Papers by Farida Zehraoui
SAPHIR: a collaborative multi‐scale, multi‐resolution Core Model Environment for the Physiome—with a prototype core model of blood pressure regulation and fluid homeostasis
The FASEB Journal, 2008
We present the current state of the SAPHIR project, a multi-resolution core modeling environment ... more We present the current state of the SAPHIR project, a multi-resolution core modeling environment (CME) in the spirit of the IUPS Physiome, with application to a prototype core model based on a modular implementation of the classic systems model by Guyton et al. (1972 Ann. Rev. Physiol. 34:13–44) and its extension by Ikeda et al. (1979 Annals Biomed. Engin. 7:135–166). This core model targets short- and long-term regulation of blood pressure and homeostasis of body fluids and major solutes. The aim is to provide a collaborative modeling environment enabling plug-and-play construction of integrated systems models with lumped-parameter sub-models at the organ/tissue level yet also allowing focus on cell- or molecular-level detailed models embedded in the larger core model. Thus, in silico exploration of gene-to-organ-to-organism scenarios is possible while keeping computation time manageable. The CME is built on the M2SL toolbox, a multi-scale, multi-resolution, multi-mode open source package developed in C++ by one of us (AH). Associated with the CME is an ontology-based database allowing exploration of the modules, parameter values, and equations.In parallel with the CME implementation of the core model, we also present stand-alone implementations in Berkeley Madonna (Ikeda model) and Simulink (Guyton model).
A3SOM, abstained explainable semi-supervised neural network based on self-organizing map
PLOS ONE
In the sea of data generated daily, unlabeled samples greatly outnumber labeled ones. This is due... more In the sea of data generated daily, unlabeled samples greatly outnumber labeled ones. This is due to the fact that, in many application areas, labels are scarce or hard to obtain. In addition, unlabeled samples might belong to new classes that are not available in the label set associated with data. In this context, we propose A3SOM, an abstained explainable semi-supervised neural network that associates a self-organizing map to dense layers in order to classify samples. Abstained classification enables the detection of new classes and class overlaps. The use of a self-organizing map in A3SOM allows integrated visualization and makes the model explainable. Along with describing our approach, this paper shows that the method is competitive with other classifiers and demonstrates the benefits of including abstention rules. A use case is presented on breast cancer subtype classification and discovery to show the relevance of our method in real-world medical problems.
Nucleic Acids Research
Recent advances have shown that some biologically active non-coding RNAs (ncRNAs) are actually tr... more Recent advances have shown that some biologically active non-coding RNAs (ncRNAs) are actually translated into polypeptides that have a physiological function as well. This paradigm shift requires adapted computational methods to predict this new class of ‘bifunctional RNAs’. Previously, we developed IRSOM, an open-source algorithm to classify non-coding and coding RNAs. Here, we use the binary statistical model of IRSOM as a ternary classifier, called IRSOM2, to identify bifunctional RNAs as a rejection of the two other classes. We present its easy-to-use web interface, which allows users to perform predictions on large datasets of RNA sequences in a short time, to re-train the model with their own data, and to visualize and analyze the classification results thanks to the implementation of self-organizing maps (SOM). We also propose a new benchmark of experimentally validated RNAs that play both protein-coding and non-coding roles, in different organisms. Thus, IRSOM2 showed promi...
Explainable Ensemble Classification Model based on Argumentation
HAL (Le Centre pour la Communication Scientifique Directe), May 29, 2023
HAL (Le Centre pour la Communication Scientifique Directe), Jun 27, 2022
De façon générale, l'éthique propose de s'interroger sur les valeurs morales et les principes mor... more De façon générale, l'éthique propose de s'interroger sur les valeurs morales et les principes moraux qui devraient orienter nos ac>ons, dans différentes situa>ons, dans le but d'agir conformément à ceux-ci.
Self-organizing maps with supervised layer
2017 12th International Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization (WSOM)
We present in this paper a new approach of supervised self organizing map (SOM). We added a super... more We present in this paper a new approach of supervised self organizing map (SOM). We added a supervised perceptron layer to the classical SOM approach. This combination allows the classification of new patterns by taking into account all the map prototypes without changing the SOM organization. We also propose to associate two reject options to our supervised SOM. This allows to improve the results reliability and to discover new classes in applications where some classes are unknown. We obtain two variants of supervised SOM with rejection that have been evaluated on different datasets. The results indicate that our approaches are competitive with most popular supervised leaning algorithms like support vector machines and random forest.
