Définition d'un état zero et mise en place d'un système de suivi écologique permanent de l'arboretum de la station forestière de Mandraka

Mandraka possède de nombreux écosystèmes, dominés surtout par les forêts. Cette zone est située sur la première falaise orientale malgache et affiche des reliefs accidentés (pentes supérieures à 60%). Elle est exposée à un régime climatique à forte influence orientale se traduisant par une humidité permanente et un régime cyclonique fréquent. Les paramètres stationnels sont ainsi rudes, or ils sont écologiquement très importants car plusieurs caractéristiques physionomiques et comportementales des espèces forestières en dépendent. Cette étude s'intéresse à la station forestière de Mandraka, particulièrement à l'arboretum. Ce dernier fût créé dans les années cinquante et est actuellement géré par le Département des Eaux et Forêts. Ce site est actuellement à vocation pédagogique et écotouristique. Son état écologique est inconnu jusqu'à maintenant, et depuis sa création, aucun système n'a été mis en place pour mesurer et suivre sa viabilité. D'où, l'intitulé de ce travail de mémoire : « Définition d'un état zéro et mise en place d'un système de suivi écologique permanent de l'arboretum de la station forestière de Mandraka ». Les objectifs étant de collecter des données concernant l'état écologique actuel du site, d'identifier des indicateurs de suivi pour mesurer sa viabilité, et d'inclure un système de suivi écologique permanent dans une proposition de plan d'aménagement pour l'arboretum. Le suivi est en effet un outil très important pour l'analyse des ressources forestières. Il permet de cadrer toutes les interventions. Les diverses analyses menées lors de cette étude ont révélé une viabilité moyenne de l'arboretum. Cela est induit par une qualité de peuplement moyennement stable, une mortalité élevée (plus de 14%), et un potentiel d'avenir très faible, voire inexistant (taux de régénération à 0%). L'envahissement de la forêt artificielle par les espèces autochtones constitue la pression la plus importante de cet arboretum vu qu'il se trouve au milieu des forêts naturelles. L'analyse sylvicole effectuée sur les deux types dendrologiques révèle que les peuplements de conifères présentent des caractéristiques sylvicoles plus favorables que les feuillus. Ce niveau moyen de viabilité de l'arboretum implique ainsi la proposition d'un plan d'aménagement pour l'améliorer; le suivi est une activité primordiale, d'où la proposition d'un plan de suivi permanent pour l'arboretum. A noter que malgré la considération du critère de représentativité pour l'échantillonnage, toutes les questions ne pourront être répondues du fait que plusieurs mosaïques de peuplements artificiels de différentes espèces constituent l'arboretum, et que chacune de ces espèces plantées ont leurs propres caractéristiques. La mise en place des plots permanents d'observation ne servira ainsi que de référence (Etat zéro), mais on propose de prévoir un suivi intégral ainsi que diverses interventions pour l'arboretum en général. Ce travail constitue ainsi une base de données pour l'arboretum et pour la station forestière de Mandraka, mais il ne représente qu'une des facettes à prendre en considération dans une finalité d'amélioration de la viabilité. L'élaboration de cartes thématiques et d'évolution spatio-temporelle à l'issue de SIG (Système d'Information Géographique) permettra d'enrichir les informations établies et admettra un suivi plus précis.

Analyzing spatio-temporal patterns of genuine cross-correlations

In multivariate time series analysis, the equal-time cross-correlation is a classic and computationally efficient measure for quantifying linear interrelations between data channels. When the cross-correlation coefficient is estimated using a finite amount of data points, its non-random part may be strongly contaminated by a sizable random contribution, such that no reliable conclusion can be drawn about genuine mutual interdependencies. The random correlations are determined by the signals' frequency content and the amount of data points used. Here, we introduce adjusted correlation matrices that can be employed to disentangle random from non-random contributions to each matrix element independently of the signal frequencies. Extending our previous work these matrices allow analyzing spatial patterns of genuine cross-correlation in multivariate data regardless of confounding influences. The performance is illustrated by example of model systems with known interdependence patterns. Finally, we apply the methods to electroencephalographic (EEG) data with epileptic seizure activity.

