Nadav Schneider
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Nadav Schneider has a total of 22 co-authors. Here are the top co-authors by number of shared publications:
- Elad Fisher
Technology Division, Soreq NRC, Yavne 81800, Israel
3 shared publications Alimi Roger
2 shared publications- Gal Oren
Stanford University
2 shared publications - Itzik Klein
University of Haifa
2 shared publications - Miki Vizel
Department of Physics, Institute of Nanotechnology and Advanced Materials, Bar-Ilan University, Ramat-Gan 5290002, Israel
2 shared publications - Niranjan Hasabnis
Code Metal, Boston, Massachusetts, USA
2 shared publications
Latest Publications
- Open Access
- 22 July 2026
- Published by Elsevier in SSRN Electronic Journal
Generalized Smartphone-Based Magnetic Mapping with MutualInformation-Guided Deep Learning
Smartphone magnetometers provide a low-cost and widely available sensing platform for near-surface magnetic surveying and geoscientific mapping. Previous studies have demonstrated their potential under controlled acquisition settings. Dense, preplanned surveys remain the standard for producing high-quality magnetic maps, but smartphones create an opportunity to extend data collection toward more flexible acquisition modes, where observations are collected opportunistically by multiple users and devices along irregular, partially overlapping trajectories. Such crowdsourced acquisition may supplement conventional surveying, support rapid coverage of bounded regions of interest, or exploit measurements collected as a by-product of movement through the area. However, transforming these sparse, heterogeneous, and spatially incomplete observations into interpretable magnetic maps constitutes a computational geoscience challenge, especially when reconstruction must remain effective across geographically distinct magnetic backgrounds.To address this problem, we present GSmartMagDL, a generalized framework for crowdsourced smartphone magnetic map reconstruction. The framework combines probabilistic occupancy-grid fusion of multi-trajectory smartphone measurements with a mutual information-guided deep reconstruction network that emphasizes spatially informative regions under sparse and heterogeneous input. We construct a physics-consistent simulation-based dataset containing diverse localized magnetic anomaly patterns, heterogeneous trajectory--device scans, and fused observation maps under different regional geomagnetic backgrounds. The proposed approach is evaluated against Kriging interpolation and a deep-learning reconstruction baseline, including held-out geographic validation without location-specific retraining. Across complementary metrics measuring pointwise accuracy, structural consistency, and perceptual map quality, GSmartMagDL outperforms the baseline approaches and produces reconstructed magnetic maps that better preserve localized magnetic structure, spatial coherence, and robustness across devices, trajectories, and geographically varying magnetic backgrounds. These results indicate that smartphone magnetic mapping can be extended beyond controlled single-device surveys toward practical crowdsourced acquisition while maintaining map features relevant to magnetic interpretation and near-surface geoscience applications.
- 22 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessGeneralized Smartphone-Based Magnetic Mapping with MutualInformation-Guided Deep Learning
Smartphone magnetometers provide a low-cost and widely available sensing platform for near-surface magnetic surveying and geoscientific mapping. Previous studies have demonstrated their potential under controlled acquisition settings. Dense, preplanned surveys remain the standard for producing high-quality magnetic maps, but smartphones create an opportunity to extend data collection toward more flexible acquisition modes, where observations are collected opportunistically by multiple users and devices along irregular, partially overlapping trajectories. Such crowdsourced acquisition may supplement conventional surveying, support rapid coverage of bounded regions of interest, or exploit measurements collected as a by-product of movement through the area. However, transforming these sparse, heterogeneous, and spatially incomplete observations into interpretable magnetic maps constitutes a computational geoscience challenge, especially when reconstruction must remain effective across geographically distinct magnetic backgrounds.To address this problem, we present GSmartMagDL, a generalized framework for crowdsourced smartphone magnetic map reconstruction. The framework combines probabilistic occupancy-grid fusion of multi-trajectory smartphone measurements with a mutual information-guided deep reconstruction network that emphasizes spatially informative regions under sparse and heterogeneous input. We construct a physics-consistent simulation-based dataset containing diverse localized magnetic anomaly patterns, heterogeneous trajectory--device scans, and fused observation maps under different regional geomagnetic backgrounds. The proposed approach is evaluated against Kriging interpolation and a deep-learning reconstruction baseline, including held-out geographic validation without location-specific retraining. Across complementary metrics measuring pointwise accuracy, structural consistency, and perceptual map quality, GSmartMagDL outperforms the baseline approaches and produces reconstructed magnetic maps that better preserve localized magnetic structure, spatial coherence, and robustness across devices, trajectories, and geographically varying magnetic backgrounds. These results indicate that smartphone magnetic mapping can be extended beyond controlled single-device surveys toward practical crowdsourced acquisition while maintaining map features relevant to magnetic interpretation and near-surface geoscience applications.
