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: Computer science & Service (business). The organization has 13363 authors who have published 17317 publications receiving 266589 citations.

Method and system for integrating transaction mechanisms over multiple internet sites

Abstract: A method for effecting transactions across multiple vendors in an integrated environment, wherein the user may purchase each of a plurality of items the user finds independent of the vendors. The user's selections are received from the user and mapped to selected set of vendors. When the user is finished, she invokes a check-out application to fill in one or many order entry forms for each of the relevant vendors whose goods the user selected during the course of shopping. The check-out application uses common information, such as name, address and credit card number, previously provided by the user in order to fill in the order entry forms for each vendor without requiring the user to fill in these forms. Finally, the check-out application tracks confirmation numbers in a common information store.

Environmental impacts of Brazil's Tucuruí Dam: unlearned lessons for hydroelectric development in Amazonia.

Abstract: Brazil's Tucurui Dam provides valuable lessons for improving decision-making on major public works in Amazonia and elsewhere. Together with social impacts, which were reviewed in a companion paper, the project's environmental costs are substantial. Monetary costs include costs of construction and maintenance and opportunity costs of natural resources (such as timber) and of the money invested by the Brazilian government. Environmental costs include forest loss, leading to both loss of natural ecosystems and to greenhouse gas emissions. Aquatic ecosystems are heavily affected by the blockage of fish migration and by creation of anoxic environments. Decay of vegetation left in the reservoir creates anoxic water that can corrode turbines, as well as producing methane and providing conditions for methylation of mercury. Defoliants were considered for removing forest in the submergence area but plans were aborted amid a public controversy. Another controversy surrounded impacts of defoliants used to prevent regrowth along the transmission line. Mitigation measures included archaeological and faunal salvage and creation of a "gene bank" on an island in the reservoir. Decision-making in the case of Tucurui was virtually uninfluenced by environmental studies, which were done concurrently with construction. The dam predates Brazil's 1986 requirement of an Environmental Impact Assessment. Despite limitations, research results provide valuable information for future dams. Extensive public-relations use of the research effort and of mitigation measures such as faunal salvage were evident. Decision-making was closely linked to the influence of construction firms, the military, and foreign financial interests in both the construction project and the use of the resulting electrical power (most of which is used for aluminum smelting). Social and environmental costs received virtually no consideration when decisions were made, an outcome facilitated by a curtain of secrecy surrounding many aspects of the project. Despite improvements in Brazil's system of environmental impact assessment since the Tucurui reservoir was filled in 1984, many essential features of the decision-making system remain unchanged.

The Geometry of Culture: Analyzing the Meanings of Class through Word Embeddings:

Abstract: We argue word embedding models are a useful tool for the study of culture using a historical analysis of shared understandings of social class as an empirical case. Word embeddings represent semant...

Adaptive neural networks for efficient inference

Abstract: We present an approach to adaptively utilize deep neural networks in order to reduce the evaluation time on new examples without loss of accuracy. Rather than attempting to redesign or approximate existing networks, we propose two schemes that adaptively utilize networks. We first pose an adaptive network evaluation scheme, where we learn a system to adaptively choose the components of a deep network to be evaluated for each example. By allowing examples correctly classified using early layers of the system to exit, we avoid the computational time associated with full evaluation of the network. We extend this to learn a network selection system that adaptively selects the network to be evaluated for each example. We show that computational time can be dramatically reduced by exploiting the fact that many examples can be correctly classified using relatively efficient networks and that complex, computationally costly networks are only necessary for a small fraction of examples. We pose a global objective for learning an adaptive early exit or network selection policy and solve it by reducing the policy learning problem to a layer-by-layer weighted binary classification problem. Empirically, these approaches yield dramatic reductions in computational cost, with up to a 2.8x speedup on state-of-the-art networks from the ImageNet image recognition challenge with minimal (< 1%) loss of top5 accuracy.