Spatio-temporal credit assignment in population learning

Learning by reinforcement is important in shaping animal behavior, and in particular in behavioral decision making. Such decision making is likely to involve the integration of many synaptic events in space and time. However, using a single reinforcement signal to modulate synaptic plasticity, as suggested in classical reinforcement learning algorithms, a twofold problem arises. Different synapses will have contributed differently to the behavioral decision, and even for one and the same synapse, releases at different times may have had different effects. Here we present a plasticity rule which solves this spatio-temporal credit assignment problem in a population of spiking neurons. The learning rule is spike-time dependent and maximizes the expected reward by following its stochastic gradient. Synaptic plasticity is modulated not only by the reward, but also by a population feedback signal. While this additional signal solves the spatial component of the problem, the temporal one is solved by means of synaptic eligibility traces. In contrast to temporal difference (TD) based approaches to reinforcement learning, our rule is explicit with regard to the assumed biophysical mechanisms. Neurotransmitter concentrations determine plasticity and learning occurs fully online. Further, it works even if the task to be learned is non-Markovian, i.e. when reinforcement is not determined by the current state of the system but may also depend on past events. The performance of the model is assessed by studying three non-Markovian tasks. In the first task, the reward is delayed beyond the last action with non-related stimuli and actions appearing in between. The second task involves an action sequence which is itself extended in time and reward is only delivered at the last action, as it is the case in any type of board-game. The third task is the inspection game that has been studied in neuroeconomics, where an inspector tries to prevent a worker from shirking. Applying our algorithm to this game yields a learning behavior which is consistent with behavioral data from humans and monkeys, revealing themselves properties of a mixed Nash equilibrium. The examples show that our neuronal implementation of reward based learning copes with delayed and stochastic reward delivery, and also with the learning of mixed strategies in two-opponent games.

Spatio-temporal credit assignment in population learning

Learning by reinforcement is important in shaping animal behavior. But behavioral decision making is likely to involve the integration of many synaptic events in space and time. So in using a single reinforcement signal to modulate synaptic plasticity a twofold problem arises. Different synapses will have contributed differently to the behavioral decision and, even for one and the same synapse, releases at different times may have had different effects. Here we present a plasticity rule which solves this spatio-temporal credit assignment problem in a population of spiking neurons. The learning rule is spike time dependent and maximizes the expected reward by following its stochastic gradient. Synaptic plasticity is modulated not only by the reward but by a population feedback signal as well. While this additional signal solves the spatial component of the problem, the temporal one is solved by means of synaptic eligibility traces. In contrast to temporal difference based approaches to reinforcement learning, our rule is explicit with regard to the assumed biophysical mechanisms. Neurotransmitter concentrations determine plasticity and learning occurs fully online. Further, it works even if the task to be learned is non-Markovian, i.e. when reinforcement is not determined by the current state of the system but may also depend on past events. The performance of the model is assessed by studying three non-Markovian tasks. In the first task the reward is delayed beyond the last action with non-related stimuli and actions appearing in between. The second one involves an action sequence which is itself extended in time and reward is only delivered at the last action, as is the case in any type of board-game. The third is the inspection game that has been studied in neuroeconomics. It only has a mixed Nash equilibrium and exemplifies that the model also copes with stochastic reward delivery and the learning of mixed strategies.