HAL (Le Centre pour la Communication Scientifique Directe), Jul 15, 2021
Ce rapport de recherche a été rédigé dans le cadre d'un projet de collaboration entre le service ... more Ce rapport de recherche a été rédigé dans le cadre d'un projet de collaboration entre le service de cardiologie du CHSF (Centre hospitalier Sud Francilien) et le laboratoire IBISC (Informatique, BioInformatique, Systèmes Complexes). Il regroupe, notamment, des travaux réalisés dans le cadre du stage de Master 2 en Informatique de Gilles Bellido (Année 2016-2017) et des stages de 2 ème année d'école d'ingénieur (ENSIIE) de Laure Bedu (Année 2016-2017) et Quentin Picholet (Année 2018-2019). Qu'ils soient remerciés pour leur contribution. 4 Perspectives et conclusion Appendices A Organigramme de la consultation semi-urgente B Questionnaire lors de l'appel téléphonique C Questionnaire lors de la consultation semi-urgente
Bioinformatics, 2022
Motivation Medical care is becoming more and more specific to patients’ needs due to the increase... more Motivation Medical care is becoming more and more specific to patients’ needs due to the increased availability of omics data. The application to these data of sophisticated machine learning models, in particular deep learning (DL), can improve the field of precision medicine. However, their use in clinics is limited as their predictions are not accompanied by an explanation. The production of accurate and intelligible predictions can benefit from the inclusion of domain knowledge. Therefore, knowledge-based DL models appear to be a promising solution. Results In this article, we propose GraphGONet, where the Gene Ontology is encapsulated in the hidden layers of a new self-explaining neural network. Each neuron in the layers represents a biological concept, combining the gene expression profile of a patient and the information from its neighboring neurons. The experiments described in the article confirm that our model not only performs as accurately as the state-of-the-art (non-exp...
Towards a Transparent Deep Ensemble Method Based on Multiagent Argumentation
Ensemble methods improve the machine learning results by combining different models. However, one... more Ensemble methods improve the machine learning results by combining different models. However, one of the major drawbacks of these approaches is their opacity, as they do not provide results explanation and they do not allow prior knowledge integration. As the use of machine learning increases in critical areas, the explanation of classification results and the ability to introduce domain knowledge inside the learned model have become a necessity. In this paper, we present a new deep ensemble method based on argumentation that combines machine learning algorithms with a multiagent system in order to explain the results of classification and to allow injecting prior knowledge. The idea is to extract arguments from classifiers and combine the classifiers using argumentation. This allows to exploit the internal knowledge of each classifier, to provide an explanation for the decisions and facilitate integration of domain knowledge. The results demonstrate that our method effectively impr...
Systèmes d'apprentissage connexionnistes et raisonnement à partir de cas pour la classification et le classement de séquence
La these porte sur l'utilisation de techniques d'apprentissage automatique pour le traite... more La these porte sur l'utilisation de techniques d'apprentissage automatique pour le traitement de sequences. Ce travail est motive par une application reelle qui consiste a modeliser le comportement d'un internaute a partir de traces de navigations. Nous avons d'abord propose un systeme de raisonnement a partir de cas (RaPC) > pour le classement ou la prediction a partir de sequences. Des mesures de maintenance sont associees aux cas pour faire face au bruit contenu dans les donnees et pour effectuer la reduction de la base de cas. Nous avons ensuite propose plusieurs modeles de cartes auto-organisatrices (SOM) pour la classification et le classement de sequences. Nous nous sommes interesses a l'insertion des proprietes de plasticite et de stabilite dans une carte SOM ainsi qu'au traitement des sequences temporelles. A la fin de cette etude, nous avons travaille sur la possibilite de cooperation des systemes connexionnistes avec l'approche de RaPC afin ...
8 : Théorie de résonance adaptative (Adaptive Resonance Theory)
Pour realiser un systeme qui permette d'acquerir en continu de nouvelles connaissances (plast... more Pour realiser un systeme qui permette d'acquerir en continu de nouvelles connaissances (plasticite) tout en continuant a memoriser les anciennes (stabilite), Carpenter et Grossberg ont propose une theorie dite de la resonance adaptative ART. Cette theorie est le resultat d'une tentative pour comprendre a quel point les systemes biologiques sont capables de maintenir la plasticite durant toute leur vie, sans compromettre la stabilite des modeles precedemment construits. Les mecanismes d'apprentissage qui sont biologiquement fondes doivent pouvoir garder les connaissances deja acquises tout en continuant a apprendre les nouveaux evenements de l'environnement. C'est ce qui est appele par Grossberg . Les modeles construits en utilisant cette theorie sont parmi les rares modeles qui peuvent apprendre dans un environnement variant continument. Ces modeles protegent les connaissances acquises en rendant l'apprentissage conditionnel. Pour ce faire, ils distinguent la...
