DiagnoseNET: Automatic Framework to Scale Neural Networks on Heterogeneous Systems.

Website: https://diagnosenet.github.io/

• Software Family : transfer;
• Audience: community;
• Evolution and maintenance: basic;
• Free Description: DiagnoseNET is a platform oriented to design a green intelligence medical workflow for deploying medical diagnostic tools with minimal infrastructure requirements and low power consumption. The first application built was to automate the unsupervised patient phenotype representation workflow trained on a mini-cluster of Nvidia Jetson TX2. The Python code is available on Github and freely distributed.

Indago

• Software Family : transfer;
• Audience: partners;
• Evolution and maintenance: lts, long term support;
• Duration of the Development (Duration): 1.8 years;

• Free Description: Indago implements a textual graph clustering method based on a joint analysis of the graph structure and the content exchanged between each nodes. This allows to reach a better segmentation than what could be obtained with traditional methods. Indago's main applications are built around communication network analysis, including social networks. However, Indago can be applied on any graph-structured textual network. Thus, Indago have been tested on various data, such as tweet corpus, mail networks, scientific paper co-publication network, etc.

  The software is used as a fully autonomous SaaS platform with 2 parts :

  1. A Python kernel that is responsible for the actual data processing.
  2. A web application that handles collecting, pre-processing and saving the data, such as providing a set of visualization for the interpretation of the results.

  Indago is deployed internally on the Inria network and used mainly by the development team for testing and research purposes. We also build tailored versions for industrial or academic partners that use the software externally (with contractual agreements).

Topix

Website: https://topix.mi.parisdescartes.fr/

• Software Family : research;
• Audience: universe;
• Evolution and maintenance: lts, long term support;
• Free Description: Topix is an innovative AI-based solution allowing to summarize massive and possibly extremely sparse data bases involving text. Topix is a versatile technology that can be applied in a large variety of situations where large matrices of texts / comments / reviews are written by users on products or addressed to other individuals (bi-partite networks). The typical use case consists in an e-commerce company interested in understanding the relationship between its users and the sold products thanks to the analysis of user comments. A simultaneous clustering (co-clustering) of users and products is produced by the Topix software, based on the key emerging topics from the reviews and by the underlying model. The Topix demonstration platform allows you to upload your own data on the website, in a totally secured framework, and let the AI-based software analyze them for you. The platform also provides some typical use cases to give a better idea of what Topix can do.

Cassiopy

Website: https://pypi.org/project/cassiopy/

• Software Family : vehicle;
• Audience: community;
• Evolution and maintenance: basic;
• Free Description: CassioPy is a python library for clustering using the Skew-t distribution. This model is designed to handle data with skewness, providing more accurate clustering results in many real-world scenarios where data may not follow a normal distribution.

7.1.1 Sparse GEMINI for Joint Discriminative Clustering and Feature Selection

Participants: Louis Ohl, Pierre-Alexandre Mattei, Charles Bouveyron, Arnaud Droit, Frédéric Precioso.

Keywords: Clustering, Deep learning, Sparsity, Feature Selection

Collaborations: Mickael Leclercq

Feature selection in clustering is a hard task which involves simultaneously the discovery of relevant clusters as well as relevant variables with respect to these clusters. While feature selection algorithms are often model-based through optimized model selection or strong assumptions on the data distribution, we introduce a discriminative clustering model trying to maximize a geometry-aware generalization of the mutual information called GEMINI with a simple ${\ell }_1$ penalty: the Sparse GEMINI 20. This algorithm avoids the burden of combinatorial feature subset exploration and is easily scalable to high-dimensional data and large amounts of samples while only designing a discriminative clustering model (see Figure 1). We demonstrate the performances of Sparse GEMINI on synthetic datasets and large-scale datasets. Our results show that Sparse GEMINI is a competitive algorithm and has the ability to select relevant subsets of variables with respect to the clustering without using relevance criteria or prior hypotheses.

7.1.2 A Deep Dynamic Latent Block Model for the Co-clustering of Zero-Inflated Data Matrices

Participants: Giulia Marchello, Marco Corneli, Charles Bouveyron.

Keywords: Co-clustering, Latent Block Model, zero-inflated distributions, dynamic systems, VEM algorithm.

Collaborations: Regional Center of Pharmacovigilance (RCPV) of Nice.

The simultaneous clustering of observations and features of data sets (known as co-clustering) has recently emerged as a central machine learning application to summarize massive data sets. However, most existing models focus on continuous and dense data in stationary scenarios, where cluster assignments do not evolve over time. In 18, we introduce a novel latent block model for the dynamic co-clustering of data matrices with high sparsity. To properly model this type of data, we assume that the observations follow a time and block dependent mixture of zero-inflated distributions, thus combining stochastic processes with the time-varying sparsity modeling. To detect abrupt changes in the dynamics of both cluster memberships and data sparsity, the mixing and sparsity proportions are modeled through systems of ordinary differential equations. The inference relies on an original variational procedure whose maximization step trains fully connected neural networks in order to solve the dynamical systems. Numerical experiments on simulated data sets demonstrate the effectiveness of the proposed methodology in the context of count data. The proposed method, called $Z{I}_{𝒫}$-dLBM, was then applied to two real data sets. The first is the data set on the London Bike sharing system while the second is a pharmacovigilance data set, on adverse drug reaction (ADR) reported to the Regional Center of Pharmacovigilance (RCPV) in Nice, France. Figure 2 shows some of the main results obtained through the application of $Z{I}_{𝒫}$-dLBM on the pharmacovigilance data set.

7.1.3 Deep Latent Position Topic Model for Clustering and Representation of Networks with Textual Edges

Participants: Charles Bouveyron, Rémi Boutin, Pierre Latouche.

Keywords: generative models, clustering, networks, text, topic modeling

Numerical interactions leading to users sharing textual content published by others are naturally represented by a network where the individuals are associated with the nodes and the exchanged texts with the edges. To understand those heterogeneous and complex data structures, clustering nodes into homogeneous groups as well as rendering a comprehensible visualization of the data is mandatory. To address both issues, we introduced Deep-LPTM, a model-based clustering strategy relying on a variational graph auto-encoder approach as well as a probabilistic model to characterize the topics of discussion. Deep-LPTM allows to build a joint representation of the nodes and of the edges in two embeddings spaces. The parameters are inferred using a variational inference algorithm. We also introduce IC2L, a model selection criterion specifically designed to choose models with relevant clustering and visualization properties. An extensive benchmark study on synthetic data is provided. In particular, we find that Deep-LPTM better recovers the partitions of the nodes than the state-of-the art ETSBM and STBM (see Figure 3). Eventually, the emails of the Enron company are analyzed and visualizations of the results are presented, with meaningful highlights of the graph structure.