Spatio-temporal credit assignment in population learning

We present a model for plasticity induction in reinforcement learning which is based on a cascade of synaptic memory traces. In the cascade of these so called eligibility traces presynaptic input is first corre lated with postsynaptic events, next with the behavioral decisions and finally with the external reinforcement. A population of leaky integrate and fire neurons endowed with this plasticity scheme is studied by simulation on different tasks. For operant co nditioning with delayed reinforcement, learning succeeds even when the delay is so large that the delivered reward reflects the appropriateness, not of the immediately preceeding response, but of a decision made earlier on in the stimulus - decision sequence . So the proposed model does not rely on the temporal contiguity between decision and pertinent reward and thus provides a viable means of addressing the temporal credit assignment problem. In the same task, learning speeds up with increasing population si ze, showing that the plasticity cascade simultaneously addresses the spatial problem of assigning credit to the different population neurons. Simulations on other task such as sequential decision making serve to highlight the robustness of the proposed sch eme and, further, contrast its performance to that of temporal difference based approaches to reinforcement learning.

Reconstruction and 3D visualisation based on objective real 3D based documentation

Reconstructions based directly upon forensic evidence alone are called primary information. Historically this consists of documentation of findings by verbal protocols, photographs and other visual means. Currently modern imaging techniques such as 3D surface scanning and radiological methods (Computer Tomography, Magnetic Resonance Imaging) are also applied. Secondary interpretation is based on facts and the examiner's experience. Usually such reconstructive expertises are given in written form, and are often enhanced by sketches. However, narrative interpretations can, especially in complex courses of action, be difficult to present and can be misunderstood. In this report we demonstrate the use of graphic reconstruction of secondary interpretation with supporting pictorial evidence, applying digital visualisation (using 'Poser') or scientific animation (using '3D Studio Max', 'Maya') and present methods of clearly distinguishing between factual documentation and examiners' interpretation based on three cases. The first case involved a pedestrian who was initially struck by a car on a motorway and was then run over by a second car. The second case involved a suicidal gunshot to the head with a rifle, in which the trigger was pushed with a rod. The third case dealt with a collision between two motorcycles. Pictorial reconstruction of the secondary interpretation of these cases has several advantages. The images enable an immediate overview, give rise to enhanced clarity, and compel the examiner to look at all details if he or she is to create a complete image.

Spatio-Temporal Credit Assignment in Neuronal Population Learning

n learning from trial and error, animals need to relate behavioral decisions to environmental reinforcement even though it may be difficult to assign credit to a particular decision when outcomes are uncertain or subject to delays. When considering the biophysical basis of learning, the credit-assignment problem is compounded because the behavioral decisions themselves result from the spatio-temporal aggregation of many synaptic releases. We present a model of plasticity induction for reinforcement learning in a population of leaky integrate and fire neurons which is based on a cascade of synaptic memory traces. Each synaptic cascade correlates presynaptic input first with postsynaptic events, next with the behavioral decisions and finally with external reinforcement. For operant conditioning, learning succeeds even when reinforcement is delivered with a delay so large that temporal contiguity between decision and pertinent reward is lost due to intervening decisions which are themselves subject to delayed reinforcement. This shows that the model provides a viable mechanism for temporal credit assignment. Further, learning speeds up with increasing population size, so the plasticity cascade simultaneously addresses the spatial problem of assigning credit to synapses in different population neurons. Simulations on other tasks, such as sequential decision making, serve to contrast the performance of the proposed scheme to that of temporal difference-based learning. We argue that, due to their comparative robustness, synaptic plasticity cascades are attractive basic models of reinforcement learning in the brain.

Evaluation of a portable image overlay projector for the visualisation of surgical navigation data: phantom studies

Presenting visual feedback for image-guided surgery on a monitor requires the surgeon to perform time-consuming comparisons and diversion of sight and attention away from the patient. Deficiencies in previously developed augmented reality systems for image-guided surgery have, however, prevented the general acceptance of any one technique as a viable alternative to monitor displays. This work presents an evaluation of the feasibility and versatility of a novel augmented reality approach for the visualisation of surgical planning and navigation data. The approach, which utilises a portable image overlay device, was evaluated during integration into existing surgical navigation systems and during application within simulated navigated surgery scenarios.