arXiv subjects

Vincent Génot

Publications and source records attributed to Vincent Génot.

6 recordsLinked to original sources

arXiv2024

Event Detection in Time Series: Universal Deep Learning Approach

Event detection in time series is a challenging task due to the prevalence of imbalanced datasets, rare events, and time interval-defined events. Traditional supervised deep learning methods primarily employ binary classification, where each time step is assigned a binary label indicating the presence or absence of an event. However, these methods struggle to handle these specific scenarios effectively. To address these limitations, we propose a novel supervised regression-based deep learning approach that offers several advantages over classification-based methods. Our approach, with a limited number of parameters, can effectively handle various types of events within a unified framework, including rare events and imbalanced datasets. We provide theoretical justifications for its universality and precision and demonstrate its superior performance across diverse domains, particularly for rare events and imbalanced datasets.

stat.ML↗

arXiv2023

A Comprehensive Python Library for Deep Learning-Based Event Detection in Multivariate Time Series Data and Information Retrieval in NLP

Event detection in time series data is crucial in various domains, including finance, healthcare, cybersecurity, and science. Accurately identifying events in time series data is vital for making informed decisions, detecting anomalies, and predicting future trends. Despite extensive research exploring diverse methods for event detection in time series, with deep learning approaches being among the most advanced, there is still room for improvement and innovation in this field. In this paper, we present a new deep learning supervised method for detecting events in multivariate time series data. Our method combines four distinct novelties compared to existing deep-learning supervised methods. Firstly, it is based on regression instead of binary classification. Secondly, it does not require labeled datasets where each point is labeled; instead, it only requires reference events defined as time points or intervals of time. Thirdly, it is designed to be robust by using a stacked ensemble learning meta-model that combines deep learning models, ranging from classic feed-forward neural networks (FFNs) to state-of-the-art architectures like transformers. This ensemble approach can mitigate individual model weaknesses and biases, resulting in more robust predictions. Finally, to facilitate practical implementation, we have developed a Python package to accompany our proposed method. The package, called eventdetector-ts, can be installed through the Python Package Index (PyPI). In this paper, we present our method and provide a comprehensive guide on the usage of the package. We showcase its versatility and effectiveness through different real-world use cases from natural language processing (NLP) to financial security domains.

cs.LG↗

arXiv2023

Solar-MACH: An open-source tool to analyze solar magnetic connection configurations

The Solar MAgnetic Connection HAUS tool (Solar-MACH) is an open-source tool completely written in Python that derives and visualizes the spatial configuration and solar magnetic connection of different observers (i.e., spacecraft or planets) in the heliosphere at different times. For doing this, the magnetic connection in the interplanetary space is obtained by the classic Parker Heliospheric Magnetic Field (HMF). In close vicinity of the Sun, a Potential Field Source Surface (PFSS) model can be applied to connect the HMF to the solar photosphere. Solar-MACH is especially aimed at providing publication-ready figures for the analyses of Solar Energetic Particle events (SEPs) or solar transients such as Coronal Mass Ejections (CMEs). It is provided as an installable Python package (listed on PyPI and conda-forge), but also as a web tool at solar-mach.github.io that completely runs in any web browser and requires neither Python knowledge nor installation. The development of Solar-MACH is open to everyone and takes place on GitHub, where the source code is publicly available under the BSD 3-Clause License. Established Python libraries like sunpy and pfsspy are utilized to obtain functionalities when possible. In this article, the Python code of Solar-MACH is explained, and its functionality is demonstrated using real science examples. In addition, we introduce the overarching SERPENTINE project, the umbrella under which the recent development took place.

astro-ph.SR↗

arXiv2020

MASER: A Science Ready Toolbox for Low Frequency Radio Astronomy

MASER (Measurements, Analysis, and Simulation of Emission in the Radio range) is a comprehensive infrastructure dedicated to time-dependent low frequency radio astronomy (up to about 50 MHz). The main radio sources observed in this spectral range are the Sun, the magnetized planets (Earth, Jupiter, Saturn), and our Galaxy, which are observed either from ground or space. Ground observatories can capture high resolution data streams with a high sensitivity. Conversely, space-borne instruments can observe below the ionospheric cut-off (at about 10 MHz) and can be placed closer to the studied object. Several tools have been developed in the last decade for sharing space physics data. Data visualization tools developed by various institutes are available to share, display and analyse space physics time series and spectrograms. The MASER team has selected a sub-set of those tools and applied them to low frequency radio astronomy. MASER also includes a Python software library for reading raw data from agency archives.

astro-ph.IM↗

arXiv2020

Relating streamer flows to density and magnetic structures at the Parker Solar Probe

The physical mechanisms that produce the slow solar wind are still highly debated. Parker Solar Probe's (PSP's) second solar encounter provided a new opportunity to relate in situ measurements of the nascent slow solar wind with white-light images of streamer flows. We exploit data taken by the Solar and Heliospheric Observatory (SOHO), the Solar TErrestrial RElations Observatory (STEREO) and the Wide Imager on Solar Probe to reveal for the first time a close link between imaged streamer flows and the high-density plasma measured by the Solar Wind Electrons Alphas and Protons (SWEAP) experiment. We identify different types of slow winds measured by PSP that we relate to the spacecraft's magnetic connectivity (or not) to streamer flows. SWEAP measured high-density and highly variable plasma when PSP was well connected to streamers but more tenuous wind with much weaker density variations when it exited streamer flows. STEREO imaging of the release and propagation of small transients from the Sun to PSP reveals that the spacecraft was continually impacted by the southern edge of streamer transients. The impact of specific density structures is marked by a higher occurrence of magnetic field reversals measured by the FIELDS magnetometers. Magnetic reversals originating from the streamers are associated with larger density variations compared with reversals originating outside streamers. We tentatively interpret these findings in terms of magnetic reconnection between open magnetic fields and coronal loops with different properties, providing support for the formation of a subset of the slow wind by magnetic reconnection.

astro-ph.SR↗

arXiv2006

Comment on "PIC simulations of circularly polarised Alfvén wave phase mixing: A new mechanism for electron acceleration in collisionless plasmas" by Tsiklauri et al

Tsiklauri et al. recently published a theoretical model of electron acceleration by Alfvén waves in a nonuniform collisionless plasmas. We compare their work with a series of results published earlier by an another team, of which Tsiklauri et al. were probably unaware. We show that these two series of works, apparently conducted independently, lead to the same conclusions. This reinforces the theoretical consistency of the model.

astro-ph↗