7.1.4 Kernel KMeans clustering splits for end-to-end unsupervised decision trees

Participants: Louis Ohl, Pierre-Alexandre Mattei, Arnaud Droit, Frédéric Precioso.

Keywords: Clustering, interpretability, k-means, kernel methods

Collaborations: Mickael Leclercq

Trees are convenient models for obtaining explainable predictions on relatively small datasets. Although there are many proposals for the end-to-end construction of such trees in supervised learning, learning a tree end-to-end for clustering without labels remains an open challenge. As most works focus on interpreting with trees the result of another clustering algorithm, we present in 41 a novel end-to-end trained unsupervised binary tree for clustering: Kauri. This method performs a greedy maximization of the kernel KMeans objective without requiring the definition of centroids. We compare this model on multiple datasets with recent unsupervised trees and show that Kauri performs identically when using a linear kernel. For other kernels, Kauri often outperforms the concatenation of kernel KMeans and a CART decision tree.

7.1.5 The Deep Latent Position Block Model for the Clustering of Nodes in Multi-Graphs

Participants: Seydina Ousmane Niang, Charles Bouveyron, Marco Corneli, Pierre Latouche.

Keywords: Stochastic block model, Latent position model, Clustering, Multi-Graphs, Graph convolutional network

Network data capture relationships among actors across multiple contexts, often forming clusters of individuals. These relationships frequently involve multiple types of interactions, necessitating the use of multidimensional networks, or multigraphs, to capture their full complexity. Latent position models (LPM) embed nodes based on connection probabilities, but cannot uncover heterogeneous clusters such as disassortative patterns. Stochastic block models (SBM), in contrast, excels at clustering but lack interpretative latent representations. To address these limitations, the deep latent position block model (Deep-LPBM) was introduced to provide clustering and continuous latent space representation simultaneously in unidimensional networks. In this paper, we extend this work to multidimensional networks by introducing the deep latent position block model for multidimensional networks (Deep-LPBMM). Deep-LPBMM integrates block modeling and latent embedding across multiple interaction types, allowing nodes to partially belong to several groups, which better captures overlapping clustering structures. Our model uses a deep variational autoencoder with graph convolutional networks (GCNs) for each layer and a multi-layer perceptron to merge latent representations into a unified latent embedding representing cluster partial membership probabilities and offering effective clustering and enhanced visualization.

7.1.6 Clustering by Deep Latent Position Model with Graph Convolutional Network

Participants: Charles Bouveyron, Marco Corneli, Pierre Latouche.

Keywords: Clustering, Random Graphs, Deep Learning, Latent Position Models

Collaborations: Dingge Liang

With the significant increase of interactions between individuals through numeric means, clustering of vertices in graphs has become a fundamental approach for analyzing large and complex networks. In 17, we propose the deep latent position model (DeepLPM), an end-to-end generative clustering approach which combines the widely used latent position model (LPM) for network analysis with a graph convolutional network (GCN) encoding strategy. Moreover, an original estimation algorithm is introduced to integrate the explicit optimization of the posterior clustering probabilities via variational inference and the implicit optimization using stochastic gradient descent for graph reconstruction. Numerical experiments on simulated scenarios highlight the ability of DeepLPM to self-penalize the evidence lower bound for selecting the intrinsic dimension of the latent space and the number of clusters, demonstrating its clustering capabilities compared to state-of-the-art methods. Finally, DeepLPM is further applied to an ecclesiastical network in Merovingian Gaul and to a citation network Cora to illustrate the practical interest in exploring large and complex real-world networks.

7.1.7 A Deep Dynamic Latent Block Model for the Co-clustering of Zero-Inflated Data Matrices

Participants: Charles Bouveyron, Marco Corneli.

Keywords: Dynamic Co-Clustering, Mixture Models, deep learning

Collaborations: Giulia Marchello

The simultaneous clustering of observations and features of data sets (a.k.a. co-clustering) has recently emerged as a central machine learning task to summarize massive data sets. However, most existing models focus on stationary scenarios, where cluster assignments do not evolve in time. In 18 we introduce a novel latent block model for the dynamic co-clustering of data matrices with high sparsity. The data are assumed to follow dynamic mixtures of block-dependent zeroinflated distributions. Moreover, the sparsity parameter as well as the cluster proportions are assumed to be driven by dynamic systems, whose parameters must be estimated. The inference of the model parameters relies on an original variational EM algorithm whose maximization step trains fully connected neural networks that approximate the dynamic systems. Due to the model ability to work with empty clusters, the selection of the number of clusters can be done in a (computationally) parsimonious way. Numerical experiments on simulated and real world data sets demonstrate the effectiveness of the proposed methodology in the context of count data.

7.1.8 The Multiplex Deep Latent Position Model for the Clustering of nodes in Multiview Networks

Participants: Charles Bouveyron, Marco Corneli, Pierre Latouche.

Keywords: multiview networks, latent position model, graph neural networks

Collaborations: Dingge Liang, Junping Yin

Multiplex networks capture multiple types of interactions among the same set of nodes, creating a complex, multi-relational framework. A typical example is a social network where nodes (actors) are connected by various types of ties, such as professional, familial, or social relationships. Clustering nodes in these networks is a key challenge in unsupervised learning, given the increasing prevalence of multiview data across domains. While previous research has focused on extending statistical models to handle such networks, these adaptations often struggle to fully capture complex network structures and rely on computationally intensive Markov chain Monte Carlo (MCMC) for inference, rendering them less feasible for effective network analysis. To overcome these limitations, in 37, we propose the multiplex deep latent position model (MDLPM), which generalizes and extends latent position models to multiplex networks. MDLPM combines deep learning with variational inference to effectively tackle both the modeling and computational challenges raised by multiplex networks. Unlike most existing deep learning models for graphs that require external clustering algorithms (e.g., k-means) to group nodes based on their latent embeddings, MDLPM integrates clustering directly into the learning process, enabling a fully unsupervised, end-to-end approach. This integration improves the ability to uncover and interpret clusters in multiplex networks without relying on external procedures. Numerical experiments across various synthetic data sets and two real-world networks demonstrate the performance of MDLPM compared to state-of-the-art methods, highlighting its applicability and effectiveness for multiplex network analysis.

