RENE NATOWICZ | ESIEE - Academia.edu (original) (raw)
Papers by RENE NATOWICZ
HAL (Le Centre pour la Communication Scientifique Directe), 1992
European Conference on Artificial Life, Apr 29, 1992
HAL (Le Centre pour la Communication Scientifique Directe), 1995
Many cognitive and behavioral tasks-such as interval timing, spatial navigation, motor control an... more Many cognitive and behavioral tasks-such as interval timing, spatial navigation, motor control and speech-require the execution of preciselytimed sequences of neural activation that cannot be fully explained by a succession of external stimuli. We use a reservoir computing framework to explain how such neural sequences can be generated and employed in temporal tasks. We propose a general solution for recurrent neural networks to autonomously produce rich patterns of activity by providing a multi-periodic oscillatory signal as input. We show that the model accurately learns a variety of tasks, including speech generation, motor control and spatial navigation. Further, the model performs temporal rescaling of natural spoken words and exhibits sequential neural activity commonly found in experimental data involving temporal processing. In the context of spatial navigation, the model learns and replays compressed sequences of place cells and captures features of neural activity such as the emergence of ripples and theta phase precession. Together, our findings suggest that combining oscillatory neuronal inputs with different frequencies provides a key mechanism to generate precisely timed sequences of activity in recurrent circuits of the brain. Reservoir computing | Neural oscillations | Temporal processing | Balanced networks 1. Introduction Virtually every aspect of sensory, cognitive and motor processing in biological organisms involves operations unfolding in time (1). In the brain, neuronal circuits must represent time on a variety of scales, from milliseconds to minutes and longer circadian rhythms (2). Despite increasingly sophisticated models of brain activity, time representation remains a challenging problem in computational modelling (3, 4). Recurrent neural networks offer a promising avenue to detect and produce precisely timed sequences of activity (5). However, it is challenging to train these networks due to their complexity (6), particularly when operating in a chaotic regime associated with biological neural networks (7, 8). One avenue to address this issue is with the use of reservoir computing (RC) (9, 10). Under this framework, a recurrent network (the reservoir) projects onto a read-out layer whose synaptic weights are adjusted to produce a desired response. However, while RC can capture some behavioral and cognitive processes (11-13), it often relies on biologically implausible mechanisms (14). Further, current RC implementations offer little insight to understand how the brain generates activity that does not follow a strict rhythmic pattern (1, 5). That is because RC models are either restricted to learning periodic functions, or require an aperiodic input to generate an aperiodic output, thus leaving the neural origins of aperiodic activity unresolved (5). A solution to this problem is to train the recurrent connections of the reservoir to stabilize innate patterns of activity (12), but this approach is more computationally expensive and is sensitive to structural perturbations (15). To address these limitations, we propose a biologically plausible spiking recurrent model that receives multiple independent sources of neural oscillations as input. The combination of oscillators with different periods creates a multi-periodic code that serves as a time-varying input that can largely exceed the period of any of its individual components. We show that this input can be generated endogenously by distinct subnetworks, alleviating the need to train recurrent connections of the reservoir to generate long segments of aperiodic activity. Further, no feedback from the output units to the reservoir was required during training (11, 16). Thus, multiplexing a set of oscillators into a reservoir network provides an efficient and neurophysiologically grounded means of controlling a recurrent circuit (15). Analogous mechanisms have been hypothesized in other contexts including grid cell representations (17) and interval timing (18, 19). This paper is structured as follows. First, we describe a simplified scenario where a reservoir network that receives a collection of input oscillations learns to reproduce an arbitrary time-evolving signal. Second, we extend the model to show how oscillations can be generated intrinsically by oscillatory networks that can be either embedded or external to the reservoir. Third, we show that a network can learn several tasks in parallel by "tagging" each task to a particular phase configuration of the oscillatory inputs. Fourth, we show that the activity of the reservoir network captures temporal rescaling and selectivity, two features of neural activity reported during behavioral tasks. Fifth, we train the model to reproduce natural speech at different speeds when cued by input oscillations. Finally, we employ a variant of the model to capture hippocampal activity during spatial navigation. Together, results highlight a novel role for neural oscillations in regulating temporal processing within recurrent networks of the brain. 2. Results A. A reservoir network driven by oscillations. We began with a basic implementation of our model where artificial oscillations served as input to a reservoir network (Fig. 1a)-in a later section, we will describe a more realistic version where recurrent networks generate these oscillations intrinsically. In this simplified model, two input nodes, but potentially more (Fig. S1),
ABSTRACT The authors propose a non-supervised model of formal neural networks to learn and recogn... more ABSTRACT The authors propose a non-supervised model of formal neural networks to learn and recognize temporal sequences. Time is represented by its effect on processing and not as an additional dimension of inputs. Synaptic efficacy of a connection is the integration time of the signal passing through the connection. The only parameters subject to learning are connection integration times. It is assumed that any cell of the network can have a spontaneous and an evoked activity. Under this assumption such networks can, in an unsupervised way, learn and recognize temporal sequences. An example of such a network is described and the results of the simulation are discussed
ABSTRACT We propose an architecture for segmenting sequences of images in grey levels at video ra... more ABSTRACT We propose an architecture for segmenting sequences of images in grey levels at video rate. This architecture implements on field programmable gate arrays an algorithm of image segmentation by self-organising feature maps. Performance and costs of the architecture are evaluated, reconfigurability of the proposed architecture is discussed.
