Papers by Nicolas Chaumont
Ce mémoire traite de deux problèmes en infographie, l'un en modélisation de surfaces implicit... more Ce mémoire traite de deux problèmes en infographie, l'un en modélisation de surfaces implicites et l'autre en animation d'acteurs virtuels. Dans le premier chapitre sont présentés les résultats obtenus en faisant évoluer les paramètres d'un modèle de blobbies (mixture de gaussiennes en 3D) pour interpoler un nuage de points de configuration arbitraire. Le second chapitre décrit des travaux portant sur la génération automatique d'acteurs virtuels dont la morphologie et le comportement sont optimisés simultanément à l'aide d'un algorithme génétique pour effectuer des tâches particulières. La recherche décrite dans ce chapitre s'appuie sur l'élaboration d'une plateforme d'expérimentation, destinée à investiguer des questions où la co-évolution de la morphologie et du contrôleur occupe une place centrale, en l'exploitant à travers deux expérimentations. La première reproduit partiellement des résultats antérieurs obtenus en optimisant la capacité de locomotion d'assemblages aléatoires de blocs parallélépipédiques (que l'on interprète comme étant des créatures virtuelles) reliés entre eux par des joints munis de moteurs. Dans la seconde, chaque individu a la capacité de détacher un bloc précis après un délai variable, lui permettant ainsi d'exploiter diverses stratégies de jet. Le déplacement de chaque individu est soumis à un ensemble de lois physiques (dynamique des corps rigides) recréées à l'intérieur d'un simulateur. Les limitations imposées par le modèle physique sur la liberté de mouvement des individus sont relativement proches de celles qui agissent sur les objets dans la réalité. Cette limitation est en fait l'aspect crucial à l'origine du réalisme du comportement des individus. La méthode d'évolution que nous utilisons tout au long de ce mémoire est basée sur le principe des algorithmes génétiques de type élitiste. Elle permet de manipuler des structures de données directement, contrairement à la forme canonique de l'algorithme génétique qui ne peut manipuler qu'une séquence de bits
Please see read file for data file descriptor
Fresh produce has become the primary cause of foodborne illness in the United States. A widesprea... more Fresh produce has become the primary cause of foodborne illness in the United States. A widespread concern that wildlife vector foodborne pathogens onto fresh produce fields has led to strong pressure on farmers to clear noncrop vegetation surrounding their farm fields. We combined three large datasets to demonstrate that pathogen prevalence in fresh produce is rapidly increasing, that pathogens are more common on farms closer to land suitable for livestock grazing, and that vegetation clearing is associated with increased pathogen prevalence over time. These findings contradict widespread food safety reforms that champion vegetation clearing as a pathogen mitigation strategy. More generally, our work indicates that achieving food safety and nature conservation goals in produce-growing landscapes is possible
Proceedings of the National Academy of Sciences of the United States of America, Jan 10, 2015
In 2006, a deadly Escherichia coli O157:H7 outbreak in bagged spinach was traced to California... more In 2006, a deadly Escherichia coli O157:H7 outbreak in bagged spinach was traced to California's Central Coast region, where >70% of the salad vegetables sold in the United States are produced. Although no definitive cause for the outbreak could be determined, wildlife was implicated as a disease vector. Growers were subsequently pressured to minimize the intrusion of wildlife onto their farm fields by removing surrounding noncrop vegetation. How vegetation removal actually affects foodborne pathogens remains unknown, however. We combined a fine-scale land use map with three datasets comprising ∼250,000 enterohemorrhagic E. coli (EHEC), generic E. coli, and Salmonella tests in produce, irrigation water, and rodents to quantify whether seminatural vegetation surrounding farmland is associated with foodborne pathogen prevalence in California's Central Coast region. We found that EHEC in fresh produce increased by more than an order of magnitude from 2007 to 2013, despite ex...
Genetic Programming and Evolvable Machines, Jun 7, 2016
Artificially evolving foraging behavior in simulated articulated animals has proved to be a notor... more Artificially evolving foraging behavior in simulated articulated animals has proved to be a notoriously difficult task. Here, we co-evolve the morphology and controller for virtual organisms in a three-dimensional physical environment to produce goal-directed locomotion in articulated agents. We show that following and reaching multiple food sources can evolve de novo, by evaluating each organism on multiple food sources placed on a basic pattern that is gradually randomized across generations. We devised a strategy of evolutionary ''staging'', where the best organism from a set of evolutionary experiments using a particular fitness function is used to seed a new set, with a fitness function that is progressively altered to better challenge organisms as evolution improves them. We find that an organism's efficiency at reaching the first food source does not predict its ability at finding subsequent ones because foraging efficiency crucially depends on the position of the last food source reached, an effect illustrated by ''foraging maps'' that capture the organism's controller state, body position, and orientation. Our best evolved foragers are able to reach multiple food sources over 90 % of the time on average, a behavior that is key to any biologically realistic simulation where a self-sustaining population has to survive by collecting food sources in three-dimensional, physical environments.