7.2.1 A new perspective on optimizers: leveraging Moreau-Yosida approximation in gradient-based learning

Participants: Alessandro Betti, Frédéric Precioso, Kevin Mottin.

Keywords: optimization, deep learning

Collaborations: Gabriele Ciravegna, Marco Gori, Stefano Melacci

Machine Learning (ML) heavily relies on optimization techniques built upon gradient descent. Numerous gradient-based update methods have been proposed in the scientific literature, particularly in the context of neural networks, and have gained widespread adoption as optimizers in ML software libraries. This paper 11 introduces a novel perspective by framing gradient-based update strategies using the Moreau-Yosida (MY) approximation of the loss function. Leveraging a first-order Taylor expansion, we demonstrate the concrete exploitability of the MY approximation to generalize the model update process. This enables the evaluation and comparison of regularization properties underlying popular optimizers like gradient descent with momentum, ADAGRAD, RMSprop, and ADAM. The MY-based unifying view opens up possibilities for designing new update schemes with customizable regularization properties. To illustrate this potential, we propose a case study that redefines the concept of closeness in the parameter space using network outputs. We present a proof-of-concept experimental procedure, demonstrating the effectiveness of this approach in continual learning scenarios. Specifically, we employ the well-known permuted MNIST dataset, a progressively-permuted MNIST and CIFAR-10 benchmarks, and a non i.i.d. stream. Additionally, we validate the update scheme's efficacy in an offline-learning scenario. By embracing the MY-based unifying view, we pave the way for advancements in optimization techniques for machine learning.

7.2.2 Are Ensembles Getting Better all the Time?

Participants: Damien Garreau, Pierre-Alexandre Mattei.

Keywords: Ensembles, dropout, random forests

Ensemble methods combine the predictions of several base models. We study whether or not including more models always improves their average performance. This question depends on the kind of ensemble considered, as well as the predictive metric chosen. We focus on situations where all members of the ensemble are a priori expected to perform as well, which is the case of several popular methods such as random forests or deep ensembles. In this setting, we show in 39 that ensembles are getting better all the time if, and only if, the considered loss function is convex. More precisely, in that case, the average loss of the ensemble is a decreasing function of the number of models. When the loss function is nonconvex, we show a series of results that can be summarized as: ensembles of good models keep getting better, and ensembles of bad models keep getting worse. To this end, we prove a new result on the monotonicity of tail probabilities that may be of independent interest. We illustrate our results on a medical prediction problem (diagnosing melanomas using neural nets) and a "wisdom of crowds" experiment (guessing the ratings of upcoming movies).

7.2.3 The Risks of Recourse in Binary Classification

Participants: Damien Garreau.

Keywords: Interpretability, Recourse, Machine Learning Theory

Collaborations: Hidde Fokkema, Tim van Erven

Algorithmic recourse provides explanations that help users overturn an unfavorable decision by a machine learning system. But so far very little attention has been paid to whether providing recourse is beneficial or not. We introduce in 29 an abstract learning-theoretic framework that compares the risks (i.e. expected losses) for classification with and without algorithmic recourse. This allows us to answer the question of when providing recourse is beneficial or harmful at the population level. Surprisingly, we find that there are many plausible scenarios in which providing recourse turns out to be harmful, because it pushes users to regions of higher class uncertainty and therefore leads to more mistakes. We further study whether the party deploying the classifier has an incentive to strategize in anticipation of having to provide recourse, and we find that sometimes they do, to the detriment of their users. Providing algorithmic recourse may therefore also be harmful at the systemic level. We confirm our theoretical findings in experiments on simulated and real-world data. All in all, we conclude that the current concept of algorithmic recourse is not reliably beneficial, and therefore requires rethinking.

7.2.4 Visual Objectification in Films: Towards a New AI Task for Video Interpretation

Participants: Julie Tores, Frédéric Precioso.

Keywords: Objectification, Video analysis, Interpretability, Concept Bottleneck Models

Collaborations: Lucile Sassatelli, Hui-Yin Wu, Clement Bergman, Léa Andolfi, Victor Ecrement, Thierry Devars, Magali Guaresi, Virginie Julliard, Sarah Lecossais

In film gender studies the concept of "male gaze" refers to the way the characters are portrayed on-screen as objects of desire rather than subjects. In 31 we introduce a novel video-interpretation task to detect character objectification in films. The purpose is to reveal and quantify the usage of complex temporal patterns operated in cinema to produce the cognitive perception of objectification. We introduce the ObyGaze12 dataset made of 1914 movie clips densely annotated by experts for objectification concepts identified in film studies and psychology. We evaluate recent vision models show the feasibility of the task and where the challenges remain with concept bottleneck models. Our new dataset and code are made available to the community.

7.3.1 Domain-Specific Long Text Classification from Sparse Relevant Information

Participants: Célia D'Cruz, Frédéric Precioso, Michel Riveill.

Keywords: Long Text Classification, Medical Report, Transformer, Information Extraction, Natural Language Processing

Collaborations: Jean-Marc Bereder

Large Language Models have undoubtedly revolutionized the Natural Language Processing field, the current trend being to promote one-model-for-all tasks (sentiment analysis, translation, etc.). However, the statistical mechanisms at work in the larger language models struggle to exploit the relevant information when it is very sparse, when it is a weak signal. This is the case, for example, for the classification of long domain-specific documents, when the relevance relies on a single relevant word or on very few relevant words from technical jargon. In the medical domain, it is essential to determine whether a given report contains critical information about a patient’s condition. This critical information is often based on one or few specific isolated terms. In 28, we propose a hierarchical model which exploits a short list of potential target terms to retrieve candidate sentences and represent them into the contextualized embedding of the target term(s) they contain. A pooling of the term(s) embedding(s) entails the document representation to be classified. We evaluate our model on one public medical document benchmark in English and on one private French medical dataset. We show that our narrower hierarchical model is better than larger language models for retrieving relevant long documents in a domain-specific context.

7.3.2 Detecting Brittle Decisions for Free: Leveraging Margin Consistency in Deep Robust Classifiers

Participants: Frédéric Precioso.

