- Parallel Computing and Optimization Techniques
- Logic, programming, and type systems
- Genetics and Plant Breeding
- Wheat and Barley Genetics and Pathology
- Software Engineering Research
- Magnetic properties of thin films
- Embedded Systems Design Techniques
- Magnetic Properties of Alloys
- Magnetic Properties and Applications
- Distributed and Parallel Computing Systems
- Intelligence, Security, War Strategy
- Retinal Imaging and Analysis
- Advanced Mathematical Modeling in Engineering
- Graph Theory and Algorithms
- Advanced Vision and Imaging
- Chalcogenide Semiconductor Thin Films
- Video Surveillance and Tracking Methods
- Inorganic Chemistry and Materials
- Distributed systems and fault tolerance
- Matrix Theory and Algorithms
- Quantum Dots Synthesis And Properties
- Rare-earth and actinide compounds
- Cloud Computing and Resource Management
- Electromagnetic Simulation and Numerical Methods
- Advanced optical system design
Centre National de la Recherche Scientifique
2018-2024
Université de Strasbourg
2024
University of Glasgow
2020-2023
Université de Lorraine
2023
University of Idaho
2003-2021
Robert Bosch (Germany)
2014-2019
Carl Zeiss (Germany)
2019
Sorbonne Université
2018
Laboratoire de Recherche en Informatique de Paris 6
2018
Friedrich-Alexander-Universität Erlangen-Nürnberg
2012
In dynamic magnetic resonance imaging (MRI) studies, the motion kinetics or contrast variability are often hard to predict, hampering an appropriate choice of image update rate temporal resolution. A constant azimuthal profile spacing (111.246deg), based on Golden Ratio, is investigated as optimal for reconstruction from arbitrary number profiles in radial MRI. The order evaluated and compared with a uniform distribution terms signal-to-noise ratio (SNR) artifact level. favorable...
This paper describes a computer vision based system for real-time robust traffic sign detection, tracking, and recognition. Such framework is of major interest driver assistance in an intelligent automotive cockpit environment. The proposed approach consists two components. First, signs are detected using set Haar wavelet features obtained from AdaBoost training. Compared to previously published approaches, our solution offers generic, joint modeling color shape information without the need...
The IBM System z10™ microprocessor is currently the fastest running 64-bit CISC (complex instruction set computer) microprocessor. This operates at 4.4 GHz and provides up to two times performance improvement compared with its predecessor, z9® In addition ultrahigh-frequency pipeline, offers such enhancements as a sophisticated branch-prediction structure, large second-level private cache, data-prefetch engine, hardwired decimal floating-point arithmetic unit. z10 also implements new...
The authors have formulated a numerical approach for application to micromagnetics problems and implemented the corresponding three-dimensional finite element code. program has been applied study of hysteretic behavior domain walls in variety samples. discuss method give results its small, rectangular Permalloy particles.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Optimizing programs to run efficiently on modern parallel hardware is hard but crucial for many applications. The predominantly used imperative languages - like C or OpenCL force the programmer intertwine code describing functionality and optimizations. This results in a portability nightmare that particularly problematic given accelerating trend towards specialized devices further increase efficiency. Many emerging DSLs performance demanding domains such as deep learning high-performance...
Rewriting is a principled term transformation technique with uses across theorem proving and compilation. In proving, each rewrite proof step; in compilation, rewrites optimize program term. While developing sequences manually possible, this process does not scale to larger sequences. Automated rewriting techniques, like greedy simplification or equality saturation, work well without requiring human input. Yet, they do large search spaces, limiting the complexity of tasks where automated...
A finite-element method is presented in which the magnetization linearly interpolated within each tetrahedral element and magnetostatic interaction accurately obtained by integration. The equilibrium magnetizations at nodal points can be found minimizing total energy of system. Two different minimization schemes are compared.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
We describe a numerical approach to micromagnetic problems based on the finite element method. discuss convergence difficulty in three dimensions and we propose solution which retains local character of The shape anisotropy each is treated same footing as crystalline anisotropy. Some tests validity our original method two are also described.
The oncogenic properties of murine double minute-2 (mdm2) protein over-expression, which mostly results from the interaction with tumor suppressor p53, are well described and their negative impacts on prognosis affected patients is characterized. However, clinical relevance mdm2 mRNA expression poorly investigated.In this study, 65 soft tissue sarcoma (STS) samples were analyzed for by a quantitative reverse transcription polymerase chain reaction (RT-PCR) approach using available validated...
This paper describes an integrated framework of on-road vehicle detection through knowledge fusion. In contrast to appearance-based detectors that make instant decisions, the proposed fuses appearance, geometry and motion information over multiple image frames. The vehicle/non-vehicle scene is utilized prior models obtained by learning, modeling estimation algorithms. It shown fusion largely improves robustness reliability system.
In this study, we investigate the effect of different elemental selenium to sulfur ratios on chalcopyrite phase formation in Cu(In,Ga)(S,Se)2 thin films. The films are formed by stacked layer process. structural and electronic properties solar cells analyzed means scanning electron microscopy, glow discharge optical emission spectrometry, X-ray diffraction, fluorescence, Raman spectroscopy, spectral photoluminescence as well current-voltage, quantum efficiency measurements. influence...
This article presents and compares optimized implementations of two optical flow algorithms on several target boards comprising multi-core SIMD processors GPUs. The are Horn-Schunck (HS) TV-L1, have been chosen because they both well-known, their different computational complexity accuracy. For algorithms, we made parallel implementations, while HS has also implemented each algorithm, the comparison between versions is carried out in a two-dimensional fashion: terms computing speed-in order...
The recent IBM ESA/390 CMOS line of processors, from 1997 to 1999, consisted the G4, G5, and G6 processors. architecture they implemented lacked 64-bit addressability had only a limited set arithmetic instructions. processors also data instruction bandwidth, since utilized unified cache. branch performance was good, but there were delays due conflicts in searching writing target buffer. Also, hardware compression decimal performance, though demand by database COBOL programmers. Most concerns...
This work presents a robust method of passing vehicle detection. Obstacle detection algorithms that rely on motion estimation tend to be sensitive image outliers caused by structured noise and shadows. To achieve reliable vision system, we have developed two important techniques, with information fusion dynamic scene modeling. By exploiting the uncertainty flow estimates, our scheme gives motion. In addition, also model background foreground dynamics road scenes impose coherency constraints...
Grass weeds are a major problem in winter wheat fields the Pacific Northwest (PNW). Control of these is now enhanced with use imazamox resistant cultivars, which have been rapidly adopted by growers. However, effect spray rate and timing on crop injury agronomic traits cultivars different genetic backgrounds has not adequately evaluated. Thus, experiments were conducted near Moscow Genesee, ID 2003–2004 2004–2005 growing seasons to evaluate four seven breeding lines. Wheat plants treated at...
Halide and many similar projects have demonstrated the great potential of domain specific optimizing compilers. They enable programs to be expressed at a convenient high-level, while generating high-performance code for parallel architectures. As domains interest expand towards deep learning, probabilistic programming beyond, it becomes increasingly clear that is unsustainable redesign compilers each new domain. In addition, rapid growth hardware architectures optimize poses challenges...
Generating high-performance code for diverse hardware and application domains is challenging. Functional array programming languages with patterns like map reduce have been successfully combined term rewriting to define explore optimization spaces. However, deciding what sequence of rewrites apply hard has a huge impact on the performance rewritten program. Equality saturation avoids issue by exploring many possible ways rewrites, efficiently representing equivalent programs in an e-graph...