arXiv subjects

Tristan Glatard

Publications and source records attributed to Tristan Glatard.

At least 19 recordsLinked to original sources

arXiv2026

Uncertain but Useful: Leveraging CNN Training Variability into Data Augmentation

Deep learning (DL) has transformed neuroimaging by delivering state-of-the-art performance with reduced computation times. Yet, the numerical uncertainty inherent to DL training remains largely underexplored despite its potential to significantly impact the reliability of model outcomes. We show that training the FastSurfer segmentation model introduces substantial numerical uncertainty that exceeds its non-DL counterpart (FreeSurfer 7.3.2) in cortical regions, potentially impacting downstream clinical results. We also characterize this training-time uncertainty using random seed perturbations and demonstrate that seed-induced variability is structurally comparable to numerical variability. We then show that seed variability can be leveraged as a data augmentation technique through ensembling to improve downstream brain age regression performance. These findings position numerical uncertainty during DL training as a substantive factor in neuroimaging reliability, with measurable consequences for downstream tasks, and demonstrate that it can simultaneously be harnessed as a data augmentation technique.

math.NA↗

arXiv2026

Fuzzy PyTorch: Rapid Numerical Variability Evaluation for Deep Learning Models

We introduce Fuzzy PyTorch, a framework for rapid evaluation of numerical variability in deep learning (DL) models. As DL is increasingly applied to diverse tasks, understanding variability from floating-point arithmetic is essential to ensure robust and reliable performance. Tools assessing such variability must be scalable, efficient, and integrate seamlessly with existing frameworks while minimizing code modifications. Fuzzy PyTorch enables this by integrating stochastic arithmetic into PyTorch through Probabilistic Rounding with Instruction Set Management, a novel library interfacing with Verificarlo, a numerical analysis compiler. The library offers stochastic rounding mode and a novel mode; up-down rounding. Comparative evaluations show Fuzzy PyTorch maintains model performance and achieves runtime reductions of 5x to 60x versus Verrou, a state-of-the-art tool. We further demonstrate scalability by running models from 1 to 341 million parameters, confirming applicability across small and large DL architectures. Overall, Fuzzy PyTorch provides an efficient, scalable, and practical solution for assessing numerical variability in deep learning, enabling researchers and practitioners to quantify and manage floating-point uncertainty without compromising performance or computational efficiency.

cs.LG↗

arXiv2026

Conservative & Aggressive NaNs Accelerate U-Nets for Neuroimaging

Deep learning models for neuroimaging increasingly rely on large architectures, making efficiency a persistent concern despite advances in hardware. Through an analysis of numerical uncertainty of convolutional neural networks (CNNs), we observe that many operations are applied to values dominated by numerical noise and have negligible influence on model outputs. In some models, up to two-thirds of convolution operations appear redundant. We introduce Conservative & Aggressive NaNs, two novel variants of max pooling and unpooling that identify numerically unstable voxels and replace them with NaNs, allowing subsequent layers to skip computations on irrelevant data. Both methods are implemented within PyTorch and require no architectural changes. We evaluate these approaches on four CNN models spanning neuroimaging and image classification tasks. For inputs containing at least 50% NaNs, we observe consistent runtime improvements; for data with more than two-thirds NaNs )common in several neuroimaging settings) we achieve an average inference speedup of 1.67x. Conservative NaNs reduces convolution operations by an average of 30% across models and datasets, with no measurable performance degradation, and can skip up to 64.64% of convolutions in specific layers. Aggressive NaNs can skip up to 69.30% of convolutions but may occasionally affect performance. Overall, these methods demonstrate that numerical uncertainty can be exploited to reduce redundant computation and improve inference efficiency in CNNs.

cs.LG↗

arXiv2026

LLM-Augmented Knowledge Base Construction For Root Cause Analysis

Communications networks now form the backbone of our digital world, with fast and reliable connectivity. However, even with appropriate redundancy and failover mechanisms, it is difficult to guarantee "five 9s" (99.999 %) reliability, requiring rapid and accurate root cause analysis (RCA) during outages. In the event of an outage, rapid and accurate RCA becomes essential to restore service and prevent future disruptions. This study evaluates three Large Language Model (LLM) methodologies - Fine-Tuning, RAG, and a Hybrid approach - for constructing a Root Cause Analysis (RCA) Knowledge Base from support tickets. We compare their performance using a comprehensive suite of lexical and semantic similarity metrics. Our experiments on a real industrial dataset demonstrate that the generated knowledge base provides an excellent starting point for accelerating RCA tasks and improving network resilience.

