Decision fusion for the classification of hyperspectral data: Outcome of the 2008 GRS-S data fusion contest

The 2008 Data Fusion Contest organized by the IEEE Geoscience and Remote Sensing Data Fusion Technical Committee deals with the classification of high-resolution hyperspectral data from an urban area. Unlike in the previous issues of the contest, the goal was not only to identify the best algorithm but also to provide a collaborative effort: The decision fusion of the best individual algorithms was aiming at further improving the classification performances, and the best algorithms were ranked according to their relative contribution to the decision fusion. This paper presents the five awarded algorithms and the conclusions of the contest, stressing the importance of decision fusion, dimension reduction, and supervised classification methods, such as neural networks and support vector machines.

Multi-modal Change Detection, Application to the Detection of Flooded Areas: Outcome of the 2009-2010 data fusion contest

The 2009-2010 Data Fusion Contest organized by the Data Fusion Technical Committee of the IEEE Geoscience and Remote Sensing Society was focused on the detection of flooded areas using multi-temporal and multi-modal images. Both high spatial resolution optical and synthetic aperture radar data were provided. The goal was not only to identify the best algorithms (in terms of accuracy), but also to investigate the further improvement derived from decision fusion. This paper presents the four awarded algorithms and the conclusions of the contest, investigating both supervised and unsupervised methods and the use of multi-modal data for flood detection. Interestingly, a simple unsupervised change detection method provided similar accuracy as supervised approaches, and a digital elevation model-based predictive method yielded a comparable projected change detection map without using post-event data.

Quantitative comparison of reconstruction methods for intra-voxel fiber recovery from diffusion MRI.

Validation is arguably the bottleneck in the diffusion magnetic resonance imaging (MRI) community. This paper evaluates and compares 20 algorithms for recovering the local intra-voxel fiber structure from diffusion MRI data and is based on the results of the "HARDI reconstruction challenge" organized in the context of the "ISBI 2012" conference. Evaluated methods encompass a mixture of classical techniques well known in the literature such as diffusion tensor, Q-Ball and diffusion spectrum imaging, algorithms inspired by the recent theory of compressed sensing and also brand new approaches proposed for the first time at this contest. To quantitatively compare the methods under controlled conditions, two datasets with known ground-truth were synthetically generated and two main criteria were used to evaluate the quality of the reconstructions in every voxel: correct assessment of the number of fiber populations and angular accuracy in their orientation. This comparative study investigates the behavior of every algorithm with varying experimental conditions and highlights strengths and weaknesses of each approach. This information can be useful not only for enhancing current algorithms and develop the next generation of reconstruction methods, but also to assist physicians in the choice of the most adequate technique for their studies.

The demographic benefits of belligerence and bravery: defeated group repopulation or victorious group size expansion?

Intraspecific coalitional aggression between groups of individuals is a widespread trait in the animal world. It occurs in invertebrates and vertebrates, and is prevalent in humans. What are the conditions under which coalitional aggression evolves in natural populations? In this article, I develop a mathematical model delineating conditions where natural selection can favor the coevolution of belligerence and bravery between small-scale societies. Belligerence increases an actor's group probability of trying to conquer another group and bravery increase the actors's group probability of defeating an attacked group. The model takes into account two different types of demographic scenarios that may lead to the coevolution of belligerence and bravery. Under the first, the fitness benefits driving the coevolution of belligerence and bravery come through the repopulation of defeated groups by fission of victorious ones. Under the second demographic scenario, the fitness benefits come through a temporary increase in the local carrying capacity of victorious groups, after transfer of resources from defeated groups to victorious ones. The analysis of the model suggests that the selective pressures on belligerence and bravery are stronger when defeated groups can be repopulated by victorious ones. The analysis also suggests that, depending on the shape of the contest success function, costly bravery can evolve in groups of any size.