- Open Access
- 22 July 2026
- Published by Elsevier in SSRN Electronic Journal
Generalized Smartphone-Based Magnetic Mapping with MutualInformation-Guided Deep Learning
Smartphone magnetometers provide a low-cost and widely available sensing platform for near-surface magnetic surveying and geoscientific mapping. Previous studies have demonstrated their potential under controlled acquisition settings. Dense, preplanned surveys remain the standard for producing high-quality magnetic maps, but smartphones create an opportunity to extend data collection toward more flexible acquisition modes, where observations are collected opportunistically by multiple users and devices along irregular, partially overlapping trajectories. Such crowdsourced acquisition may supplement conventional surveying, support rapid coverage of bounded regions of interest, or exploit measurements collected as a by-product of movement through the area. However, transforming these sparse, heterogeneous, and spatially incomplete observations into interpretable magnetic maps constitutes a computational geoscience challenge, especially when reconstruction must remain effective across geographically distinct magnetic backgrounds.To address this problem, we present GSmartMagDL, a generalized framework for crowdsourced smartphone magnetic map reconstruction. The framework combines probabilistic occupancy-grid fusion of multi-trajectory smartphone measurements with a mutual information-guided deep reconstruction network that emphasizes spatially informative regions under sparse and heterogeneous input. We construct a physics-consistent simulation-based dataset containing diverse localized magnetic anomaly patterns, heterogeneous trajectory--device scans, and fused observation maps under different regional geomagnetic backgrounds. The proposed approach is evaluated against Kriging interpolation and a deep-learning reconstruction baseline, including held-out geographic validation without location-specific retraining. Across complementary metrics measuring pointwise accuracy, structural consistency, and perceptual map quality, GSmartMagDL outperforms the baseline approaches and produces reconstructed magnetic maps that better preserve localized magnetic structure, spatial coherence, and robustness across devices, trajectories, and geographically varying magnetic backgrounds. These results indicate that smartphone magnetic mapping can be extended beyond controlled single-device surveys toward practical crowdsourced acquisition while maintaining map features relevant to magnetic interpretation and near-surface geoscience applications.
- 22 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessGeneralized Smartphone-Based Magnetic Mapping with MutualInformation-Guided Deep Learning
Smartphone magnetometers provide a low-cost and widely available sensing platform for near-surface magnetic surveying and geoscientific mapping. Previous studies have demonstrated their potential under controlled acquisition settings. Dense, preplanned surveys remain the standard for producing high-quality magnetic maps, but smartphones create an opportunity to extend data collection toward more flexible acquisition modes, where observations are collected opportunistically by multiple users and devices along irregular, partially overlapping trajectories. Such crowdsourced acquisition may supplement conventional surveying, support rapid coverage of bounded regions of interest, or exploit measurements collected as a by-product of movement through the area. However, transforming these sparse, heterogeneous, and spatially incomplete observations into interpretable magnetic maps constitutes a computational geoscience challenge, especially when reconstruction must remain effective across geographically distinct magnetic backgrounds.To address this problem, we present GSmartMagDL, a generalized framework for crowdsourced smartphone magnetic map reconstruction. The framework combines probabilistic occupancy-grid fusion of multi-trajectory smartphone measurements with a mutual information-guided deep reconstruction network that emphasizes spatially informative regions under sparse and heterogeneous input. We construct a physics-consistent simulation-based dataset containing diverse localized magnetic anomaly patterns, heterogeneous trajectory--device scans, and fused observation maps under different regional geomagnetic backgrounds. The proposed approach is evaluated against Kriging interpolation and a deep-learning reconstruction baseline, including held-out geographic validation without location-specific retraining. Across complementary metrics measuring pointwise accuracy, structural consistency, and perceptual map quality, GSmartMagDL outperforms the baseline approaches and produces reconstructed magnetic maps that better preserve localized magnetic structure, spatial coherence, and robustness across devices, trajectories, and geographically varying magnetic backgrounds. These results indicate that smartphone magnetic mapping can be extended beyond controlled single-device surveys toward practical crowdsourced acquisition while maintaining map features relevant to magnetic interpretation and near-surface geoscience applications.