Soft Computing, Mar 12, 2010
Breast Cancer Research and Treatment, Aug 14, 2012
Simulation Practice and Theory, Apr 1, 2000
Calcium plays an essential role as a messenger and as a factor in cardiac contraction. In the pre... more Calcium plays an essential role as a messenger and as a factor in cardiac contraction. In the present study, a model for Ca 2 handling in cardiac cells is presented. After the identi®cation of the sarcoplasmic reticulum (SR) parameters, the SERCA pump and ryanodine channels activities, a comparison is made between experimental and calculated responses. The model's parameters were identi®ed using optimization methods. This identi®cation is based on the response of the digitonin permeabilized cells. The model deals with the dynamics of the calcium exchange between the dierent compartments of the cell. Cell compartments involved are the SR, the cytosol and the extra-cellular medium. The dierent components of the mathematical models are discussed and compared. The modeling and simulation are run within Wlab, 1 a freeware for modeling and simulation of dynamic systems.
Mechanical Engineering, 2018
This paper addresses the problem of optimal control and scheduling of Networked Control Systems o... more This paper addresses the problem of optimal control and scheduling of Networked Control Systems over limited bandwidth deterministic networks using some insight on the interplay between the control and information theory. The motivation is related to the necessity of choice a communication sequence which maximize control signal impact on the plant behavior. The solution is obtained by decomposing the overall problem in a twofold one. The first level problem aims obtaining the periodic off-line or static scheduling function of control signals based on system properties, communication constrains, periodicity of scheduling sequence, performance criteria and maximization of the degree of reachability/observability of the periodic system. A Mixed Integer Quadratic Programming (MIQP) problem is formulated and solved obtaining a periodic and stable NCS. The solution of the second level problem is based on the structure of the static scheduling function obtained from the first level solutio...
2011 14th International IEEE Conference on Intelligent Transportation Systems (ITSC), 2011
... and H. Mounier are with Laboratoire des Signaux et Systemes, CNRS/Supelec, 91192, Gif-Sur-Yve... more ... and H. Mounier are with Laboratoire des Signaux et Systemes, CNRS/Supelec, 91192, Gif-Sur-Yvettes Cedex, France silviu.niculescu@lss ... terrain information ahead of the controlled vehicle can be obtained from ITS like systems SYTADIN [29] or Traffic and Radar Software [30 ...