The combination of social and personal contexts affects dominance hierarchy development in shore crabs, Carcinus maenas

Many animals that live in groups maintain competitive relationships, yet avoid continual fighting, by forming dominance hierarchies. We compare predictions of stochastic, individual-based models with empirical experimental evidence using shore crabs to test competing hypotheses regarding hierarchy development. The models test (1) what information individuals use when deciding to fight or retreat, (2) how past experience affects current resource-holding potential, and (3) how individuals deal with changes to the social environment. First, we conclude that crabs assess only their own state and not their opponent's when deciding to fight or retreat. Second, willingness to enter, and performance in, aggressive contests are influenced by previous contest outcomes. Winning increases the likelihood of both fighting and winning future interactions, while losing has the opposite effect. Third, when groups with established dominance hierarchies dissolve and new groups form, individuals reassess their ranks, showing no memory of previous rank or group affiliation. With every change in group composition, individuals fight for their new ranks. This iterative process carries over as groups dissolve and form, which has important implications for the relationship between ability and hierarchy rank. We conclude that dominance hierarchies emerge through an interaction of individual and social factors, and discuss these findings in terms of an underlying mechanism. Overall, our results are consistent with crabs using a cumulative assessment strategy iterated across changes in group composition, in which aggression is constrained by an absolute threshold in energy spent and damage received while fighting.

Effects of hatching asynchrony on sibling negotiation, begging, jostling for position and within-brood food allocation in the barn owl Tyto alba

When siblings differ markedly in their need for food, they may benefit from signalling to each other their willingness to contest the next indivisible food item delivered by the parents. This sib-sib communication system, referred to as 'sibling negotiation', may allow them to adjust optimally to investment in begging. Using barn owl (Two alba) broods. I assessed the role of within-brood age hierarchy on sibling negotiation, and in turn on jostling for position where parents predictably deliver food (i.e. nest-box entrance), begging and within-brood food allocation. More specifically, I examined three predictions derived from a game-theoretical model of sibling negotiation where a senior and a junior sibling compete for food resources (Roulin, 2002a, Johnstone and Roulin, 2003): (1) begging effort invested by the senior sibling should be less sensitive to the junior sibling's negotiation than vice versa; (2) the junior should invest less effort in sibling negotiation than its senior sibling but a similar amount of effort in begging; and (3) within-brood food allocation should be directly related to begging but only indirectly to sibling negotiation. Two-chick broods were created and vocalization in the absence (negotiation signals directed to siblings) and presence (begging signals directed to parents) of parents was recorded. In support of the first prediction, juniors begged at a low cadence after their senior sibling negotiated intensely, probably because negotiation reflects prospective investment in begging and hence willingness to compete. In contrast, the begging of senior siblings was not sensitive to their junior sibling's negotiation. In contrast to the second prediction, juniors negotiated and begged more intensely than their senior sibling apparently because they were hungrier rather than younger. In line with the third prediction, juniors monopolized food delivered by their parents when their senior sibling begged at a low level. The begging cadence of both the junior and senior sibling, the junior's negotiation cadence, the difference in age between the two nest-mates and jostling for position were not associated with the likelihood of monopolizing food. In conclusion, sibling negotiation appears to influence begging behaviour, which, in turn, affects within-brood food allocation. Juniors may negotiate to challenge their senior siblings, and thereby determine whether seniors are less hungry before deciding to beg for food. In contrast, seniors may negotiate to deter juniors from begging.

Nestling barn owls assess short-term variation in the amount of vocally competing siblings.

Assessing the amount of rivals is crucial to optimally adjust investment into a contest. If laboratory animals show numerical abilities, little is known about the ecological implications particularly in young animals. The two to nine barn owl (Tyto alba) siblings vocally compete for priority of access to food resources before parents actually deliver them. In dyads, the individual that vocalizes at the highest rate in the absence of parents deters its siblings from competing for next delivered prey. We tested the novel hypothesis that to optimally adjust vocal investment, barn owl nestlings assess how many of their siblings are currently competing. To singleton owlets, we broadcasted a fixed global number of calls emitted by one, two or four pre-recorded unfamiliar nestlings. We could thus distinguish the independent effect on singletons' vocal behavior of the global number of calls produced by a brood from the number of competitors that produced these calls. Overall, nestlings retreated more from vocal contest when facing more competitors. However, in front of one highly motivated competitor, nestlings refrained from vocalizing to a larger extent than when competing against more but less motivated individuals. Therefore, young animals assess variation in the number of currently competing siblings based on individual-specific vocal cues.