Keywords: adversarial robustness, empirical robustness estimation, classification, vulnerability detection

Collaborations: Jonas Ngnawé, Sabyasachi Sahoo, Yann Pequignot, Christian Gagné

Despite extensive research on adversarial training strategies to improve robustness, the decisions of even the most robust deep learning models can still be quite sensitive to imperceptible perturbations, creating serious risks when deploying them for high-stakes real-world applications. While detecting such cases may be critical, evaluating a model's vulnerability at a per-instance level using adversarial attacks is computationally too intensive and unsuitable for real-time deployment scenarios. The input space margin is the exact score to detect non-robust samples and is intractable for deep neural networks. In 30, we introduce the concept of margin consistency – a property that links the input space margins and the logit margins in robust models – for efficient detection of vulnerable samples. First, we establish that margin consistency is a necessary and sufficient condition to use a model's logit margin as a score for identifying non-robust samples. Next, through comprehensive empirical analysis of various robustly trained models on CIFAR10 and CIFAR100 datasets, we show that they indicate high margin consistency with a strong correlation between their input space margins and the logit margins. Then, we show that we can effectively and confidently use the logit margin to detect brittle decisions with such models. Finally, we address cases where the model is not sufficiently margin-consistent by learning a pseudo-margin from the feature representation. Our findings highlight the potential of leveraging deep representations to assess adversarial vulnerability in deployment scenarios efficiently. Figure 5 presents an overview of the MapEX method.

7.4.1 Mind the map! Accounting for existing map information when estimating online HDMaps from sensor data

Participants: Rémy Sun, Li Yang, Diane Lingrand, Frédéric Precioso.

Keywords: Autonomous Driving, HDMaps, Online HDMap estimation

Collaborations: ANR Project MultiTrans

Online High Definition Map (HDMap) estimation from sensors offers a low-cost alternative to manually acquired HDMaps. As such, it promises to lighten costs for already HDMap-reliant Autonomous Driving systems, and potentially even spread their use to new systems. We proposed to improve online HDMap estimation by accounting for already existing maps. We identify 3 reasonable types of useful existing maps (minimalist, noisy, and outdated). We also introduce MapEX (see Fig. 6), a novel online HDMap estimation framework that accounts for existing maps. MapEX achieves this by encoding map elements into query tokens and by refining the matching algorithm used to train classic query based map estimation models. We demonstrate that MapEX brings significant improvements on the nuScenes dataset. For instance, given noisy maps, MapEX improves by 38% over the MapTRv2 detector it is based on and by 16% over the current SOTA (state-of-the-art).

7.4.2 Model-based clustering with Missing Not At Random Data

Participants: Aude Sportisse.

Keywords: model-based clustering, generative models, missing data

Collaborations: Christophe Biernacki (Inria Lille), Claire Boyer (Sorbonne Unviersité), Julie Josse (Inria Montpellier) Matthieu Marbac (Ensai Rennes)

Model-based unsupervised learning, as any learning task, stalls as soon as missing data occurs. This is even more true when the missing data are informative, or said missing not at random (MNAR). In 25, we propose model-based clustering algorithms designed to handle very general types of missing data, including MNAR data. To do so, we introduce a mixture model for different types of data (continuous, count, categorical and mixed) to jointly model the data distribution and the MNAR mechanism, remaining vigilant to the relative degrees of freedom of each. Several MNAR models are discussed, for which the cause of the missingness can depend on both the values of the missing variable themselves and on the class membership. However, we focus on a specific MNAR model, called MNARz, for which the missingness only depends on the class membership. We first underline its ease of estimation, by showing that the statistical inference can be carried out on the data matrix concatenated with the missing mask considering finally a standard MAR (missing at random) mechanism. Consequently, we propose to perform clustering using the Expectation Maximization algorithm, specially developed for this simplified reinterpretation. Finally, we assess the numerical performances of the proposed methods on synthetic data and on the real medical registry TraumaBase as well.

7.4.3 A Model-Based Clustering Approach for Chemical Toxicity Assessment Using Cell Painting Data

Participants: Grigoryan Mariam, Vincent Vandewalle.

Collaborations: David Rouquié (Bayer Crop Science)

Keywords: Model-based Clustering, Toxicity assessment, Anomaly detection

Risk assessment of chemicals relies heavily on toxicity studies conducted on laboratory animals. However, those studies are lengthy, costly, and are not always fully relevant to human physiology. This is where human-derived, cell-based assays, such as the Cell Painting technique, could play a crucial role. This high-content imaging approach allows for analyzing several cellular compartments, offering a comprehensive view of how these chemicals impact cell morphology and organelle structure.

In 44, we propose an answer to the complex dose-response analysis in high dimensions using an intermediate clustering step of the cells within each well of the assay plates from Cell Painting experiments. The clustering model is a Gaussian mixture model assuming that each well contains several classes of cells with class proportions depending on the well. That class-specific parameters are the same across wells. This enables to summarize the information of each well by its specific class proportions. In the second step, we use these estimated class proportions to compare wells treated with chemical compounds at different concentrations. Applying this approach to real data from Cell Painting experiments allows the identification of consistent patterns across experimental conditions and provides insights into how varying concentrations affect cell behavior. This model will be extended to consider information from several cell lines.

7.4.4 Towards a fully automated underwater census for fish assemblages in the Mediterranean Sea

Participants: Kilian Bürgi, Charles Bouveyron, Diane Lingrand, Frédéric Precioso

Keywords: Diver operated video, Automated UVC, Deep learning, Object detection, Marine biology, Marine protected areas

Collaborations: Cecile Sabourault, Benoit Derijard

In marine biology and ecology the collection of data relies mostly on diver operations which is labour-intensive and financially costly to operate leading to reduced frequencies of data collection missions. Technological advances in the past decade have made it available for unmanned robots collecting data which resulted in numerous amounts of videos that need to be manually evaluated and analyzed which created a new bottleneck. In this study 35, we explored the possibilities and differences between a manual and automated analysis of collected videos by divers simulating a remotely operated vehicle (ROV). We discuss the difference between collecting data by diver and by videos and found that both methods added species to the overall species pool and that the automation was successful. This proof of concept will be used in future studies to fasten the process of data analysis and allow more frequent data collections creating more robust data for ecological decision making.

7.4.5 Automated Counting of Fish in Diver Operated Videos (DOV) for Biodiversity Assessments

Participants: Kilian Bürgi, Charles Bouveyron, Diane Lingrand, Frédéric Precioso.