HAL (Le Centre pour la Communication Scientifique Directe), 1992
Background & Purpose: Isoforms of metabolic enzymes share similar catalytic activities. M... more Background & Purpose: Isoforms of metabolic enzymes share similar catalytic activities. Most cells express several isoforms that likely represent redundancy and contribute robustness to biochemical processes. The distribution of isoforms varies by tissue and differentiation stage and may be altered during neoplastic transformation. Our hypothesis is iosenzyme expression changes in neoplastic cells, particularly the loss of isoenzyme diversity, can render cancer cells more vulnerable to metabolism targeted therapies. Our goal was to identify isoenzyme expression shifts in breast cancer compared to normal breast tissue to define metabolic processes in the cancer that rely on the expression of only one or a few enzyme isoforms due to loss of isoenzyme diversity. Methods: Metabolic enzymes (n=1,267) were identified in the KEGG database and we established pair-wise isoform expression distributions in breast cancer gene expression data sets (n=1,081 cancers) and in 40 normal breast samples. We looked for isoenzyme pairs that showed loss of expression in one isoform in cancer compared to normal while the other isoform was preserved or overexpressed. We assessed the isoform expression relationships in breast cancer cell lines (n=18) to identify experimental models for functional studies and used siRNAs to knock down the preserved isoform in cell lines that showed loss of the other isoform. Results: We identified 98 pairs of isoenzymes that showed reduced expression in one isoform in at least 20% of cancer patients compared to normal tissue (< 3 standard deviations below the mean of normal expression, the lower boundary of the same gene in normal tissues), while the expression of the second, larger than its lower boundary in normal tissues. These included PAPP2B/PAPP2C, COX7A1/COX7A2, ALDOC/ALDOA, LDHB/LDHA, and ME3/ME2 where the first member of the pair showed reduced expression in 83.7%, 48.2%, 33.0%, 49.4%, and 46.9% of cancers, respectively. Overall cancers shifted towards expressing isoforms that are normally high in muscle for many energy metabolism related enzymes. We tested the functional implication of knocking down ME2 (malic enzyme 2) expression in MDA435 cells that showed reduced ME3 expression but high ME2 levels. We observed a 54.0%, growth inhibition compared to scrambled siRNA control. In the control cell lines, MDA231 and MDA 468 that have preserved ME3 and ME2 expression, ME2 knockdown had no significant growth inhibitory effect. Conclusions: We demonstrate shifts in isoenzyme expression in breast cancers compare to normal breast. These shifts suggest adaptation to hypoxic environment, rapid metabolism and potential metabolic vulnerabilities. Our siRNA experiments with anti ME2 siRNA provide proof of concept that loss of isoenzyme diversity and primary reliance on a single or few isoenzymes to provide a metabolic function in neoplastic cells may be exploited therapeutically. Citation Format: Weiwei Shi, Aihua Gong, Tingting Jiang, Rene Natowicz, Lajos Pusztai. Metabolic vulnerability of breast cancer based on shifts in the expression of metabolic enzyme isoforms. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 1899. doi:10.1158/1538-7445.AM2013-1899
16th International IEEE Conference on Intelligent Transportation Systems (ITSC 2013), 2013
International audienc
2019 International Conference in Engineering Applications (ICEA), 2019
Old-fashioned street lighting strategies, i.e. always on street lights, can pose a significant en... more Old-fashioned street lighting strategies, i.e. always on street lights, can pose a significant energy wasting, especially in the areas with low amount of traffic at night. This paper presents review of several Smart Street Lighting (SmSL) strategies, which promise energy saving. For selected strategy, also known as Light on Demand (LoD), a energy consumption models for defined scenario are developed and the results are presented.
ABSTRACT Nous proposons un algorithme de sélection de caractéristiques (feature selection) à haut... more ABSTRACT Nous proposons un algorithme de sélection de caractéristiques (feature selection) à haute dimension et son application en onco-pharmacogénomique pour le cancer du sein. Dans cette application, nous devons sélectionner un ensemble de gènes dont les niveaux d'expression permettent une prédiction efficace de la réponse des patientes à un traitement de chimiothérapie préopératoire. L'algorithme de sélection de caractéristiques que nous proposons est issu d'une heuristique d'optimisation de type line search, développée pour les problèmes à grande dimensionnalité. Cet algorithme développé pour des espaces continus est ici transposé aux espaces discrets et appliqué à la sélection de sous-ensembles de sondes à ADN.