Plate tectonics of the Altaids

Abstract: The Altaids consist in a huge accretionary-type belt extending from Siberia through Mon-golia, northern China, Kyrgyzstan and Kazakhstan. They were formed from the Vendian through the Jurassic by the accretion of numerous displaced and exotic terranes (e.g. island arc, ribbon microcontinent, seamount, basaltic plateau, back-arc basin). The number, nature and origin of the terranes differ according to the palaeotectonic models of the different authors. Thanks to a geo- dynamic study (i.e. definition of tectonic settings and elaboration of geodynamic scenarios) and plate tectonics modelling, this work aims to present an alternative model explaining the Palaeozoic palaeotectonic evolution of the Altaids. Based on a large set of compiled geological data related to palaeogeography and geodyna¬mic (e.g. sedimentology, stratigraphy, palaeobiogeography, palaeomagnetism, magmatism, me- tamorphism, tectonic...), a partly new classification of the terranes and sutures implicated in the formation of the Altaids is proposed. In the aim to elaborate plate tectonics reconstructions, it is necessary to fragment the present arrangement of continents into consistent geological units. To avoid confusion with existing terminology (e.g. tectonic units, tectono-stratigraphic units, micro- continents, terranes, blocks...), the new concept of "Geodynamic Units (GDU)" was introduced. A terrane may be formed by a set of GDUs. It consists of a continental and/or oceanic fragment which has its own kinematic and geodynamic evolution for a given period. With the same ap-proach, the life span and type of the disappeared oceans is inferred thanks to the study of the mate-rial contained in suture zones. The interpretation of the tectonic settings within the GDUs comple-ted by the restoration of oceans leads to the elaboration of geodynamic scenarios. Since the Wilson cycle was presented in 1967, numerous works demonstrated that the continental growth is more complex and results from diverse geodynamic scenarios. The identification of these scenarios and their exploitation enable to elaborate plate tectonics models. The models are self-constraining (i.e. space and time constraints) and contest or confirm in turn the geodynamic scenarios which were initially proposed. The Altaids can be divided into three domains: (1) the Peri-Siberian, (2) the Kazakhstan, and (3) the Tarim-North China domains. The Peri-Siberian Domain consists of displaced (i.e. Sayan Terrane Tuva-Mongolian, Lake-Khamsara Terrane) and exotic terranes (i.e. Altai-Mongolian and Khangai-Argunsky Terrane) accreted to Siberia from the Vendian through the Ordovician. Fol-lowing the accretion of these terranes, the newly formed Siberia active margin remained active un-til its part collision with the Kazakhstan Superterrane in the Carboniferous. The eastern part of the active margin (i.e. East Mongolia) continued to act until the Permian when the North-China Tarim Superterrane collided with it. The geodynamic evolution of the eastern part of the Peri-Siberian Domain (i.e. Eastern Mongolia and Siberia) is complicated by the opening of the Mongol-Okhotsk Ocean in the Silurian. The Kazakhstan Domain is composed of several continental terranes of East Gondwana origin amalgamated together during the Ordovician-Silurian time. After these different orogenic events, the Kazakhstan Superterrane evolved as a single superterrane until its collision with a Tarim-North China related-terrane (i.e. Tianshan-Hanshan Terrane) and Siberian Continent during the Devonian. This new organisation of the continents imply a continued active margin from Siberia, to North China through the Kazakhstan Superterrane and the closure of the Junggar- Balkash Ocean which implied the oroclinal bending of the Kazakhstan Superterrane during the entire Carboniferous. The formation history of the Tarim-North China Domain is less complex. The Cambrian northern passive margin became active in the Ordovician. In the Silurian, the South Tianshan back-arc Ocean was open and led to the formation of the Tianshan-Hanshan Terrane which collided with the Kazakhstan Superterrane during the Devonian. The collision between Siberia and the eastern part of the Tarim-North China continents (i.e. Inner Mongolia), implied by the closure of the Solonker Ocean, took place in the Permian. Since this time, the major part of the Altaids was formed, the Mongol-Okhotsk Ocean only was still open and closed during the Jurassic. Résumé: La chaîne des Altaïdes est une importante chaîne d'accrétion qui s'étend en Sibérie, Mon-golie, Chine du Nord, Kirghizstan