- Open Access
- 3 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-6
MPIrigen: MPI Code Generation through Domain-Specific Language Models
The imperative need to scale computation across numerous nodes highlights the significance of efficient parallel computing, particularly in the realm of Message Passing Interface (MPI) integration. While MPI serves as a cornerstone for large-scale parallelism, its seamless integration into codebases, especially concerning domain decomposition, has proven challenging. Static tools aimed at addressing this challenge have exhibited limited effectiveness and scalability. On the other hand, contemporary language models designed for programming problems have demonstrated utility in parallel programming tasks such as OpenMP pragma generation. However, the challenging parallel programming task of generating MPI-based parallel programs has remained unexplored. This study first investigates the performance of state-of-the-art language models in generating MPI-based parallel programs. Findings reveal that widely used models such as GPT-3.5 and PolyCoder (specialized multi-lingual code models) exhibit notable performance degradation when generating MPI-based programs compared to general-purpose programs. In contrast, domain-specific models such as MonoCoder, which are pre-trained on MPI-related programming languages of C and C++, outperform larger models. Subsequently, we introduce a dedicated downstream task of MPI-based program generation by fine-tuning MonoCoder on HPCorpusMPI. We call the resulting model as MPIrigen. We propose an innovative preprocessing for completion only after observing the whole code, thus enabling better completion with a wider context. Comparative analysis against GPT-3.5 zero-shot performance, using a novel HPC-oriented evaluation method, demonstrates that MPIrigen excels in generating accurate MPI functions calls. The success of this tailored solution underscores the importance of domain-specific fine-tuning in optimizing language models for parallel computing code generation, paving the way for a new generation of automatic parallelization tools. The sources of this work are available at our GitHub MPIrigen repository.
- 3 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Open AccessMPIrigen: MPI Code Generation through Domain-Specific Language Models
The imperative need to scale computation across numerous nodes highlights the significance of efficient parallel computing, particularly in the realm of Message Passing Interface (MPI) integration. While MPI serves as a cornerstone for large-scale parallelism, its seamless integration into codebases, especially concerning domain decomposition, has proven challenging. Static tools aimed at addressing this challenge have exhibited limited effectiveness and scalability. On the other hand, contemporary language models designed for programming problems have demonstrated utility in parallel programming tasks such as OpenMP pragma generation. However, the challenging parallel programming task of generating MPI-based parallel programs has remained unexplored. This study first investigates the performance of state-of-the-art language models in generating MPI-based parallel programs. Findings reveal that widely used models such as GPT-3.5 and PolyCoder (specialized multi-lingual code models) exhibit notable performance degradation when generating MPI-based programs compared to general-purpose programs. In contrast, domain-specific models such as MonoCoder, which are pre-trained on MPI-related programming languages of C and C++, outperform larger models. Subsequently, we introduce a dedicated downstream task of MPI-based program generation by fine-tuning MonoCoder on HPCorpusMPI. We call the resulting model as MPIrigen. We propose an innovative preprocessing for completion only after observing the whole code, thus enabling better completion with a wider context. Comparative analysis against GPT-3.5 zero-shot performance, using a novel HPC-oriented evaluation method, demonstrates that MPIrigen excels in generating accurate MPI functions calls. The success of this tailored solution underscores the importance of domain-specific fine-tuning in optimizing language models for parallel computing code generation, paving the way for a new generation of automatic parallelization tools. The sources of this work are available at our GitHub MPIrigen repository.
- Open Access
- 12 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 2-10
MPI-RICAL: Data-Driven MPI Distributed Parallelism Assistance with Transformers
Computational science has made rapid progress in recent years, leading to ever increasing demand for supercomputing resources. For scientific applications that leverage such resources, Message Passing Interface (MPI) plays a crucial role in enabling distributed memory parallelization across multiple nodes. However, parallelizing MPI code manually, and specifically, performing domain decomposition, is a challenging and error-prone task. In this paper, we address this problem by developing MPI-rical, a novel data-driven, programming-assistance tool that assists programmers in writing domain decomposition based distributed memory parallelization code using MPI. Specifically, we leverage Transformer architecture — the invention that led to advancements in the field of natural language processing (NLP) — with a supervised language model to suggest MPI functions and their proper locations in the code on the fly. In addition to the novel model for MPI-based parallel programming, in this paper, we also introduce MPICodeCorpus, the first publicly-available corpus of MPI-based parallel programs that is created by mining more than 15,000 open-source repositories on GitHub. Experimental results demonstrate the effectiveness of MPI-rical on both dataset from MPICodeCorpus and more importantly, on a compiled benchmark of MPI-based parallel programs for numerical computations that represent real-world scientific applications. Specifically, MPI-rical achieves F1 scores between 0.87-0.91 on these programs, demonstrating its accuracy in suggesting correct MPI functions at appropriate code locations. The source code used in this work, as well as other relevant sources, are available at: https://github.com/Scientific-Computing-Lab-NRCN/MPI-rical.