Abstract. This paper presents an evolutionary wrapper method for feature selection that uses a no... more Abstract. This paper presents an evolutionary wrapper method for feature selection that uses a non-parametric density estimation method and a Bayesian Classifier. Non-parametric methods are a good alternative for scarce and sparse data, as in Bioinformatics problems, since they do not make any assumptions about its structure and all the information come from data itself. Results show that local modeling provides small and relevant subsets of features when comparing to results available on literature. 1
Journal of Clinical Oncology, 2013
1013 Background: The goal of this study was to develop a method to quantify intratumor heterogene... more 1013 Background: The goal of this study was to develop a method to quantify intratumor heterogeneity of cancers using gene expression data. We compared gene expression heterogeneity between different molecular subtypes of breast cancer and between basal like cancers with or without pathologic complete response (pCR) to neoadjuvant chemotherapy. Methods: Affymetrix U133A gene expression data of 335 stage I-III breast cancers were analyzed. Molecular class was assigned using the PAM50 predictor. All patients received neoadjuvant chemotherapy. We measured tumor heterogeneity by the Gini index (GI) calculated individually for each case over the expression of all probe sets and random subsets. The GI was used as a metric of inequality of gene expression distributions between cases. The higher the GI, the greater the inequality of the expression distribution. Results: Basal like cancers (n=138) had greater heterogeneity than luminal cancers (n=197) (mean GI values 24.51 vs 23.05, p<0.0...
Cancer Research, 2013
Background: The goal of this analysis was to develop a computational tool that integrates gene ex... more Background: The goal of this analysis was to develop a computational tool that integrates gene expression, DNA copy number and sequence abnormalities in individual cancers in the framework of biological processes. Methods and Findings: We used the hierarchical structure of the Gene Ontology (GO) database to create a reference network and projected mRNA expression, DNA copy number and mutation anomalies detected in single samples into this space. We applied our method to 59 breast cancers where all three types of molecular data were available. Each cancer had a large number of disturbed biological processes. Locomotion, multicellular organismal process and signal transduction pathways were the most commonly affected GO terms but the individual molecular events were different from case-to-case. Estrogen receptor-positive and -negative cancers had different repertoire of anomalies. We tested the functional impact of 27 mRNAs that had overexpression in cancer with variable frequency ( Conclusions We developed a free, on-line software tool (http://netgoplot.org) to display the complex genomic abnormalities in individual cancers in the biological framework of the GO biological processes. We show that each cancer harbors a large number and unique combination of pathway anomalies, the individual molecular events that underlie the pathway level disturbances vary from case-to-case. Our in vitro experiments indicate that even rare case-specific molecular abnormalities can play a functional role and driver events may vary from case to case depending on the constellation of other molecular anomalies. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr PD4-6.
HAL (Le Centre pour la Communication Scientifique Directe), 1992
European Conference on Artificial Life, Apr 29, 1992
HAL (Le Centre pour la Communication Scientifique Directe), 1995
Many cognitive and behavioral tasks-such as interval timing, spatial navigation, motor control an... more Many cognitive and behavioral tasks-such as interval timing, spatial navigation, motor control and speech-require the execution of preciselytimed sequences of neural activation that cannot be fully explained by a succession of external stimuli. We use a reservoir computing framework to explain how such neural sequences can be generated and employed in temporal tasks. We propose a general solution for recurrent neural networks to autonomously produce rich patterns of activity by providing a multi-periodic oscillatory signal as input. We show that the model accurately learns a variety of tasks, including speech generation, motor control and spatial navigation. Further, the model performs temporal rescaling of natural spoken words and exhibits sequential neural activity commonly found in experimental data involving temporal processing. In the context of spatial navigation, the model learns and replays compressed sequences of place cells and captures features of neural activity such as the emergence of ripples and theta phase precession. Together, our findings suggest that combining oscillatory neuronal inputs with different frequencies provides a key mechanism to generate precisely timed sequences of activity in recurrent circuits of the brain. Reservoir computing | Neural oscillations | Temporal processing | Balanced networks 1. Introduction Virtually every aspect of sensory, cognitive and motor processing in biological organisms involves operations unfolding in time (1). In the brain, neuronal circuits must represent time on a variety of scales, from milliseconds to minutes and longer circadian rhythms (2). Despite increasingly sophisticated models of brain activity, time representation remains a challenging problem in computational modelling (3, 4). Recurrent neural networks offer a promising avenue to detect and produce precisely timed sequences of activity (5). However, it is challenging to train these networks due to their complexity (6), particularly when operating in a chaotic regime associated with biological neural networks (7, 8). One avenue to address this issue is with the use of reservoir computing (RC) (9, 10). Under this framework, a recurrent network (the reservoir) projects onto a read-out layer whose synaptic weights are adjusted to produce a desired response. However, while RC can capture some behavioral and cognitive processes (11-13), it often relies on biologically implausible mechanisms (14). Further, current RC implementations offer little insight to understand how the brain generates activity that does not follow a strict rhythmic pattern (1, 5). That is because RC models are either restricted to learning periodic functions, or require an aperiodic input to generate an aperiodic output, thus leaving the neural origins of aperiodic activity unresolved (5). A solution to this problem is to train the recurrent connections of the reservoir to stabilize innate patterns of activity (12), but this approach is more computationally expensive and is sensitive to structural perturbations (15). To address these limitations, we propose a biologically plausible spiking recurrent model that receives multiple independent sources of neural oscillations as input. The combination of oscillators with different periods creates a multi-periodic code that serves as a time-varying input that can largely exceed the period of any of its individual components. We show that this input can be generated endogenously by distinct subnetworks, alleviating the need to train recurrent connections of the reservoir to generate long segments of aperiodic activity. Further, no feedback from the output units to the reservoir was required during training (11, 16). Thus, multiplexing a set of oscillators into a reservoir network provides an efficient and neurophysiologically grounded means of controlling a recurrent circuit (15). Analogous mechanisms have been hypothesized in other contexts including grid cell representations (17) and interval timing (18, 19). This paper is structured as follows. First, we describe a simplified scenario where a reservoir network that receives a collection of input oscillations learns to reproduce an arbitrary time-evolving signal. Second, we extend the model to show how oscillations can be generated intrinsically by oscillatory networks that can be either embedded or external to the reservoir. Third, we show that a network can learn several tasks in parallel by "tagging" each task to a particular phase configuration of the oscillatory inputs. Fourth, we show that the activity of the reservoir network captures temporal rescaling and selectivity, two features of neural activity reported during behavioral tasks. Fifth, we train the model to reproduce natural speech at different speeds when cued by input oscillations. Finally, we employ a variant of the model to capture hippocampal activity during spatial navigation. Together, results highlight a novel role for neural oscillations in regulating temporal processing within recurrent networks of the brain. 2. Results A. A reservoir network driven by oscillations. We began with a basic implementation of our model where artificial oscillations served as input to a reservoir network (Fig. 1a)-in a later section, we will describe a more realistic version where recurrent networks generate these oscillations intrinsically. In this simplified model, two input nodes, but potentially more (Fig. S1),
ABSTRACT The authors propose a non-supervised model of formal neural networks to learn and recogn... more ABSTRACT The authors propose a non-supervised model of formal neural networks to learn and recognize temporal sequences. Time is represented by its effect on processing and not as an additional dimension of inputs. Synaptic efficacy of a connection is the integration time of the signal passing through the connection. The only parameters subject to learning are connection integration times. It is assumed that any cell of the network can have a spontaneous and an evoked activity. Under this assumption such networks can, in an unsupervised way, learn and recognize temporal sequences. An example of such a network is described and the results of the simulation are discussed
ABSTRACT We propose an architecture for segmenting sequences of images in grey levels at video ra... more ABSTRACT We propose an architecture for segmenting sequences of images in grey levels at video rate. This architecture implements on field programmable gate arrays an algorithm of image segmentation by self-organising feature maps. Performance and costs of the architecture are evaluated, reconfigurability of the proposed architecture is discussed.
Soft Computing, Mar 12, 2010
Breast Cancer Research and Treatment, Aug 14, 2012
Simulation Practice and Theory, Apr 1, 2000
Calcium plays an essential role as a messenger and as a factor in cardiac contraction. In the pre... more Calcium plays an essential role as a messenger and as a factor in cardiac contraction. In the present study, a model for Ca 2 handling in cardiac cells is presented. After the identi®cation of the sarcoplasmic reticulum (SR) parameters, the SERCA pump and ryanodine channels activities, a comparison is made between experimental and calculated responses. The model's parameters were identi®ed using optimization methods. This identi®cation is based on the response of the digitonin permeabilized cells. The model deals with the dynamics of the calcium exchange between the dierent compartments of the cell. Cell compartments involved are the SR, the cytosol and the extra-cellular medium. The dierent components of the mathematical models are discussed and compared. The modeling and simulation are run within Wlab, 1 a freeware for modeling and simulation of dynamic systems.