BMC Bioinformatics, 2020
Background The use of predictive gene signatures to assist clinical decision is becoming more and... more Background The use of predictive gene signatures to assist clinical decision is becoming more and more important. Deep learning has a huge potential in the prediction of phenotype from gene expression profiles. However, neural networks are viewed as black boxes, where accurate predictions are provided without any explanation. The requirements for these models to become interpretable are increasing, especially in the medical field. Results We focus on explaining the predictions of a deep neural network model built from gene expression data. The most important neurons and genes influencing the predictions are identified and linked to biological knowledge. Our experiments on cancer prediction show that: (1) deep learning approach outperforms classical machine learning methods on large training sets; (2) our approach produces interpretations more coherent with biology than the state-of-the-art based approaches; (3) we can provide a comprehensive explanation of the predictions for biolog...
Bioinformatics, 2014
Motivation: Identifying the set of genes differentially expressed along time is an important task... more Motivation: Identifying the set of genes differentially expressed along time is an important task in two-sample time course experiments. Furthermore, estimating at which time periods the differential expression is present can provide additional insight into temporal gene functions. The current differential detection methods are designed to detect difference along observation time intervals or on single measurement points, warranting dense measurements along time to characterize the full temporal differential expression patterns. Results: We propose a novel Bayesian likelihood ratio test to estimate the differential expression time periods. Applying the ratio test to systems of genes provides the temporal response timings and durations of gene expression to a biological condition. We introduce a novel non-stationary Gaussian process as the underlying expression model, with major improvements on model fitness on perturbation and stress experiments. The method is robust to uneven or sp...
Vers Un Physiome XML Pour Le Physiome Rénal
In biology, an enormous quantity of data has been generated over the last 20 years and is stored ... more In biology, an enormous quantity of data has been generated over the last 20 years and is stored in various sources around the world. Until very recently, each source was dedicated to a specific type of data. Now however, biologists have an increasing need to integrate these different ...
Localized Multiple Sources Self-Organizing Map
Neural Information Processing, 2018
We present in this paper a new approach based on unsupervised self organizing maps called MSSOM. ... more We present in this paper a new approach based on unsupervised self organizing maps called MSSOM. This approach combines multiple heterogeneous data sources and learns the weights of each source at the level of clusters instead of learning the same source weights for the whole space. This allows to improve the performances of our model especially in applications where a local feature selection is important. We evaluate our method using several artificial and real datasets and show competitive results compared to the state-of-art.
MS-LSTMEA: Predicting Clinical Events for Hypertension Using Multi-Sources LSTM Explainable Approach
SSRN Electronic Journal
This paper presents a new growing neural network for sequences clustering and classification. Thi... more This paper presents a new growing neural network for sequences clustering and classification. This network is a self organizing map (SOM), which has the properties of stability and plasticity. The stability concerns the preservation of previously learned knowledge and the plasticity concerns the adaptation to any change in the input environment. These properties are obtained using Adaptive Resonance Theory. In order to take into account the temporal information (the dynamics) and the correlation of the patterns contained in the sequences, the inputs of the map are modelled using their associated dynamic covariance matrices. This new model is inspired from the field of speaker recognition. We have modified a covariance matrix in order to represent a temporal order in the sequence. The experimentations show that our approach is better than some other temporal self organizing map for user's Web navigation classification.
CBR system for sequence prediction: casep
BMC Bioinformatics
Background With the rapid advancement of genomic sequencing techniques, massive production of gen... more Background With the rapid advancement of genomic sequencing techniques, massive production of gene expression data is becoming possible, which prompts the development of precision medicine. Deep learning is a promising approach for phenotype prediction (clinical diagnosis, prognosis, and drug response) based on gene expression profile. Existing deep learning models are usually considered as black-boxes that provide accurate predictions but are not interpretable. However, accuracy and interpretation are both essential for precision medicine. In addition, most models do not integrate the knowledge of the domain. Hence, making deep learning models interpretable for medical applications using prior biological knowledge is the main focus of this paper. Results In this paper, we propose a new self-explainable deep learning model, called Deep GONet, integrating the Gene Ontology into the hierarchical architecture of the neural network. This model is based on a fully-connected architecture ...
Uploads
Papers by Farida Zehraoui