a more accurate theoretical model of beam transport for the CEBAF accelerator.