cs.CL↗

arXiv2025

Numerical Uncertainty in Linear Registration: An Experimental Study

While linear registration is a critical step in MRI preprocessing pipelines, its numerical uncertainty is understudied. Using Monte-Carlo Arithmetic (MCA) simulations, we assessed the most commonly used linear registration tools within major software packages (SPM, FSL, and ANTs) across multiple image similarity measures, two brain templates, and both healthy control (HC, n=50) and Parkinson's Disease (PD, n=50) cohorts. Our findings highlight the influence of linear registration tools and similarity measures on numerical stability. Among the evaluated tools and with default similarity measures, SPM exhibited the highest stability. FSL and ANTs showed greater and similar ranges of variability, with ANTs demonstrating particular sensitivity to numerical perturbations that occasionally led to registration failure. Furthermore, no significant differences were observed between healthy and PD cohorts, suggesting that numerical stability analyses obtained with healthy subjects may generalise to clinical populations. Finally, we also demonstrated how numerical uncertainty measures may support automated quality control (QC) of linear registration results. Overall, our experimental results characterize the numerical stability of linear registration experimentally and can serve as a basis for future uncertainty analyses.

q-bio.QM↗

arXiv2025

Predicting Parkinson's disease trajectory using clinical and functional MRI features: a reproduction and replication study

Parkinson's disease (PD) is a common neurodegenerative disorder with a poorly understood physiopathology and no established biomarkers for the diagnosis of early stages and for prediction of disease progression. Several neuroimaging biomarkers have been studied recently, but these are susceptible to several sources of variability related for instance to cohort selection or image analysis. In this context, an evaluation of the robustness of such biomarkers to variations in the data processing workflow is essential. This study is part of a larger project investigating the replicability of potential neuroimaging biomarkers of PD. Here, we attempt to reproduce (re-implementing the experiments with the same data, same method) and replicate (different data and/or method) the models described in [1] to predict individual's PD current state and progression using demographic, clinical and neuroimaging features (fALFF and ReHo extracted from resting-state fMRI). We use the Parkinson's Progression Markers Initiative dataset (PPMI, ppmi-info.org), as in [1] and aim to reproduce the original cohort, imaging features and machine learning models as closely as possible using the information available in the paper and the code. We also investigated methodological variations in cohort selection, feature extraction pipelines and sets of input features. Different criteria were used to evaluate the reproduction and compare the reproduced results with the original ones. Notably, we obtained significantly better than chance performance using the analysis pipeline closest to that in the original study (R2 \> 0), which is consistent with its findings. Moreover, using derived data provided by the authors of the original study, we were able to make an exact reproduction and managed to obtain results that were close to the original ones. The challenges encountered while reproducing and replicating the original work are likely explained by the complexity of neuroimaging studies, in particular in clinical settings. We provide recommendations to further facilitate the reproducibility of such studies in the future.

q-bio.NC↗

arXiv2024

Training Compute-Optimal Vision Transformers for Brain Encoding

The optimal training of a vision transformer for brain encoding depends on three factors: model size, data size, and computational resources. This study investigates these three pillars, focusing on the effects of data scaling, model scaling, and high-performance computing on brain encoding results. Using VideoGPT to extract efficient spatiotemporal features from videos and training a Ridge model to predict brain activity based on these features, we conducted benchmark experiments with varying data sizes (10k, 100k, 1M, 6M) and different model configurations of GPT-2, including hidden layer dimensions, number of layers, and number of attention heads. We also evaluated the effects of training models with 32-bit vs 16-bit floating point representations. Our results demonstrate that increasing the hidden layer dimensions significantly improves brain encoding performance, as evidenced by higher Pearson correlation coefficients across all subjects. In contrast, the number of attention heads does not have a significant effect on the encoding results. Additionally, increasing the number of layers shows some improvement in brain encoding correlations, but the trend is not as consistent as that observed with hidden layer dimensions. The data scaling results show that larger training datasets lead to improved brain encoding performance, with the highest Pearson correlation coefficients observed for the largest dataset size (6M). These findings highlight that the effects of data scaling are more significant compared to model scaling in enhancing brain encoding performance. Furthermore, we explored the impact of floating-point precision by comparing 32-bit and 16-bit representations. Training with 16-bit precision yielded the same brain encoding accuracy as 32-bit, while reducing training time by 1.17 times, demonstrating its efficiency for high-performance computing tasks.