Mechanical Engineering, 2018
This paper addresses the problem of optimal control and scheduling of Networked Control Systems o... more This paper addresses the problem of optimal control and scheduling of Networked Control Systems over limited bandwidth deterministic networks using some insight on the interplay between the control and information theory. The motivation is related to the necessity of choice a communication sequence which maximize control signal impact on the plant behavior. The solution is obtained by decomposing the overall problem in a twofold one. The first level problem aims obtaining the periodic off-line or static scheduling function of control signals based on system properties, communication constrains, periodicity of scheduling sequence, performance criteria and maximization of the degree of reachability/observability of the periodic system. A Mixed Integer Quadratic Programming (MIQP) problem is formulated and solved obtaining a periodic and stable NCS. The solution of the second level problem is based on the structure of the static scheduling function obtained from the first level solutio...
2011 14th International IEEE Conference on Intelligent Transportation Systems (ITSC), 2011
... and H. Mounier are with Laboratoire des Signaux et Systemes, CNRS/Supelec, 91192, Gif-Sur-Yve... more ... and H. Mounier are with Laboratoire des Signaux et Systemes, CNRS/Supelec, 91192, Gif-Sur-Yvettes Cedex, France silviu.niculescu@lss ... terrain information ahead of the controlled vehicle can be obtained from ITS like systems SYTADIN [29] or Traffic and Radar Software [30 ...
HAL (Le Centre pour la Communication Scientifique Directe), 1992
Background & Purpose: Isoforms of metabolic enzymes share similar catalytic activities. M... more Background & Purpose: Isoforms of metabolic enzymes share similar catalytic activities. Most cells express several isoforms that likely represent redundancy and contribute robustness to biochemical processes. The distribution of isoforms varies by tissue and differentiation stage and may be altered during neoplastic transformation. Our hypothesis is iosenzyme expression changes in neoplastic cells, particularly the loss of isoenzyme diversity, can render cancer cells more vulnerable to metabolism targeted therapies. Our goal was to identify isoenzyme expression shifts in breast cancer compared to normal breast tissue to define metabolic processes in the cancer that rely on the expression of only one or a few enzyme isoforms due to loss of isoenzyme diversity. Methods: Metabolic enzymes (n=1,267) were identified in the KEGG database and we established pair-wise isoform expression distributions in breast cancer gene expression data sets (n=1,081 cancers) and in 40 normal breast samples. We looked for isoenzyme pairs that showed loss of expression in one isoform in cancer compared to normal while the other isoform was preserved or overexpressed. We assessed the isoform expression relationships in breast cancer cell lines (n=18) to identify experimental models for functional studies and used siRNAs to knock down the preserved isoform in cell lines that showed loss of the other isoform. Results: We identified 98 pairs of isoenzymes that showed reduced expression in one isoform in at least 20% of cancer patients compared to normal tissue (< 3 standard deviations below the mean of normal expression, the lower boundary of the same gene in normal tissues), while the expression of the second, larger than its lower boundary in normal tissues. These included PAPP2B/PAPP2C, COX7A1/COX7A2, ALDOC/ALDOA, LDHB/LDHA, and ME3/ME2 where the first member of the pair showed reduced expression in 83.7%, 48.2%, 33.0%, 49.4%, and 46.9% of cancers, respectively. Overall cancers shifted towards expressing isoforms that are normally high in muscle for many energy metabolism related enzymes. We tested the functional implication of knocking down ME2 (malic enzyme 2) expression in MDA435 cells that showed reduced ME3 expression but high ME2 levels. We observed a 54.0%, growth inhibition compared to scrambled siRNA control. In the control cell lines, MDA231 and MDA 468 that have preserved ME3 and ME2 expression, ME2 knockdown had no significant growth inhibitory effect. Conclusions: We demonstrate shifts in isoenzyme expression in breast cancers compare to normal breast. These shifts suggest adaptation to hypoxic environment, rapid metabolism and potential metabolic vulnerabilities. Our siRNA experiments with anti ME2 siRNA provide proof of concept that loss of isoenzyme diversity and primary reliance on a single or few isoenzymes to provide a metabolic function in neoplastic cells may be exploited therapeutically. Citation Format: Weiwei Shi, Aihua Gong, Tingting Jiang, Rene Natowicz, Lajos Pusztai. Metabolic vulnerability of breast cancer based on shifts in the expression of metabolic enzyme isoforms. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 1899. doi:10.1158/1538-7445.AM2013-1899
16th International IEEE Conference on Intelligent Transportation Systems (ITSC 2013), 2013
International audienc
2019 International Conference in Engineering Applications (ICEA), 2019
Old-fashioned street lighting strategies, i.e. always on street lights, can pose a significant en... more Old-fashioned street lighting strategies, i.e. always on street lights, can pose a significant energy wasting, especially in the areas with low amount of traffic at night. This paper presents review of several Smart Street Lighting (SmSL) strategies, which promise energy saving. For selected strategy, also known as Light on Demand (LoD), a energy consumption models for defined scenario are developed and the results are presented.
