arXiv subjects

Christophe Guyeux

Publications and source records attributed to Christophe Guyeux.

At least 19 recordsLinked to original sources

arXiv2026

Data-Driven Fire-Zone Segmentation for Improved Short-Term Wildfire Prediction

Wildfire prediction models typically discretize study areas into uniform grids, ignoring the heterogeneous spatial distribution of ignitions. We challenge this paradigm by showing that how data is discretized matters more than which model is used. We propose an unsupervised fire-zone segmentation algorithm combining watershed detection with K-means clustering to define prediction units directly from historical fire patterns. Experiments across six French departments and six forecasting models show that fire-zone segmentation consistently outperforms grid-based approaches, with mean IoU improvements of +3--6% depending on spatial scale. The method is computationally lightweight (<10s per configuration) and fully parallelizable. Our results demonstrate that optimizing spatial discretization yields significant, reproducible performance gains for short-term wildfire forecasting.

cs.LG↗

arXiv2026

Risk Is Not the Target: A Monotonic Framework for Evaluating Wildfire Operational Risk Signals

Evaluating wildfire risk systems using standard machine-learning metrics such as F1-score or IoU is fundamentally flawed: these metrics assess event prediction accuracy, not the operational coherence of a continuous risk signal. This work proposes a novel monotonic evaluation framework that measures whether increases in a predicted risk score consistently correspond to increases in observed operational load, such as number of fires, intervention time, and deployed resources. Moreover, we compare three structurally different approaches on the French Alpes-Maritimes department: the expert-based DFE index, GRU- based predictive models, and FARS, a hybrid multi-agent system combining predictive AI with LLM-based reasoning. Experimental results reveal that the DFE, despite poor classification metrics, exhibits the most balanced monotonic behavior across the full risk scale. GRU models achieve strong local monotonicity but fail to produce well-distributed risk levels. FARS inherits and reveals the structural limitations of upstream signals rather than correcting them. The central finding is a paradigm shift: a good risk model does not predict fires accurately, but one whose ordinal scale meaningfully explains operational dynamics, as proved in this paper. Code of the monotonic framework is available on github.

cs.AI↗

arXiv2026

Proof of Concept: Multi-Target Wildfire Risk Prediction and Large Language Model Synthesis

Current state-of-the-art approaches to wildfire risk assessment often overlook operational needs, limiting their practical value for first responders and firefighting services. Effective wildfire management requires a multi-target analysis that captures the diverse dimensions of wildfire risk, including meteorological danger, ignition activity, intervention complexity, and resource mobilization, rather than relying on a single predictive indicator. In this proof of concept, we propose the development of a hybrid framework that combines predictive models for each risk dimension with large language models (LLMs) to synthesize heterogeneous outputs into structured, actionable reports.

cs.LG↗

arXiv2026

Extreme-value forest fire prediction A study of the Loss Function in an Ordinality Scheme

Wildfires are highly imbalanced natural hazards in both space and severity, making the prediction of extreme events particularly challenging. In this work, we introduce the first ordinal classification framework for forecasting wildfire severity levels directly aligned with operational decision-making in France. Our study investigates the influence of loss-function design on the ability of neural models to predict rare yet critical high-severity fire occurrences. We compare standard cross-entropy with several ordinal-aware objectives, including the proposed probabilistic TDeGPD loss derived from a truncated discrete exponentiated Generalized Pareto Distribution. Through extensive benchmarking over multiple architectures and real operational data, we show that ordinal supervision substantially improves model performance over conventional approaches. In particular, the Weighted Kappa Loss (WKLoss) achieves the best overall results, with more than +0.1 IoU (Intersection Over Union) gain on the most extreme severity classes while maintaining competitive calibration quality. However, performance remains limited for the rarest events due to their extremely low representation in the dataset. These findings highlight the importance of integrating both severity ordering, data imbalance considerations, and seasonality risk into wildfire forecasting systems. Future work will focus on incorporating seasonal dynamics and uncertainty information into training to further improve the reliability of extreme-event prediction.

cs.LG↗

arXiv2025

Localized Forest Fire Risk Prediction: A Department-Aware Approach for Operational Decision Support

Forest fire prediction involves estimating the likelihood of fire ignition or related risk levels in a specific area over a defined time period. With climate change intensifying fire behavior and frequency, accurate prediction has become one of the most pressing challenges in Artificial Intelligence (AI). Traditionally, fire ignition is approached as a binary classification task in the literature. However, this formulation oversimplifies the problem, especially from the perspective of end-users such as firefighters. In general, as is the case in France, firefighting units are organized by department, each with its terrain, climate conditions, and historical experience with fire events. Consequently, fire risk should be modeled in a way that is sensitive to local conditions and does not assume uniform risk across all regions. This paper proposes a new approach that tailors fire risk assessment to departmental contexts, offering more actionable and region-specific predictions for operational use. With this, we present the first national-scale AI benchmark for metropolitan France using state-of-the-art AI models on a relatively unexplored dataset. Finally, we offer a summary of important future works that should be taken into account. Supplementary materials are available on GitHub.