eess.IV↗

arXiv2024

Performance comparison of Dask and Apache Spark on HPC systems for Neuroimaging

The general increase in data size and data sharing motivates the adoption of Big Data strategies in several scientific disciplines. However, while several options are available, no particular guidelines exist for selecting a Big Data engine. In this paper, we compare the runtime performance of two popular Big Data engines with Python APIs, Apache Spark, and Dask, in processing neuroimaging pipelines. Our experiments use three synthetic \HL{neuroimaging} applications to process the \SI{606}{\gibi\byte} BigBrain image and an actual pipeline to process data from thousands of anatomical images. We benchmark these applications on a dedicated HPC cluster running the Lustre file system while using varying combinations of the number of nodes, file size, and task duration. Our results show that although there are slight differences between Dask and Spark, the performance of the engines is comparable for data-intensive applications. However, Spark requires more memory than Dask, which can lead to slower runtime depending on configuration and infrastructure. In general, the limiting factor was the data transfer time. While both engines are suitable for neuroimaging, more efforts need to be put to reduce the data transfer time and the memory footprint of applications.

cs.DC↗

arXiv2024

An Analysis of Performance Bottlenecks in MRI Pre-Processing

Magnetic Resonance Image (MRI) pre-processing is a critical step for neuroimaging analysis. However, the computational cost of MRI pre-processing pipelines is a major bottleneck for large cohort studies and some clinical applications. While High-Performance Computing (HPC) and, more recently, Deep Learning have been adopted to accelerate the computations, these techniques require costly hardware and are not accessible to all researchers. Therefore, it is important to understand the performance bottlenecks of MRI pre-processing pipelines to improve their performance. Using Intel VTune profiler, we characterized the bottlenecks of several commonly used MRI-preprocessing pipelines from the ANTs, FSL, and FreeSurfer toolboxes. We found that few functions contributed to most of the CPU time, and that linear interpolation was the largest contributor. Data access was also a substantial bottleneck. We identified a bug in the ITK library that impacts the performance of ANTs pipeline in single-precision and a potential issue with the OpenMP scaling in FreeSurfer recon-all. Our results provide a reference for future efforts to optimize MRI pre-processing pipelines.

cs.PF↗

arXiv2024

Hierarchical storage management in user space for neuroimaging applications

Neuroimaging open-data initiatives have led to increased availability of large scientific datasets. While these datasets are shifting the processing bottleneck from compute-intensive to data-intensive, current standardized analysis tools have yet to adopt strategies that mitigate the costs associated with large data transfers. A major challenge in adapting neuroimaging applications for data-intensive processing is that they must be entirely rewritten. To facilitate data management for standardized neuroimaging tools, we developed Sea, a library that intercepts and redirects application read and write calls to minimize data transfer time. In this paper, we investigate the performance of Sea on three preprocessing pipelines implemented using standard toolboxes (FSL, SPM and AFNI), using three neuroimaging datasets of different sizes (OpenNeuro's ds001545, PREVENT-AD and the HCP dataset) on two high-performance computing clusters. Our results demonstrate that Sea provides large speedups (up to 32X) when the shared file system's (e.g. Lustre) performance is deteriorated. When the shared file system is not overburdened by other users, performance is unaffected by Sea, suggesting that Sea's overhead is minimal even in cases where its benefits are limited. Overall, Sea is beneficial, even when performance gain is minimal, as it can be used to limit the number of files created on parallel file systems.

cs.DC↗

arXiv2024

Scaling up ridge regression for brain encoding in a massive individual fMRI dataset

