Amazon.com

About: Amazon.com is a company organization based out in Seattle, Washington, United States. It is known for research contribution in the topics: Service (business) & Service provider. The organization has 13363 authors who have published 17317 publications receiving 266589 citations.

Leaf development and demography explain photosynthetic seasonality in Amazon evergreen forests.

Abstract: In evergreen tropical forests, the extent, magnitude, and controls on photosynthetic seasonality are poorly resolved and inadequately represented in Earth system models. Combining camera observations with ecosystem carbon dioxide fluxes at forests across rainfall gradients in Amazonia, we show that aggregate canopy phenology, not seasonality of climate drivers, is the primary cause of photosynthetic seasonality in these forests. Specifically, synchronization of new leaf growth with dry season litterfall shifts canopy composition toward younger, more light-use efficient leaves, explaining large seasonal increases (~27%) in ecosystem photosynthesis. Coordinated leaf development and demography thus reconcile seemingly disparate observations at different scales and indicate that accounting for leaf-level phenology is critical for accurately simulating ecosystem-scale responses to climate change.

Configuring intercommunications between computing nodes

Abstract: Techniques are described for configuring intercommunications between multiple computing nodes, such as multiple virtual machine nodes hosted on one or more physical computing machines or systems. In some situations, virtual networks may be established and maintained for groups of computing nodes, such as those operated by or on behalf of various users. Such virtual networks may be established in some situations by automatically configuring various communication manager modules to associate communications from a computing node belonging to a virtual network with one or more networking identifiers associated with the virtual network, such that other communication manager modules may appropriately forward or otherwise process such communications.

Using virtual networking devices and routing information to initiate external actions

Abstract: Techniques are described for providing managed virtual computer networks that have a configured logical network topology with virtual networking devices, such as by a network-accessible configurable network service, with corresponding networking functionality provided for communications between multiple computing nodes of the virtual computer network by emulating functionality that would be provided by the virtual networking devices if they were physically present. In some situations, the networking functionality provided for a managed computer network of a client includes receiving routing communications directed to the virtual networking devices, and using included routing information to identify and initiate external actions whose effects are not related to how network communications between computing nodes of the managed computer network are configured to be routed or otherwise forwarded through the managed computer network, such as external actions that affect devices that are not part of the managed computer network, or other types of external actions.

Basin-wide variations in foliar properties of Amazonian forest: phylogeny, soils and climate.

Abstract: . We analysed 1040 individual trees, located in 62 plots across the Amazon Basin for leaf mass per unit area (MA), foliar carbon isotopic composition (δ13C) and leaf level concentrations of C, N, P, Ca, Mg, K and Al. All trees were identified to the species level with the dataset containing 58 families, 236 genera and 508 species, distributed across a wide range of soil types and precipitation regimes. Some foliar characteristics such as MA, [C], [N] and [Mg] emerge as highly constrained by the taxonomic affiliation of tree species, but with others such as [P], [K], [Ca] and δ13C also strongly influenced by site growing conditions. By removing the environmental contribution to trait variation, we find that intrinsic values of most trait pairs coordinate, although different species (characterised by different trait suites) are found at discrete locations along a common axis of coordination. Species that tend to occupy higher fertility soils are characterised by a lower MA and have a higher intrinsic [N], [P], [K], [Mg] and δ13C than their lower fertility counterparts. Despite this consistency, different scaling patterns were observed between low and high fertility sites. Inter-relationships are thus substantially modified by growth environment. Analysing the environmental component of trait variation, we found soil fertility to be the most important predictor, influencing all leaf nutrient concentrations and δ13C and reducing MA. Mean annual temperature was negatively associated with leaf level [N], [P] and [K] concentrations. Total annual precipitation positively influences MA, [C] and δ13C, but with a negative impact on [Mg]. These results provide a first basis for understanding the relationship between the physiological functioning and distribution of tree species across Amazonia.

Born Again Neural Networks

Abstract: Knowledge Distillation (KD) consists of transferring “knowledge” from one machine learning model (the teacher) to another (the student). Commonly, the teacher is a high-capacity model with formidable performance, while the student is more compact. By transferring knowledge, one hopes to benefit from the student’s compactness, without sacrificing too much performance. We study KD from a new perspective: rather than compressing models, we train students parameterized identically to their teachers. Surprisingly, these Born-Again Networks (BANs), outperform their teachers significantly, both on computer vision and language modeling tasks. Our experiments with BANs based on DenseNets demonstrate state-of-the-art performance on the CIFAR-10 (3.5%) and CIFAR-100 (15.5%) datasets, by validation error. Additional experiments explore two distillation objectives: (i) Confidence-Weighted by Teacher Max (CWTM) and (ii) Dark Knowledge with Permuted Predictions (DKPP). Both methods elucidate the essential components of KD, demonstrating the effect of the teacher outputs on both predicted and non-predicted classes.