cs.LG↗

arXiv2021

Constructing Higher-Dimensional Digital Chaotic Systems via Loop-State Contraction Algorithm

In recent years, the generation of rigorously provable chaos in finite precision digital domain has made a lot of progress in theory and practice, this article is a part of it. It aims to improve and expand the theoretical and application framework of higher-dimensional digital chaotic system (HDDCS). In this study, topological mixing for HDDCS is strictly proved theoretically at first. Topological mixing implies Devaney's definition of chaos in compact space, but not vise versa. Therefore, the proof of topological mixing improves and expands the theoretical framework of HDDCS. Then, a general design method for constructing HDDCS via loop-state contraction algorithm is given. The construction of the iterative function uncontrolled by random sequences (hereafter called iterative function) is the starting point of this research. On this basis, this article put forward a general design method to solve the problem of HDDCS construction, and a number of examples illustrate the effectiveness and feasibility of this method. The adjacency matrix corresponding to the designed HDDCS is used to construct the chaotic echo state network (ESN) for the Mackey-Glass time series prediction. The prediction accuracy is improved with the increase of the size of the reservoir and the dimension of HDDCS, indicating that higher-dimensional systems have better prediction performance than lower-dimensional systems.

nlin.CD↗

arXiv2018

Online shortest paths with confidence intervals for routing in a time varying random network

The increase in the world's population and rising standards of living is leading to an ever-increasing number of vehicles on the roads, and with it ever-increasing difficulties in traffic management. This traffic management in transport networks can be clearly optimized by using information and communication technologies referred as Intelligent Transport Systems (ITS). This management problem is usually reformulated as finding the shortest path in a time varying random graph. In this article, an online shortest path computation using stochastic gradient descent is proposed. This routing algorithm for ITS traffic management is based on the online Frank-Wolfe approach. Our improvement enables to find a confidence interval for the shortest path, by using the stochastic gradient algorithm for approximate Bayesian inference.

cs.DS↗

arXiv2018

Average performance analysis of the stochastic gradient method for online PCA

This paper studies the complexity of the stochastic gradient algorithm for PCA when the data are observed in a streaming setting. We also propose an online approach for selecting the learning rate. Simulation experiments confirm the practical relevance of the plain stochastic gradient approach and that drastic improvements can be achieved by learning the learning rate.

math.ST↗

arXiv2017

On the topology effects in wireless sensor networks based prognostics and health management

In this work, we consider the usage of wireless sensor networks (WSN) to monitor an area of interest, in order to diagnose on real time its state. Each sensor node forwards information about relevant features towards the sink where the data is processed. Nevertheless, energy conservation is a key issue in the design of such networks and once a sensor exhausts its resources, it will be dropped from the network. This will lead to broken links and data loss. It is therefore important to keep the network running for as long as possible by preserving the energy held by the nodes. Indeed, saving the quality of service (QoS) of a wireless sensor network for a long period is very important in order to ensure accurate data. Then, the area diagnosing will be more accurate. From another side, packet transmission is the phase that consumes the highest amount of energy comparing to other activities in the network. Therefore, we can see that the network topology has an important impact on energy efficiency, and thus on data and diagnosis accuracies. In this paper, we study and compare four network topologies: distributed, hierarchical, centralized, and decentralized topology and show their impact on the resulting estimation of diagnostics. We have used six diagnostic algorithms, to evaluate both prognostic and health management with the variation of type of topology in WSN.

cs.DC↗

arXiv2017

An improved watermarking scheme for Internet applications

In this paper, a data hiding scheme ready for Internet applications is proposed. An existing scheme based on chaotic iterations is improved, to respond to some major Internet security concerns, such as digital rights management, communication over hidden channels, and social search engines. By using Reed Solomon error correcting codes and wavelets domain, we show that this data hiding scheme can be improved to solve issues and requirements raised by these Internet fields.

cs.MM↗

arXiv2017

Diffusion and confusion of chaotic iteration based hash functions

To guarantee the integrity and security of data transmitted through the Internet, hash functions are fundamental tools. But recent researches have shown that security flaws exist in the most widely used hash functions. So a new way to improve their security performance is urgently demanded. In this article, we propose new hash functions based on chaotic iterations, which have chaotic properties as defined by Devaney. The corresponding diffusion and confusion analyzes are provided and a comparative study between the proposed hash functions is carried out, to make their use more applicable in any security context.

nlin.CD↗

arXiv2017

Summary of Topological Study of Chaotic CBC Mode of Operation

In cryptography, block ciphers are the most fundamental elements in many symmetric-key encryption systems. The Cipher Block Chaining, denoted CBC, presents one of the most famous mode of operation that uses a block cipher to provide confidentiality or authenticity. In this research work, we intend to summarize our results that have been detailed in our previous series of articles. The goal of this series has been to obtain a complete topological study of the CBC block cipher mode of operation after proving his chaotic behavior according to the reputed definition of Devaney.

