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.

User identification by gesture recognition

Abstract: A user can be identified and/or authenticated to an electronic device by analyzing aspects of a motion or gesture made by that user. At least one imaging element of the device can capture image information including the motion or gesture, and can determine time-dependent information about that motion or gesture in two or three dimensions of space. The time-dependent information can be used to identify varying speeds, motions, and other such aspects that are indicative of a particular user. The way in which a gesture or motion is made, in addition to the motion or gesture itself, can be used to authenticate an individual user. While other persons can learn the basic gesture or motion, the way in which each person makes that gesture or motion will generally be at least slightly different, which can be used to prevent unauthorized access to sensitive information, protected functionality, or other such content.

Greenhouse-gas emissions from Amazonian hydroelectric reservoirs: the example of Brazil's Tucuruí Dam as compared to fossil fuel alternatives

Abstract: Hydroelectric dams in tropical forest areas emit carbon dioxide and methane. How these emissions and their impacts should be calculated, and how comparisons should be made with global warming contributions of alternative energy sources such as fossil fuels, can lead to sharp differences in conclusions on the relative advantages of these options. The example of Brazil's Tucurui Dam is examined to clarify these differences. The present paper extends an earlier analysis to 100 years and explores the differences between these and comparable fossil fuel emissions.Factors considered here in calculating emissions for Tucurui Dam include the initial stock and distribution of carbon, decay rates and pathways (leading to carbon dioxide and methane), and losses of power in transmission lines. Factors not considered include forest degradation on islands and reservoir shores, nitrous oxide sources in drawdown zones and transmission lines, additional methane emission pathways for release from standing trees, water passing through the turbines, etc. Construction-phase emissions are also not included; neither are emissions from deforestation by people displaced by and attracted to the project. A complete accounting of the alternative landscape is also lacking. Standardization of the level of reliability of the electricity supply is needed to compare hydroelectric and thermoelectric options.Types of emission calculations commonly used include the ultimate contribution to emissions, the annual balance of emissions in a given year, and emissions over a long time horizon (such as 100 years). The timing of emissions differs between hydroelectric and thermal generation, hydro producing a large pulse of carbon dioxide emissions in the first years after filling the reservoir while thermal produces a constant flux of gases in proportion to the power generated. The impacts of emissions are related to the atmospheric load (stocks) of the gases rather than to the emissions (flows), and therefore last over a long time. According to the calculations in the present paper, the average carbon dioxide molecule in the atmospheric load contributed by Tucurui was present in the atmosphere 15 years earlier than the average molecule in the comparable load from fossil fuel generation. This means that, considering a 100-year time horizon, a tonne of CO2 emitted by Tucurui has 15% more global warming impact than a tonne emitted by fossil fuel, assuming no discounting. If discounting is applied, then the relative impact of the hydroelectric option is increased.Time preference, either by discounting or by an alternative procedure, is a key factor affecting the attractiveness of hydroelectric power. At low annual discount rates (say 1–2%), the attractiveness of Tucurui, although less than without discounting, is still 3–4 times better than fossil-fuel generation. If the discount rate reaches 15%, the situation is reversed, and fossil-fuel generation becomes more attractive from a global-warming perspective. Tucurui, with a power density (installed capacity/reservoir area) of 1.63 W m-2 is better than both the 0.81 W m-2 average for Brazilian Amazonia's 5500 km2 of existing reservoirs and the 1 W m-2 estimated by Brazil's electrical authorities as the mean for all planned hydroelectric development in the region.

Can 3D Pose Be Learned from 2D Projections Alone

Abstract: 3D pose estimation from a single image is a challenging task in computer vision. We present a weakly supervised approach to estimate 3D pose points, given only 2D pose landmarks. Our method does not require correspondences between 2D and 3D points to build explicit 3D priors. We utilize an adversarial framework to impose a prior on the 3D structure, learned solely from their random 2D projections. Given a set of 2D pose landmarks, the generator network hypothesizes their depths to obtain a 3D skeleton. We propose a novel Random Projection layer, which randomly projects the generated 3D skeleton and sends the resulting 2D pose to the discriminator. The discriminator improves by discriminating between the generated poses and pose samples from a real distribution of 2D poses. Training does not require correspondence between the 2D inputs to either the generator or the discriminator. We apply our approach to the task of 3D human pose estimation. Results on Human3.6M dataset demonstrates that our approach outperforms many previous supervised and weakly supervised approaches.

Adaptive algorithms for online convex optimization with long-term constraints

Abstract: We present an adaptive online gradient descent algorithm to solve online convex optimization problems with long-term constraints, which are constraints that need to be satisfied when accumulated over a finite number of rounds T, but can be violated in intermediate rounds. For some user-defined trade-off parameter β ∈ (0; 1), the proposed algorithm achieves cumulative regret bounds of O(Tmax{β,1-β}) and O(T1-β/2), respectively for the loss and the constraint violations. Our results hold for convex losses, can handle arbitrary convex constraints and rely on a single computationally efficient algorithm. Our contributions generalize over the best known cumulative regret bounds of Mahdavi et al. (2012a), which are respectively O(T1/2) and O(T3/4) for general convex domains, and respectively O(T2/3) and O(T2/3) when the domain is further restricted to be a polyhedral set. We supplement the analysis with experiments validating the performance of our algorithm in practice.

Personalized recommendations based on item usage

Abstract: This disclosure describes systems and associated processes for generating recommendations for users based on usage, among other things. These systems and processes are described in the context of an interactive computing system that enables users to download applications for mobile devices (such as phones) or for other computing devices. Users' interactions with applications once they are downloaded can be observed and tracked, with such usage data being collected and provided to the interactive computing system. The interactive computing system can include a recommendation system or service that processes the usage data from a plurality of users to detect usage patterns. Using these usage patterns, among possibly other data, the recommendation system can recommend applications to users for download.