ABSTRACT Nous proposons un algorithme de sélection de caractéristiques (feature selection) à haut... more ABSTRACT Nous proposons un algorithme de sélection de caractéristiques (feature selection) à haute dimension et son application en onco-pharmacogénomique pour le cancer du sein. Dans cette application, nous devons sélectionner un ensemble de gènes dont les niveaux d'expression permettent une prédiction efficace de la réponse des patientes à un traitement de chimiothérapie préopératoire. L'algorithme de sélection de caractéristiques que nous proposons est issu d'une heuristique d'optimisation de type line search, développée pour les problèmes à grande dimensionnalité. Cet algorithme développé pour des espaces continus est ici transposé aux espaces discrets et appliqué à la sélection de sous-ensembles de sondes à ADN.
Abstract. This paper presents an evolutionary wrapper method for feature selection that uses a no... more Abstract. This paper presents an evolutionary wrapper method for feature selection that uses a non-parametric density estimation method and a Bayesian Classifier. Non-parametric methods are a good alternative for scarce and sparse data, as in Bioinformatics problems, since they do not make any assumptions about its structure and all the information come from data itself. Results show that local modeling provides small and relevant subsets of features when comparing to results available on literature. 1
Journal of Clinical Oncology, 2013
1013 Background: The goal of this study was to develop a method to quantify intratumor heterogene... more 1013 Background: The goal of this study was to develop a method to quantify intratumor heterogeneity of cancers using gene expression data. We compared gene expression heterogeneity between different molecular subtypes of breast cancer and between basal like cancers with or without pathologic complete response (pCR) to neoadjuvant chemotherapy. Methods: Affymetrix U133A gene expression data of 335 stage I-III breast cancers were analyzed. Molecular class was assigned using the PAM50 predictor. All patients received neoadjuvant chemotherapy. We measured tumor heterogeneity by the Gini index (GI) calculated individually for each case over the expression of all probe sets and random subsets. The GI was used as a metric of inequality of gene expression distributions between cases. The higher the GI, the greater the inequality of the expression distribution. Results: Basal like cancers (n=138) had greater heterogeneity than luminal cancers (n=197) (mean GI values 24.51 vs 23.05, p<0.0...
Cancer Research, 2013
Background: The goal of this analysis was to develop a computational tool that integrates gene ex... more Background: The goal of this analysis was to develop a computational tool that integrates gene expression, DNA copy number and sequence abnormalities in individual cancers in the framework of biological processes. Methods and Findings: We used the hierarchical structure of the Gene Ontology (GO) database to create a reference network and projected mRNA expression, DNA copy number and mutation anomalies detected in single samples into this space. We applied our method to 59 breast cancers where all three types of molecular data were available. Each cancer had a large number of disturbed biological processes. Locomotion, multicellular organismal process and signal transduction pathways were the most commonly affected GO terms but the individual molecular events were different from case-to-case. Estrogen receptor-positive and -negative cancers had different repertoire of anomalies. We tested the functional impact of 27 mRNAs that had overexpression in cancer with variable frequency ( Conclusions We developed a free, on-line software tool (http://netgoplot.org) to display the complex genomic abnormalities in individual cancers in the biological framework of the GO biological processes. We show that each cancer harbors a large number and unique combination of pathway anomalies, the individual molecular events that underlie the pathway level disturbances vary from case-to-case. Our in vitro experiments indicate that even rare case-specific molecular abnormalities can play a functional role and driver events may vary from case to case depending on the constellation of other molecular anomalies. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr PD4-6.