Brain encoding with neuroimaging data is an established analysis aimed at predicting human brain activity directly from complex stimuli features such as movie frames. Typically, these features are the latent space representation from an artificial neural network, and the stimuli are image, audio, or text inputs. Ridge regression is a popular prediction model for brain encoding due to its good out-of-sample generalization performance. However, training a ridge regression model can be highly time-consuming when dealing with large-scale deep functional magnetic resonance imaging (fMRI) datasets that include many space-time samples of brain activity. This paper evaluates different parallelization techniques to reduce the training time of brain encoding with ridge regression on the CNeuroMod Friends dataset, one of the largest deep fMRI resource currently available. With multi-threading, our results show that the Intel Math Kernel Library (MKL) significantly outperforms the OpenBLAS library, being 1.9 times faster using 32 threads on a single machine. We then evaluated the Dask multi-CPU implementation of ridge regression readily available in scikit-learn (MultiOutput), and we proposed a new "batch" version of Dask parallelization, motivated by a time complexity analysis. In line with our theoretical analysis, MultiOutput parallelization was found to be impractical, i.e., slower than multi-threading on a single machine. In contrast, the Batch-MultiOutput regression scaled well across compute nodes and threads, providing speed-ups of up to 33 times with 8 compute nodes and 32 threads compared to a single-threaded scikit-learn execution. Batch parallelization using Dask thus emerges as a scalable approach for brain encoding with ridge regression on high-performance computing systems using scikit-learn and large fMRI datasets.

cs.LG↗

arXiv2024

Numerical Stability of DeepGOPlus Inference

Convolutional neural networks (CNNs) are currently among the most widely-used deep neural network (DNN) architectures available and achieve state-of-the-art performance for many problems. Originally applied to computer vision tasks, CNNs work well with any data with a spatial relationship, besides images, and have been applied to different fields. However, recent works have highlighted numerical stability challenges in DNNs, which also relates to their known sensitivity to noise injection. These challenges can jeopardise their performance and reliability. This paper investigates DeepGOPlus, a CNN that predicts protein function. DeepGOPlus has achieved state-of-the-art performance and can successfully take advantage and annotate the abounding protein sequences emerging in proteomics. We determine the numerical stability of the model's inference stage by quantifying the numerical uncertainty resulting from perturbations of the underlying floating-point data. In addition, we explore the opportunity to use reduced-precision floating point formats for DeepGOPlus inference, to reduce memory consumption and latency. This is achieved by instrumenting DeepGOPlus' execution using Monte Carlo Arithmetic, a technique that experimentally quantifies floating point operation errors and VPREC, a tool that emulates results with customizable floating point precision formats. Focus is placed on the inference stage as it is the primary deliverable of the DeepGOPlus model, widely applicable across different environments. All in all, our results show that although the DeepGOPlus CNN is very stable numerically, it can only be selectively implemented with lower-precision floating-point formats. We conclude that predictions obtained from the pre-trained DeepGOPlus model are very reliable numerically, and use existing floating-point formats efficiently.

cs.LG↗

arXiv2023

Classification of Anomalies in Telecommunication Network KPI Time Series

The increasing complexity and scale of telecommunication networks have led to a growing interest in automated anomaly detection systems. However, the classification of anomalies detected on network Key Performance Indicators (KPI) has received less attention, resulting in a lack of information about anomaly characteristics and classification processes. To address this gap, this paper proposes a modular anomaly classification framework. The framework assumes separate entities for the anomaly classifier and the detector, allowing for a distinct treatment of anomaly detection and classification tasks on time series. The objectives of this study are (1) to develop a time series simulator that generates synthetic time series resembling real-world network KPI behavior, (2) to build a detection model to identify anomalies in the time series, (3) to build classification models that accurately categorize detected anomalies into predefined classes (4) to evaluate the classification framework performance on simulated and real-world network KPI time series. This study has demonstrated the good performance of the anomaly classification models trained on simulated anomalies when applied to real-world network time series data.

