No abstract

The Northern Vosges and the Pays de Bitche (north-east France) are regions rich in recent industrial inheritance which history is well-known. On the other hand, the ancient history of these regions is not well known and the relationships between human populations and their environment remain unexplored until now for ancient times. The multidisciplinary palaeoenvironmental study that we carried out on the site of the bog-pond located below the ruins of the medieval castle of Waldeck, has made it possible to reconstruct the history of vegetation since 6 600 cal. BP.. Throughout the Holocene, the succession of forest vegetation (pine and hazelnut forests, reduced oak forest, beech forest, oak-beech forest) was largely dominated by pine. The human presence, tenuous during the Neolithic period, is well marked from the Bronze Age onwards with the introduction of crops and livestock crops in the catchment area. From the Middle Ages, anthropic pressure highly increased with the building, in the 13th century, of Waldeck Castle, which led to a major opening of the area. The Modern period is characterized by a gradual return of the forest, while anthropogenic pressure is decreasing. Over time, the occupation phases have been interspersed with abandonment phases during which human activities regress or disappear. Finally, the rarefaction analysis carried out on pollen data shows that human presence has led to a gradual increase in plant diversity, which peaked in the Middle Ages. As a result, the forest has lost some of its resilience to human disturbance over time.

No abstract

Automatic classification of aquatic microorganisms is based on the morphological features extracted from individual images. The current works on their classification do not consider the inter-class similarity and intra-class variance that causes misclassification. We are particularly interested in the case where variance within a class occurs due to discrete visual changes in microscopic images. In this paper, we propose to account for it by partitioning the classes with high variance based on the visual features. Our algorithm automatically decides the optimal number of sub-classes to be created and consider each of them as a separate class for training. This way, the network learns finer-grained visual features. Our experiments on two databases of freshwater benthic diatoms and marine plankton show that our method can outperform the state-of-theart approaches for classification of these aquatic microorganisms.