arXiv (Cornell University), Mar 9, 2011
One of the hallmarks of biological organisms is their ability to integrate disparate information ... more One of the hallmarks of biological organisms is their ability to integrate disparate information sources to optimize their behavior in complex environments. How this capability can be quantified and related to the functional complexity of an organism remains a challenging problem, in particular since organismal functional complexity is not well-defined. We present here several candidate measures that quantify information and integration, and study their dependence on fitness as an artificial agent ("animat") evolves over thousands of generations to solve a navigation task in a simple, simulated environment. We compare the ability of these measures to predict high fitness with more conventional information-theoretic processing measures. As the animat adapts by increasing its "fit" to the world, information integration and processing increase commensurately along the evolutionary line of descent. We suggest that the correlation of fitness with information integration and with processing measures implies that high fitness requires both information processing as well as integration, but that information integration may be a better measure when the task requires memory. A correlation of measures of information integration (but also information processing) and fitness strongly suggests that these measures reflect the functional complexity of the animat, and that such measures can be used to quantify functional complexity even in the absence of fitness data.
arXiv (Cornell University), Dec 21, 2011
Marine Policy, Dec 1, 2015
Concern about the visibility of large infrastructure development often drives public opposition t... more Concern about the visibility of large infrastructure development often drives public opposition to these projects. However, insufficient analytical tools to assess visibility across a large number of alternate sites prior to siting typically results in the omission of visibility in multi-criteria siting processes, leading to inferior site selection and often costly litigation. This paper presents an approach for deriving visibility maps based on the location and duration of viewing by residents and visitors and demonstrates its use in illuminating tradeoffs by comparing these maps to wind energy value maps in the context of offshore wind energy development in the Northeastern United States.
arXiv (Cornell University), Dec 21, 2011
Ce mémoire traite de deux problèmes en infographie, l'un en modélisation de surfaces implicit... more Ce mémoire traite de deux problèmes en infographie, l'un en modélisation de surfaces implicites et l'autre en animation d'acteurs virtuels. Dans le premier chapitre sont présentés les résultats obtenus en faisant évoluer les paramètres d'un modèle de blobbies (mixture de gaussiennes en 3D) pour interpoler un nuage de points de configuration arbitraire. Le second chapitre décrit des travaux portant sur la génération automatique d'acteurs virtuels dont la morphologie et le comportement sont optimisés simultanément à l'aide d'un algorithme génétique pour effectuer des tâches particulières. La recherche décrite dans ce chapitre s'appuie sur l'élaboration d'une plateforme d'expérimentation, destinée à investiguer des questions où la co-évolution de la morphologie et du contrôleur occupe une place centrale, en l'exploitant à travers deux expérimentations. La première reproduit partiellement des résultats antérieurs obtenus en optimisant la capacité de locomotion d'assemblages aléatoires de blocs parallélépipédiques (que l'on interprète comme étant des créatures virtuelles) reliés entre eux par des joints munis de moteurs. Dans la seconde, chaque individu a la capacité de détacher un bloc précis après un délai variable, lui permettant ainsi d'exploiter diverses stratégies de jet. Le déplacement de chaque individu est soumis à un ensemble de lois physiques (dynamique des corps rigides) recréées à l'intérieur d'un simulateur. Les limitations imposées par le modèle physique sur la liberté de mouvement des individus sont relativement proches de celles qui agissent sur les objets dans la réalité. Cette limitation est en fait l'aspect crucial à l'origine du réalisme du comportement des individus. La méthode d'évolution que nous utilisons tout au long de ce mémoire est basée sur le principe des algorithmes génétiques de type élitiste. Elle permet de manipuler des structures de données directement, contrairement à la forme canonique de l'algorithme génétique qui ne peut manipuler qu'une séquence de bits
Electronic Supplement to the manuscript "Evolution of Sustained Foraging in Three-Dimensiona... more Electronic Supplement to the manuscript "Evolution of Sustained Foraging in Three-Dimensional Environments with Physics" in the journal "Genetic Programming and Evolvable Machines". The supplement describes the EVO platform and the genetic language that is used in EVO.
One of the hallmarks of biological organisms is their ability to integrate disparate information ... more One of the hallmarks of biological organisms is their ability to integrate disparate information sources to optimize their behavior in complex environments. How this capability can be quantified and related to the functional complexity of an organism remains a challenging problem, in particular since organismal functional complexity is not well-defined. We present here several candidate measures that quantify information and integration, and study their dependence on fitness as an artificial agent ("animat") evolves over thousands of generations to solve a navigation task in a simple, simulated environment. We compare the ability of these measures to predict high fitness with more conventional information-theoretic processing measures. As the animat adapts by increasing its "fit" to the world, information integration and processing increase commensurately along the evolutionary line of descent. We suggest that the correlation of fitness with information integration and with processing measures implies that high fitness requires both information processing as well as integration, but that information integration may be a better measure when the task requires memory. A correlation of measures of information integration (but also information processing) and fitness strongly suggests that these measures reflect the functional complexity of the animat, and that such measures can be used to quantify functional complexity even in the absence of fitness data.