Keywords: Underwater video, Fish count prediction, Temporal convolutional network, Object detection, Marine biology, Marine conservation

Collaborations: Cecile Sabourault, Benoit Derijard

Counting fish in moving underwater videos relies on labour-intensive manual counting or imprecise metrics from stationary cameras, while there is a great potential to use better methods to receive a better fish count. For this purpose, we explored traditional methods of counting fish as well as introduced three new methods to count fish from computer vision derived data (single frame detections). This resulted in a holistic and fully automated pipeline for fish abundance extraction 36. The following different methods are proposed on transect data of three Mediterranean species with different ecological niches: 1) traditional N${}_{max}$, 2) 1d k-means clustering method, 3) an intuitive clustering approach N${}_{Heuristic}$ and 4) a Temporal Convolutional Neural Networks (TCN) counting method. Our results show evidence of underestimation by the traditional N${}_{max}$ while the other methods showed better overall results with the proposed N${}_{Heuristic}$ and TCN methods best representing the reality. With an absolute variation comparable to inter-observer variation, we demonstrated reliable methods for quantifying fish counts.

7.4.6 Topological data analysis and multiple kernel learning for species identification of modern and archaeological small ruminants

Participants: Marco Corneli, Davide Adamo.

Keywords: Shape Analysis, Point Clouds, Classification, Multiple Kernel Learning, Topological Data Analysis

Collaborations: Manon Vuillien, Emmanuelle Vila, Agraw Amane, Thierry Argant, et al

The faunal remains from numerous Holocene archaeological sites across southwest Asia frequently include the bones of several wild and domestic ungulates, such as sheep, goats, ibexes, roe deer and gazelles. These assemblages may provide insight into hunting and animal husbandry strategies and offer paleoecological information on ancient human societies. However, the skeletons of these taxa are highly similar in appearance, which presents a challenge for accurate identification based on their bones. This paper 43 presents a case study to test the potential of topological data analysis (TDA) and multiple kernel learning (MKL) for inter-specific identification of 150 3D astragali belonging to modern and archaeological specimens. The joint application of TDA and MKL demonstrated remarkable efficacy in accurately identifying wild species, with a correct identification rate of approximately 90%. In contrast, the identification of domestic species exhibited a lower success rate, at approximately 60%. The misidentification of sheep and goat species is attributed to the morphological variability of domestic breeds. Moreover, while these methods assist in clearly identifying wild taxa from one another, they also highlight their morphological diversity. In this context, TDA and MKL could be invaluable for investigating intra-specific variability in domestic and wild animals. These methods offer a means of expanding our understanding of past domestic animal selection practices and techniques. They also facilitate an investigation into the morphological evolution of wild animal populations over time (see Fig. 7).

.2	.2	.3
(a)	(b)	(c)

Point cloud

Persistence diagram (PD) 3D data analysis pipeline via Topological Data Analysis (TDA).

7.4.7 A Bayesian approach for clustering and exact finite-sample model selection in longitudinal data mixtures

Participants: Marco Corneli.

Keywords: Clustering, Longitudinal Data, Mixture Models, Bayesian Model Selection

Collaborations: Elena Erosheva, Marco Lorenzi, Xunlei Qian

In 14, we consider mixtures of longitudinal trajectories, where one trajectory contains measurements over time of the variable of interest for one individual and each individual belongs to one cluster. The number of clusters as well as individual cluster memberships are unknown and must be inferred. We propose an original Bayesian clustering framework that allows us to obtain an exact finite-sample model selection criterion for selecting the number of clusters. Our finite-sample approach is more flexible and parsimonious than asymptotic alternatives such as Bayesian information criterion or integrated classification likelihood criterion in the choice of the number of clusters. Moreover, our approach has other desirable qualities: (i) it keeps the computational effort of the clustering algorithm under control and (ii) it generalizes to several families of regression mixture models, from linear to purely non-parametric. We test our method on simulated datasets as well as on a real world dataset from the Alzheimer’s disease neuroimaging initative database.

.48	.48
(a)	(b)

7.4.8 AI-Enhanced Prediction of Aortic Stenosis Progression: Insights From the PROGRESSA Study

Participants: Louis Ohl, Pierre-Alexandre Mattei, Frederic Precioso, Arnaud Droit.

Keywords: Cardiology, supervised learning

Collaborations: Melissa Sanabria, Lionel Tastet, Simon Pelletier, Mickael Leclercq, Lara Hermann, Nancy Coté, Philippe Pibarot

Aortic valve stenosis (AS) is a progressive chronic disease with progression rates that vary in patients and therefore difficult to predict. The aim of this study 24 was to predict the progression of AS using comprehensive and longitudinal patient data . Machine learning algorithms were trained on a data set of 303 patients enrolled in the PROGRESSA (Metabolic Determinants of the Progression of Aortic Stenosis) study who underwent clinical and echocardiographic follow-up on an annual basis. Performance of the models was measured to predict disease progression over long (next 5 years) and short (next 2 years) terms and was compared to a standard clinical model with usually used features in clinical settings based on logistic regression. For each annual follow-up visit including baseline, we trained various supervised learning algorithms in predicting disease progression at 2- and 5-year terms. At both terms, LightGBM consistently outperformed other models with the highest average area under curves across patient visits (0.85 at 2 years, 0.83 at 5 years). Recurrent neural network-based models (Gated Recurrent Unit and Long Short-Term Memory) and XGBoost also demonstrated strong predictive capabilities, while the clinical model showed the lowest performance. This study demonstrates how an artificial intelligence-guided approach in clinical routine could help enhance risk stratification of AS. It presents models based on multisource comprehensive data to predict disease progression and clinical outcomes in patients with mild-to-moderate AS at baseline.

7.4.9 BERNN: Enhancing classification of Liquid Chromatography Mass Spectrometry data with batch effect removal neural networks

Participants: Frederic Precioso, Arnaud Droit.

Keywords: Batch effect removal, Mass Spectrometry, omics

Collaborations: Melissa Sanabria, Lionel Tastet, Simon Pelletier, Mickael Leclercq, Lara Hermann, Nancy Coté, Philippe Pibarot

Liquid Chromatography Mass Spectrometry (LC-MS) is a powerful method for profiling complex biological samples. However, batch effects typically arise from differences in sample processing protocols, experimental conditions, and data acquisition techniques, significantly impacting the interpretability of results. Correcting batch effects is crucial for the reproducibility of omics research, but current methods are not optimal for the removal of batch effects without compressing the genuine biological variation under study. In 21, we propose a suite of Batch Effect Removal Neural Networks (BERNN) to remove batch effects in large LC-MS experiments, with the goal of maximizing sample classification performance between conditions. More importantly, these models must efficiently generalize in batches not seen during training. A comparison of batch effect correction methods across five diverse datasets demonstrated that BERNN models consistently showed the strongest sample classification performance. However, the model producing the greatest classification improvements did not always perform best in terms of batch effect removal. Finally, we show that the overcorrection of batch effects resulted in the loss of some essential biological variability. These findings highlight the importance of balancing batch effect removal while preserving valuable biological diversity in large-scale LC-MS experiments. Figure 9 shows the used architecture for the VAE.