cs.LG↗

arXiv2023

Mondrian Forest for Data Stream Classification Under Memory Constraints

Supervised learning algorithms generally assume the availability of enough memory to store their data model during the training and test phases. However, in the Internet of Things, this assumption is unrealistic when data comes in the form of infinite data streams, or when learning algorithms are deployed on devices with reduced amounts of memory. In this paper, we adapt the online Mondrian forest classification algorithm to work with memory constraints on data streams. In particular, we design five out-of-memory strategies to update Mondrian trees with new data points when the memory limit is reached. Moreover, we design trimming mechanisms to make Mondrian trees more robust to concept drifts under memory constraints. We evaluate our algorithms on a variety of real and simulated datasets, and we conclude with recommendations on their use in different situations: the Extend Node strategy appears as the best out-of-memory strategy in all configurations, whereas different trimming mechanisms should be adopted depending on whether a concept drift is expected. All our methods are implemented in the OrpailleCC open-source library and are ready to be used on embedded systems and connected objects.

cs.LG↗

arXiv2023

Numerical Uncertainty of Convolutional Neural Networks Inference for Structural Brain MRI Analysis

This paper investigates the numerical uncertainty of Convolutional Neural Networks (CNNs) inference for structural brain MRI analysis. It applies Random Rounding -- a stochastic arithmetic technique -- to CNN models employed in non-linear registration (SynthMorph) and whole-brain segmentation (FastSurfer), and compares the resulting numerical uncertainty to the one measured in a reference image-processing pipeline (FreeSurfer recon-all). Results obtained on 32 representative subjects show that CNN predictions are substantially more accurate numerically than traditional image-processing results (non-linear registration: 19 vs 13 significant bits on average; whole-brain segmentation: 0.99 vs 0.92 Sørensen-Dice score on average), which suggests a better reproducibility of CNN results across execution environments.

eess.IV↗

arXiv2023

A numerical variability approach to results stability tests and its application to neuroimaging

Ensuring the long-term reproducibility of data analyses requires results stability tests to verify that analysis results remain within acceptable variation bounds despite inevitable software updates and hardware evolutions. This paper introduces a numerical variability approach for results stability tests, which determines acceptable variation bounds using random rounding of floating-point calculations. By applying the resulting stability test to \fmriprep, a widely-used neuroimaging tool, we show that the test is sensitive enough to detect subtle updates in image processing methods while remaining specific enough to accept numerical variations within a reference version of the application. This result contributes to enhancing the reliability and reproducibility of data analyses by providing a robust and flexible method for stability testing.

physics.med-ph↗

arXiv2022

Dynamic Ensemble Size Adjustment for Memory Constrained Mondrian Forest

Supervised learning algorithms generally assume the availability of enough memory to store data models during the training and test phases. However, this assumption is unrealistic when data comes in the form of infinite data streams, or when learning algorithms are deployed on devices with reduced amounts of memory. Such memory constraints impact the model behavior and assumptions. In this paper, we show that under memory constraints, increasing the size of a tree-based ensemble classifier can worsen its performance. In particular, we experimentally show the existence of an optimal ensemble size for a memory-bounded Mondrian forest on data streams and we design an algorithm to guide the forest toward that optimal number by using an estimation of overfitting. We tested different variations for this algorithm on a variety of real and simulated datasets, and we conclude that our method can achieve up to 95% of the performance of an optimally-sized Mondrian forest for stable datasets, and can even outperform it for datasets with concept drifts. All our methods are implemented in the OrpailleCC open-source library and are ready to be used on embedded systems and connected objects.

cs.LG↗

arXiv2022

Sea: A lightweight data-placement library for Big Data scientific computing

The recent influx of open scientific data has contributed to the transitioning of scientific computing from compute intensive to data intensive. Whereas many Big Data frameworks exist that minimize the cost of data transfers, few scientific applications integrate these frameworks or adopt data-placement strategies to mitigate the costs. Scientific applications commonly rely on well-established command-line tools that would require complete reinstrumentation in order to incorporate existing frameworks. We developed Sea as a means to enable data-placement strategies for scientific applications executing on HPC clusters without the need to reinstrument workflows. Sea leverages GNU C library interception to intercept POSIX-compliant file system calls made by the applications. We designed a performance model and evaluated the performance of Sea on a synthetic data-intensive application processing a representative neuroimaging dataset (the Big Brain). Our results demonstrate that Sea significantly improves performance, up to a factor of 3$\times$.

cs.DC↗