Genetic Programming and Evolvable Machines, 2016
Artificially evolving foraging behavior in simulated articulated animals has proved to be a notor... more Artificially evolving foraging behavior in simulated articulated animals has proved to be a notoriously difficult task. Here, we co-evolve the morphology and controller for virtual organisms in a three-dimensional physical environment to produce goal-directed locomotion in articulated agents. We show that following and reaching multiple food sources can evolve de novo, by evaluating each organism on multiple food sources placed on a basic pattern that is gradually randomized across generations. We devised a strategy of evolutionary ''staging'', where the best organism from a set of evolutionary experiments using a particular fitness function is used to seed a new set, with a fitness function that is progressively altered to better challenge organisms as evolution improves them. We find that an organism's efficiency at reaching the first food source does not predict its ability at finding subsequent ones because foraging efficiency crucially depends on the position of the last food source reached, an effect illustrated by ''foraging maps'' that capture the organism's controller state, body position, and orientation. Our best evolved foragers are able to reach multiple food sources over 90 % of the time on average, a behavior that is key to any biologically realistic simulation where a self-sustaining population has to survive by collecting food sources in three-dimensional, physical environments.
Marine Policy, 2015
Concern about the visibility of large infrastructure development often drives public opposition t... more Concern about the visibility of large infrastructure development often drives public opposition to these projects. However, insufficient analytical tools to assess visibility across a large number of alternate sites prior to siting typically results in the omission of visibility in multi-criteria siting processes, leading to inferior site selection and often costly litigation. This paper presents an approach for deriving visibility maps based on the location and duration of viewing by residents and visitors and demonstrates its use in illuminating tradeoffs by comparing these maps to wind energy value maps in the context of offshore wind energy development in the Northeastern United States.
This paper addresses the problem of generating a volumetric shape to approximate an arbitrary clo... more This paper addresses the problem of generating a volumetric shape to approximate an arbitrary cloud of points. The volumetric shape is an aggregate of blobbies, computed using a genetic algorithm. The choice of this model is motivated by the simplicity and the popularity of this primitive. The algorithm is tested on clouds of points representing a bunny and a cat.
Since Bedau et al. identified the simulation of open-ended evolution in digital life media as one... more Since Bedau et al. identified the simulation of open-ended evolution in digital life media as one of the key problems in the field of Artificial Life (Bedeau et al., 2000, Artificial Life 6, p.363), no attempt has convincingly solved the problem until this day. Creating open-ended evolution ultimately boils down to creating niches: A new evolutionary feature can only be retained if there is an ecological niche in which it becomes an innovation. An environment with a limited potential for hosting niches is inherently restricted as far as evolutionary innovations and open-ended evolution are concerned. Moreover, static niches, even in a very large number, are not enough to enable open-ended evolution, they need to appear persistently.
Functional networks-be they biological, social, or technological-have characteristics that distin... more Functional networks-be they biological, social, or technological-have characteristics that distinguish them from random or non-functional networks (Barabasi and Albert, 1999; Albert and Barabasi, 2002; Alon, 2007). Most well-known among these characteristics is the (approximately) scale-free degree distribution (probability that a node has k edges) of functional networks, as opposed to the binomial distribution of Erdös-Renyi random graphs. Another feature is a short mean path length through the network (also known as the 'small-world' property) that implies that signals can travel to any part of the network quickly (Girvan and Newman, 2002). While these properties are shared by protein-protein networks (Albert, 2005), metatbolic (Jeong et al., 2000; Wagner and Fell, 2001) and signaling (Barrios-Rodiles et al., 2005) networks, the topological and graph-properties of biological neural networks, in particular animal brains, are much less studied. Reigl et al. (2004) studied the connectivity patterns and computational modules in the nematode C. elegans brain, and found that it was structured into small computational modules that are over-represented with respect to an equivalent random network, yet with a degree distribution that is neither scale-free nor Poissonian.
PLoS Computational Biology, 2011
One of the hallmarks of biological organisms is their ability to integrate disparate information ... more One of the hallmarks of biological organisms is their ability to integrate disparate information sources to optimize their behavior in complex environments. How this capability can be quantified and related to the functional complexity of an organism remains a challenging problem, in particular since organismal functional complexity is not well-defined. We present here several candidate measures that quantify information and integration, and study their dependence on fitness as an artificial agent ("animat") evolves over thousands of generations to solve a navigation task in a simple, simulated environment. We compare the ability of these measures to predict high fitness with more conventional information-theoretic processing measures. As the animat adapts by increasing its "fit" to the world, information integration and processing increase commensurately along the evolutionary line of descent. We suggest that the correlation of fitness with information integration and with processing measures implies that high fitness requires both information processing as well as integration, but that information integration may be a better measure when the task requires memory. A correlation of measures of information integration (but also information processing) and fitness strongly suggests that these measures reflect the functional complexity of the animat, and that such measures can be used to quantify functional complexity even in the absence of fitness data.
Uploads
Papers by Nicolas Chaumont