7.4.10 An artificial intelligence algorithm for co-clustering to help in pharmacovigilance before and during the COVID-19 pandemic

Participants: Charles Bouveyron, Marco Corneli, Alexandre Destere.

Keywords: Pharmacovigilance, co-clustering, COVID-19

Collaborations: Giulia Marchello, Diane Merino, Nouha Ben Othman, Alexandre O. Gérard, Thibaud Lavrut, Delphine Viard, Fanny Rocher, Milou-Daniel Drici

Monitoring drug safety in real-world settings is the primary aim of pharmacovigilance. Frequent adverse drug reactions (ADRs) are usually identified during drug development. Rare ones are mostly characterized through post-marketing scrutiny, increasingly with the use of data mining and disproportionality approaches, which lead to new drug safety signals. Nonetheless, waves of excessive numbers of reports, often stirred up by social media, may overwhelm and distort this process, as observed recently with levothyroxine or COVID-19 vaccines. As human resources become rarer in the field of pharmacovigilance, we aimed to evaluate the performance of an unsupervised co-clustering method to help the monitoring of drug safety. In 15, a dynamic latent block model (dLBM), based on a time-dependent co-clustering generative method, was used to summarize all regional ADR reports ($n\hspace{0.17em}=\hspace{0.17em}45,269$) issued between January 1, 2012 and February 28, 2022. After analysis of their intra and extra interrelationships, all reports were grouped into different cluster types (time, drug, ADR). Our model clustered all reports in 10 time, 10 ADR and 9 drug collections. Based on such clustering, three prominent societal problems were detected, subsequent to public health concerns about drug safety, including a prominent media hype about the perceived safety of COVID-19 vaccines. The dLBM also highlighted some specific drug–ADR relationships, such as the association between antiplatelets, anticoagulants and bleeding. Co-clustering and dLBM appear as promising tools to explore large pharmacovigilance databases. They allow, without supervision, the detection, exploration and strengthening of safety signals, facilitating the analysis of massive upsurges of reports.

9.2.1 Visits of international scientists

9.3.1 H2020 projects

Maasai is one of the 3IA-UCA research teams of AI4Media, one of the 4 ICT-48 Center of Excellence in Artificial Intelligence which has started in September 2020. There are 30 partners (Universities and companies), and 3IA-UCA received about 325k€.

10.1.1 Scientific events: organisation

• GeMSS 2024: Pierre-Alexandre Mattei co-founded, co-organized, and co-taught the Generative Modelling Summer School (GeMSS) since 2023. The 2024 edition was held in Eindhoven (June 24th to 28th). More details on the GeMSS website.
• GenU 2024: Pierre-Alexandre Mattei co-founded in 2019 the workshop on Generative Models and Uncertainty Quantification (GenU). This small-scale workshop has been held physically in Copenhagen in the Fall. The 2024 edition was on September 18-19, 2023. More details on the GenU website.
• SophI.A Summit 2024: It is an AI conference that brings together researchers and companies doing AI, held every Fall in Sophia Antipolis. In 2024, Diane Lingrand and Pierre-Alexandre Mattei were members of the scientific committee, and Charles Bouveyron gave the opening speech. More details on the website.
• Deep learning school 2024: the team organized, under the guidance of F. Precioso, the 5th edition of this summer school on Deep Learning. The School took place from July 1st to July 5th 2024 and welcomed about 100 participants (School website).

10.1.2 Scientific publishing

Member of editorial boards

• Charles Bouveyron is Associate Editor for the Annals of Applied Statistics since 2016.
• Pierre-Alexandre Mattei is Area Chair since 2024 for the conferences NeurIPS and ICML.

Reviewing activities All permanent members of the team are serving as reviewers for the most important journals and conferences in statistical and machine learning, such as (non exhaustive list):

• International journals:
  • Annals of Applied Statistics
  • Statistics and Computing
  • Journal of the Royal Statistical Society, Series C
  • Journal of Machine Learning Research
  • Transactions on Machine Learning Research
• International Conferences
  • Neural Information Processing Systems (NeurIPS)
  • International Conference on Machine Learning (ICML)
  • International Conference on Learning Representations (ICLR)
  • International Joint Conference on Artificial Intelligence (IJCAI)
  • International Conference on Artificial Intelligence and Statistics (AISTATS)
  • IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

10.1.3 Leadership within the scientific community

• Charles Bouveyron is the Director of the Institut 3IA Côte d'Azur since January 2021 and of the EFELIA Côte d'Azur education program since September 2022.
• Vincent Vandewalle is the Deputy Scientific director of the EFELIA Côte d'Azur education program since September 2022.

10.1.4 Scientific expertise

• Charles Bouveyron is member of the Scientific Orientation Council of Centre Antoine Lacassagne, Unicancer center of Nice.

10.1.5 Research administration

Charles Bouveyron is the director the Institut 3IA Côte d'Azur and the EFELIA Côte d'Azur project. He also leaded the demand for renewing the Institute funding , that was granted by the French governement for an additional 5 years (budget 20M€).

10.2.1 Supervision

The team has 3 senior researchers with HDR that are able to supervise Ph.D. students. Usually, the supervision of the Ph.D. students of the team is jointly made by a senior and a junior researchers of the team. Two Ph.D. students of the team have defended their thesis in 2024.

10.2.2 Juries

C. Bouveyron, M. Riveill, F. Precioso and V. Vandewalle are full professors, D. Garreau and D. Lingrand are associate professors at Université Côte d'Azur and therefore handle usual teaching duties at the university. M. Corneli, P.-A. Mattei, and R. Sun are also teaching around 60h per year at Université Côte d'Azur. P.-A. Mattei is also teaching a graphical models course at the MVA masters from ENS Paris Saclay. M. Corneli has been hired in September 2022 on a “Chaire de Professeur Junior" on AI for Archeology and Historical Sciences.

M. Riveill is the current director of the Master of Science “Data Sciences and Artificial Intelligence" at Université Côte d'Azur, since September 2020. C. Bouveyron was the founder and first responsible (Sept. 2018 - Aug. 2020) of that MSc.