- 12 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 2-10
Open AccessMPI-RICAL: Data-Driven MPI Distributed Parallelism Assistance with Transformers
Computational science has made rapid progress in recent years, leading to ever increasing demand for supercomputing resources. For scientific applications that leverage such resources, Message Passing Interface (MPI) plays a crucial role in enabling distributed memory parallelization across multiple nodes. However, parallelizing MPI code manually, and specifically, performing domain decomposition, is a challenging and error-prone task. In this paper, we address this problem by developing MPI-rical, a novel data-driven, programming-assistance tool that assists programmers in writing domain decomposition based distributed memory parallelization code using MPI. Specifically, we leverage Transformer architecture — the invention that led to advancements in the field of natural language processing (NLP) — with a supervised language model to suggest MPI functions and their proper locations in the code on the fly. In addition to the novel model for MPI-based parallel programming, in this paper, we also introduce MPICodeCorpus, the first publicly-available corpus of MPI-based parallel programs that is created by mining more than 15,000 open-source repositories on GitHub. Experimental results demonstrate the effectiveness of MPI-rical on both dataset from MPICodeCorpus and more importantly, on a compiled benchmark of MPI-based parallel programs for numerical computations that represent real-world scientific applications. Specifically, MPI-rical achieves F1 scores between 0.87-0.91 on these programs, demonstrating its accuracy in suggesting correct MPI functions at appropriate code locations. The source code used in this work, as well as other relevant sources, are available at: https://github.com/Scientific-Computing-Lab-NRCN/MPI-rical.
- Open Access
- 29 June 2022
- Published by Cambridge University Press (CUP) in Laser and Particle Beams
- Vol. 2022
Particles Detection System with CR-39 Based on Deep Learning
We present a novel method that we call FAINE, fast artificial intelligence neutron detection system. FAINE automatically classifies tracks of fast neutrons on CR-39 detectors using a deep learning model. This method was demonstrated using a LANDAUER Neutrak® fast neutron dosimetry system, which is installed in the External Dosimetry Laboratory (EDL) at Soreq Nuclear Research Center (SNRC). In modern fast neutron dosimetry systems, after the preliminary stages of etching and imaging of the CR-39 detectors, the third stage uses various types of computer vision systems combined with a manual revision to count the CR-39 tracks and then convert them to a dose in mSv units. Our method enhances these modern systems by introducing an innovative algorithm, which uses deep learning to classify all CR-39 tracks as either real neutron tracks or any other sign such as dirt, scratches, or even cleaning remainders. This new algorithm makes the third stage of manual CR-39 tracks revision superfluous and provides a completely repeatable and accurate way of measuring either neutrons flux or dose. The experimental results show a total accuracy rate of 96.7% for the true positive tracks and true negative tracks detected by our new algorithm against the current method, which uses computer vision followed by manual revision. This algorithm is now in the process of calibration for both alpha-particles detection and fast neutron spectrometry classification and is expected to be very useful in analyzing results of proton-boron11 fusion experiments. Being fully automatic, the new algorithm will enhance the quality assurance and effectiveness of external dosimetry, will lower the uncertainty for the reported dose measurements, and might also enable lowering the system’s detection threshold.
- 29 June 2022
- Published by Cambridge University Press (CUP) in Laser and Particle Beams
- Vol. 2022
Open AccessParticles Detection System with CR-39 Based on Deep Learning
We present a novel method that we call FAINE, fast artificial intelligence neutron detection system. FAINE automatically classifies tracks of fast neutrons on CR-39 detectors using a deep learning model. This method was demonstrated using a LANDAUER Neutrak® fast neutron dosimetry system, which is installed in the External Dosimetry Laboratory (EDL) at Soreq Nuclear Research Center (SNRC). In modern fast neutron dosimetry systems, after the preliminary stages of etching and imaging of the CR-39 detectors, the third stage uses various types of computer vision systems combined with a manual revision to count the CR-39 tracks and then convert them to a dose in mSv units. Our method enhances these modern systems by introducing an innovative algorithm, which uses deep learning to classify all CR-39 tracks as either real neutron tracks or any other sign such as dirt, scratches, or even cleaning remainders. This new algorithm makes the third stage of manual CR-39 tracks revision superfluous and provides a completely repeatable and accurate way of measuring either neutrons flux or dose. The experimental results show a total accuracy rate of 96.7% for the true positive tracks and true negative tracks detected by our new algorithm against the current method, which uses computer vision followed by manual revision. This algorithm is now in the process of calibration for both alpha-particles detection and fast neutron spectrometry classification and is expected to be very useful in analyzing results of proton-boron11 fusion experiments. Being fully automatic, the new algorithm will enhance the quality assurance and effectiveness of external dosimetry, will lower the uncertainty for the reported dose measurements, and might also enable lowering the system’s detection threshold.