et Kazakhstan. Elle s'est formée durant la période du Vendian au Jurassique par l'accrétion de nombreux terranes déplacés ou exotiques (par exemple arc océa-nique, microcontinent, guyot, plateau basaltique, basin d'arrière-arc...). Le nombre, la nature ou encore l'origine diffèrent selon les modèles paléo-tectoniques proposés par les différents auteurs. Grâce à une étude géodynamique (c'est-à-dire définition des environnements tectoniques et éla-boration de scénarios géodynamiques) et à la modélisation de la tectonique des plaques, ce travail propose un modèle alternatif expliquant l'évolution paléo-tectonique des Altaïdes. Basé sur une large compilation de données géologiques pertinentes en termes de paléo-géographie et de géodynamique (par exemple sédimentologie, stratigraphie, paléo-biogéographie, paléomagnétisme, magmatisme, métamorphisme, tectonique...), une nouvelle classification des terranes et des sutures impliqués dans la formation des Altaïdes est proposée. Dans le but d'élabo¬rer des reconstructions de plaques tectoniques, il est nécessaire de fragmenter l'arrangement actuel des continents en unités tectoniques cohérentes. Afin d'éviter les confusions avec la terminolo¬gie existante (par exemple unité tectonique, unité tectono-stratigraphique, microcontinent, block, terrane...), le nouveau concept d' "Unité Géodynamique (UGD)" a été introduit. Un terrane est formé d'une ou plusieurs UGD et représente un fragment océanique ou continental défini pas sa propre cinétique et évolution géodynamique pour une période donnée. Parallèlement, la durée de vie et le type des océans disparus (c'est-à-dire principal ou secondaire) est déduite grâce à l'étude du matériel contenu dans les zones de sutures. L'interprétation des environnements tectoniques des UGD associés à la restauration des océans mène à l'élaboration de scénarios géodynamiques. Depuis que le Cycle de Wilson a été présenté en 1967, de nombreux travaux ont démontré que la croissance continentale peut résulter de divers scénarios géodynamiques. L'identification et l'ex-ploitation de ces scénarios permet finalement l'élaboration de modèles de tectonique des plaques. Les modèles sont auto-contraignants (c'est-à-dire contraintes spatiales et temporelles) et peuvent soit contester ou confirmer les scénarios géodynamiques initialement proposés. Les Altaïdes peuvent être divisées en trois domaines : (1) le Domaine Péri-Sibérien, (2) le Domaine Kazakh, et (3) le Domaine Tarim-Nord Chinois. Le Domaine Péri-Sibérien est composé de terranes déplacés (c'est-à-dire Terrane du Sayan, Tuva-Mongol et Lake-Khamsara) et exotiques (c'est-à-dire Terrane Altai-Mongol et Khangai-Argunsky) qui ont été accrétés au craton Sibérien durant la période du Vendien à l'Ordovicien. Suite à l'accrétion de ces terranes, la marge sud-est de la Sibérie nouvellement formée reste active jusqu'à sa collision partielle avec le Superterrane Ka-zakh au Carbonifère. La partie est de la marge active (c'est-à-dire Mongolie de l'est) continue son activité jusqu'au Permien lors de sa collision avec le Superterrane Tarim-Nord Chinois. L'évolu¬tion géodynamique de la partie est du Domaine Sibérien est compliquée par l'ouverture Silurienne de l'Océan Mongol-Okhotsk qui disparaîtra seulement au Jurassique. Le Domaine Kazakh est composé de plusieurs terranes d'origine est-Gondwanienne accrétés les uns avec les autres avant ou pendant le Silurien inférieur et leurs evolution successive sous la forme d'un seul superterrane. Le Superterrane Kazakh collisione avec un terrane Tarim-Nord Chinois (c'est-à-dire Terrane du Tianshan-Hanshan) durant le Dévonien et le continent Sibérien au Dévonien supérieur. Ce nouvel agencement des plaques induit une marge active continue le long des continents Sibérien, Kazakh et Nord Chinois et la fermeture de l'Océan Junggar-Balkash qui provoque le plissement oroclinal du Superterrane Kazakh durant le Carbonifère. L'histoire de la formation du Domaine Tarim-Nord Chinois est moins complexe. La marge passive nord Cambrienne devient active à l'Ordovicien et l'ouverture Silurienne du bassin d'arrière-arc du Tianshan sud mène à la formation du terrane du Tianshan-Hanshan. La collision Dévonienne entre ce dernier et le Superterrane Kazakh provoque la fermerture de l'Océan Tianshan sud. Finalement, la collision entre la Sibérie et la partie est du continent Tarim-Nord Chinois (c'est-à-dire Mongolie Intérieure) prend place durant le Permien suite à la fermeture de l'Océan Solonker. La majeure partie des Altaïdes est alors formée, seul l'Océan Mongol-Okhotsk est encore ouvert. Ce dernier se fermera seulement au Jurassique.