Since September 2022, C. Bouveyron and V. Vandewalle are respectively the Director and Deputy Scientific Director of the EFELIA Côte d'Azur program, funded by the French national plan “France 2030", through the “Compétences et Métiers d'Avenir" initiative (8M€ for 5 years). This program aims at enlarging the teaching capacities in AI of the Institut 3IA Côte d'Azur and developing new education programs for specialists and non-specialists.

All members of the team are also actively involved in the supervision of postdocs, Ph.D. students, interns and participate frequently to Ph.D. and HDR defenses. They are also frequently part of juries for the recruitment of research scientists, assistant-professors or professors.

International journals

• 11 articleA.Alessandro Betti, G.Gabriele Ciravegna, M.Marco Gori, S.Stefano Melacci, K.Kevin Mottin and F.Frédéric Precioso. A new perspective on optimizers: leveraging moreau-yosida approximation in gradient-based learning.Intelligenza Artificiale182August 2024, 301-311HALDOIback to text
• 12 articleK.Kilian Bürgi, C.Charles Bouveyron, D.Diane Lingrand, B.Benoit Derijard, F.Frédéric Precioso and C.Cécile Sabourault. Towards a fully automated underwater census for fish assemblages in the Mediterranean Sea.Ecological Informatics85March 2025, 102959HALDOI
• 13 articleC.Cyrielle Codde, F.Florence Rivals, A.Alexandre Destere, Y.Yeleen Fromage, M.Marc Labriffe, P.Pierre Marquet, C.Clément Benoist, L.Laure Ponthier, J.-F.Jean-François Faucher and J.-B.Jean-Baptiste Woillard. A machine learning approach to predict daptomycin exposure from two concentrations based on Monte Carlo simulations.Antimicrobial Agents and Chemotherapy685May 2024HALDOI
• 14 articleM.Marco Corneli, E.Elena Erosheva, X.Xunlei Qian and M.Marco Lorenzi. A Bayesian approach for clustering and exact finite-sample model selection in longitudinal data mixtures.Computational StatisticsMay 2024HALDOIback to text
• 15 articleA.Alexandre Destere, G.Giulia Marchello, D.Diane Merino, N. B.Nouha Ben Othman, A.Alexandre Gérard, T.Thibaud Lavrut, D.Delphine Viard, F.Fanny Rocher, M.Marco Corneli, C.Charles Bouveyron and M.Milou‐daniel Drici. An artificial intelligence algorithm for co‐clustering to help in pharmacovigilance before and during the COVID‐19 pandemic.British Journal of Clinical Pharmacology905February 2024, 1258-1267HALDOIback to text
• 16 articleV.Victor Hémar, S.Stéphane Bouchet, E.Emmanuel Ribeiro, P.Perrine Prier, C.Christophe Elleau, C.Caroline Solas, A.Alexandre Destere, M.Mojgan Hessamfar, C.Camille Tumiotto and F.Fabrice Bonnet. A case of lenacapavir use for preventing mother-to-child HIV transmission.Journal of Antimicrobial Chemotherapy171September 2024, 15-21HALDOI
• 17 articleD.Dingge Liang, M.Marco Corneli, C.Charles Bouveyron and P.Pierre Latouche. Clustering by Deep Latent Position Model with Graph Convolutional Network.Advances in Data Analysis and Classification2025. In press. HALDOIback to text
• 18 articleG.Giulia Marchello, M.Marco Corneli and C.Charles Bouveyron. A Deep Dynamic Latent Block Model for the Co-clustering of Zero-Inflated Data Matrices.Journal of Computational and Graphical Statistics334April 2024, 1224-1239HALDOIback to textback to text
• 19 articleD.Diane Merino, A.Alexandre Gérard, A.Alexandre Destere, H.Haitam Saidessalam, F.Florence Askenazy, F.François Montastruc, M.-D.Milou-Daniel Drici and S.Susanne Thümmler. Cardiac and metabolic safety profile of antipsychotics in youths: A WHO safety database analysis.Psychiatry Research334April 2024, 115786HALDOI
• 20 articleL.Louis Ohl, P.-A.Pierre-Alexandre Mattei, C.Charles Bouveyron, M.Mickaël Leclercq, A.Arnaud Droit and F.Frédéric Precioso. Sparse and geometry-aware generalisation of the mutual information for joint discriminative clustering and feature selection.Statistics and Computing345July 2024, 155HALDOIback to text
• 21 articleS.Simon Pelletier, M.Mickaël Leclercq, F.Florence Roux-Dalvai, M.Matthijs de Geus, S.Shannon Leslie, W.Weiwei Wang, T.Tukiet Lam, A.Angus Nairn, S.Steven Arnold, B.Becky Carlyle, F.Frédéric Precioso and A.Arnaud Droit. BERNN: Enhancing classification of Liquid Chromatography Mass Spectrometry data with batch effect removal neural networks.Nature Communications151May 2024, 3777HALDOIback to text
• 22 articleF.Florence Rivals, S.Sylvain Goutelle, C.Cyrielle Codde, R.Romain Garreau, L.Laure Ponthier, P.Pierre Marquet, T.Tristan Ferry, M.Marc Labriffe, A.Alexandre Destere and J.-B.Jean-Baptiste Woillard. A Machine Learning Algorithm to Predict the Starting Dose of Daptomycin.Clinical PharmacokineticsJuly 2024HALDOI
• 23 articleI.Isabelle Rubera, L.Laetitia Clotaire, A.Audrey Laurain, A.Alexandre Destere, L.Ludovic Martin, C.Christophe Duranton and G.Georges Leftheriotis. A Plasma Pyrophosphate Cutoff Value for Diagnosing Pseudoxanthoma Elasticum.International Journal of Molecular Sciences2512June 2024, 6502HALDOI
• 24 articleM.Melissa Sanabria, L.Lionel Tastet, S.Simon Pelletier, M.Mickael Leclercq, L.Louis Ohl, L.Lara Hermann, P.-A.Pierre-Alexandre Mattei, F.Frédéric Precioso, N.Nancy Coté, P.Philippe Pibarot and A.Arnaud Droit. AI-Enhanced Prediction of Aortic Stenosis Progression.JACC. Advances310October 2024, 101234HALDOIback to text
• 25 articleA.Aude Sportisse, M.Matthieu Marbac, F.Fabien Laporte, G.Gilles Celeux, C.Claire Boyer, J.Julie Josse and C.Christophe Biernacki. Model-based Clustering with Missing Not At Random Data.Statistics and ComputingJune 2024HALDOIback to text
• 26 articleJ.Jean‐baptiste Woillard, C.Clément Benoist, A.Alexandre Destere, M.Marc Labriffe, G.Giulia Marchello, J.Julie Josse and P.Pierre Marquet. To be or not to be, when synthetic data meet clinical pharmacology: A focused study on pharmacogenetics.CPT: Pharmacometrics and Systems PharmacologySeptember 2024, Online ahead of printHALDOI