nlin.CD↗

arXiv2017

A Cryptographic Approach for Steganography

In this research work, security concepts are formalized in steganography, and the common paradigms based on information theory are replaced by another ones inspired from cryptography, more practicable are closer than what is usually done in other cryptographic domains. These preliminaries lead to a first proof of a cryptographically secure information hiding scheme.

cs.CR↗

arXiv2017

A Robust Data Hiding Process Contributing to the Development of a Semantic Web

In this paper, a novel steganographic scheme based on chaotic iterations is proposed. This research work takes place into the information hiding framework, and focus more specifically on robust steganography. Steganographic algorithms can participate in the development of a semantic web: medias being on the Internet can be enriched by information related to their contents, authors, etc., leading to better results for the search engines that can deal with such tags. As media can be modified by users for various reasons, it is preferable that these embedding tags can resist to changes resulting from some classical transformations as for example cropping, rotation, image conversion, and so on. This is why a new robust watermarking scheme for semantic search engines is proposed in this document. For the sake of completeness, the robustness of this scheme is finally compared to existing established algorithms.

cs.MM↗

arXiv2017

An optimization technique on pseudorandom generators based on chaotic iterations

Internet communication systems involving cryptography and data hiding often require billions of random numbers. In addition to the speed of the algorithm, the quality of the pseudo-random number generator and the ease of its implementation are common practical aspects. In this work we will discuss how to improve the quality of random numbers independently from their generation algorithm. We propose an additional implementation technique in order to take advantage of some chaotic properties. The statistical quality of our solution stems from some well-defined discrete chaotic iterations that satisfy the reputed Devaney's definition of chaos, namely the chaotic iterations technique. Pursuing recent researches published in the previous International Conference on Evolving Internet (Internet 09, 10, and 11), three methods to build pseudorandom generators by using chaotic iterations are recalled. Using standard criteria named NIST and DieHARD (some famous batteries of tests), we will show that the proposed technique can improve the statistical properties of a large variety of defective pseudorandom generators, and that the issues raised by statistical tests decrease when the power of chaotic iterations increase.

cs.CR↗

arXiv2017

A Pipeline for Insertion Sequence Detection and Study for Bacterial Genome

Insertion Sequences (ISs) are small DNA segments that have the ability of moving themselves into genomes. These types of mobile genetic elements (MGEs) seem to play an essential role in genomes rearrangements and evolution of prokaryotic genomes, but the tools that deal with discovering ISs in an efficient and accurate way are still too few and not totally precise. Two main factors have big effects on IS discovery, namely: genes annotation and functionality prediction. Indeed, some specific genes called "transposases" are enzymes that are responsible of the production and catalysis for such transposition, but there is currently no fully accurate method that could decide whether a given predicted gene is either a real transposase or not. This is why authors of this article aim at designing a novel pipeline for ISs detection and classification, which embeds the most recently available tools developed in this field of research, namely OASIS (Optimized Annotation System for Insertion Sequence) and ISFinder database (an up-to-date and accurate repository of known insertion sequences). As this latter depend on predicted coding sequences, the proposed pipeline will encompass too various kinds of bacterial genes annotation tools (that is, Prokka, BASys, and Prodigal). A complete IS detection and classification pipeline is then proposed and tested on a set of 23 complete genomes of Pseudomonas aeruginosa. This pipeline can also be used as an investigator of annotation tools performance, which has led us to conclude that Prodigal is the best software for IS prediction. A deepen study regarding IS elements in P.aeruginosa has then been conducted, leading to the conclusion that close genomes inside this species have also a close numbers of IS families and groups.

q-bio.GN↗

arXiv2017

Traversing a n-cube without Balanced Hamiltonian Cycle to Generate Pseudorandom Numbers

This article presents a new class of Pseudorandom Number Generators. The generators are based on traversing a n-cube where a Balanced Hamiltonian Cycle has been removed. The construction of such generators is automatic for small number of bits, but remains an open problem when this number becomes large. A running example is used throughout the paper. Finally, first statistical experiments of these generators are presented, they show how efficient and promising the proposed approach seems.

cs.DS↗

arXiv2017

Efficient and accurate monitoring of the depth information in a Wireless Multimedia Sensor Network based surveillance

Wireless Multimedia Sensor Network (WMSN) is a promising technology capturing rich multimedia data like audio and video, which can be useful to monitor an environment under surveillance. However, many scenarios in real time monitoring requires 3D depth information. In this research work, we propose to use the disparity map that is computed from two or multiple images, in order to monitor the depth information in an object or event under surveillance using WMSN. Our system is based on distributed wireless sensors allowing us to notably reduce the computational time needed for 3D depth reconstruction, thus permitting the success of real time solutions. Each pair of sensors will capture images for a targeted place/object and will operate a Stereo Matching in order to create a Disparity Map. Disparity maps will give us the ability to decrease traffic on the bandwidth, because they are of low size. This will increase WMSN lifetime. Any event can be detected after computing the depth value for the target object in the scene, and also 3D scene reconstruction can be achieved with a disparity map and some reference(s) image(s) taken by the node(s).

cs.CV↗