Invited conferences

• 27 inproceedingsC.Christophe Biernacki and V.Vincent Vandewalle. An EM Stopping Rule for Avoiding Degeneracy in Gaussian-based Clustering with Missing Data.Advances in Intelligent Systems and ComputingSMPS 2024 - International Conference on Soft Methods in Probability and StatisticsAISC-1458Combining, Modelling and Analyzing Imprecision, Randomness and DependenceSalsburg, AustriaSeptember 2024, 27-35HALDOI

International peer-reviewed conferences

• 28 inproceedingsC.Célia D’cruz, J.-M.Jean-Marc Bereder, F.Frédéric Precioso and M.Michel Riveill. Domain-Specific Long Text Classification from Sparse Relevant Information.Frontiers in Artificial Intelligence and ApplicationsECAI 2024 - 27th european conference on artificial intelligence392: ECAI 2024cover 27th European Conference on Artificial Intelligence, 19–24 October 2024, Santiago de Compostela, Spain – Including 13th Conference on Prestigious Applications of Intelligent Systems (PAIS 2024)Santiago de Compostela, SpainIOS PressOctober 2024, 4003 - 4010HALDOIback to text
• 29 inproceedingsH.Hidde Fokkema, D.Damien Garreau and T.Tim van Erven. The Risks of Recourse in Binary Classification.Proceedings of Machine Learning Research,27th International Conference on Artificial Intelligence and Statistics (AISTATS)238Valencia (Espagne), Spain2024, 550-558HALback to text
• 30 inproceedingsJ.Jonas Ngnawé, S.Sabyasachi Sahoo, Y.Yann Pequignot, F.Frédéric Precioso and C.Christian Gagné. Detecting Brittle Decisions for Free: Leveraging Margin Consistency in Deep Robust Classifiers.NeurIPS ProceedingsNeurIPS 2024 - Thirty-eighth Annual Conference on Neural Information Processing SystemsVancouver (CA), CanadaDecember 2024HALback to text
• 31 inproceedingsJ.Julie Tores, L.Lucile Sassatelli, H.-Y.Hui-Yin Wu, C.Clement Bergman, L.Lea Andolfi, V.Victor Ecrement, F.Frédéric Precioso, T.Thierry Devars, M.Magali Guaresi, V.Virginie Julliard and S.Sarah Lecossais. Visual Objectification in Films: Towards a New AI Task for Video Interpretation.IEEE XploreCVPR 2024 - IEEE / CVF Computer Vision and Pattern Recognition Conference2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)Seattle (Washington), United StatesJune 2024, 10864-10874HALDOIback to text

Doctoral dissertations and habilitation theses

• 32 thesisB.Baptiste Pouthier. Deep and statistical learning on audio-visual data for human-machine interface on embedded systems.Université Côte d'AzurJune 2024HAL
• 33 thesisM.Mansour Zoubeirou a Mayaki. Deep learning based anomaly and drift detection : application to predictive maintenance in embedded systems.Université Côte d'AzurApril 2024HAL

Reports & preprints

• 34 miscR.Rémi Boutin, P.Pierre Latouche and C.Charles Bouveyron. The Deep Latent Position Block Model For The Block Clustering And Latent Representation Of Networks.November 2024HAL
• 35 miscK.Kilian Bürgi, C.Charles Bouveyron, D.Diane Lingrand, B.Benoit Dérijard, F.Frédéric Precioso and C.Cécile Sabourault. Towards a Fully Automated Underwater Census for Fish Assemblages in the Mediterranean Sea.June 2024HALback to text
• 36 miscK.Kilian Bürgi, R.Rémy Sun, C.Charles Bouveyron, D.Diane Lingrand, B.Benoit Dérijard, F.Frédéric Precioso and C.Cécile Sabourault. Automated Counting of Fish in Diver Operated Videos (DOV) for Biodiversity Assessments: Automated Fish Counting in DOV.January 2025HALback to text
• 37 miscD.Dingge Liang, M.Marco Corneli, C.Charles Bouveyron, P.Pierre Latouche and J.Junping Yin. The Multiplex Deep Latent Position Model for the Clustering of nodes in Multiview Networks.November 2024HALback to text
• 38 miscG.Giulia Marchello, A.Alexandre Destere, M.Marco Corneli and C.Charles Bouveyron. Deep dynamic co-clustering of count data streams: application to pharmacovigilance.January 2024HAL
• 39 misc P.-A.Pierre-Alexandre Mattei and D.Damien Garreau. Are Ensembles Getting Better all the Time? April 2024 HAL back to text
• 40 miscS. O.Seydina Ousmane Niang, C.Charles Bouveyron, M.Marco Corneli, P.Pierre Latouche and R.Rémi Boutin. The Deep Latent Position Block Model for the Clustering of Nodes in Multi-Graphs.December 2024HAL
• 41 miscL.Louis Ohl, P.-A.Pierre-Alexandre Mattei, M.Mickaël Leclercq, A.Arnaud Droit and F.Frédéric Precioso. Kernel KMeans clustering splits for end-to-end unsupervised decision trees.February 2024HALback to text
• 42 miscS.Sabyasachi Sahoo, M.Mostafa Elaraby, J.Jonas Ngnawe, Y.Yann Pequignot, F.Frédéric Precioso and C.Christian Gagné. Layerwise Early Stopping for Test Time Adaptation.April 2024HAL
• 43 miscM.Manon Vuillien, D.Davide Adamo, E.Emmanuelle Vila, A.Agraw Amane, T.Thierry Argant, D.Daniel Helmer, M.Marjan Mashkour, A.Abdelkader Moussous, O.Olivier Notter, E.Elena Rossoni-Notter, I.Isabelle Théry-Parisot and M.Marco Corneli. Topological data analysis and multiple kernel learning for species identification of modern and archaeological small ruminants.September 2024HALback to text

Other scientific publications

• 44 inproceedingsM.Mariam Grigoryan, V.Vincent Vandewalle and D.David Rouquié. A Model-Based Clustering Approach for Chemical Toxicity Assessment Using Cell Painting Data.SophI.A Summit 2024 - 7th international AI conferenceValbonne, FranceNovember 2024HALback to text