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.

Ground effect based surface sensing in automated aerial vehicles

Abstract: This disclosure describes a system and method for operating an automated aerial vehicle wherein influences of a ground effect may be utilized for sensing the ground or other surfaces. In various implementations, an operating parameter of the automated aerial vehicle may be monitored to determine when a ground effect is influencing the parameter, which correspondingly indicates a proximity to a surface (e.g., the ground). In various implementations, the ground effect based sensing techniques may be utilized for determining a proximity to the ground, as a backup for a primary sensor system, for determining if a landing location is uneven, etc.

Domain Adaptation with BERT-based Domain Classification and Data Selection.

Abstract: The performance of deep neural models can deteriorate substantially when there is a domain shift between training and test data. For example, the pre-trained BERT model can be easily fine-tuned with just one additional output layer to create a state-of-the-art model for a wide range of tasks. However, the fine-tuned BERT model suffers considerably at zero-shot when applied to a different domain. In this paper, we present a novel two-step domain adaptation framework based on curriculum learning and domain-discriminative data selection. The domain adaptation is conducted in a mostly unsupervised manner using a small target domain validation set for hyper-parameter tuning. We tested the framework on four large public datasets with different domain similarities and task types. Our framework outperforms a popular discrepancy-based domain adaptation method on most transfer tasks while consuming only a fraction of the training budget.

Multi-Factor Authentication to Access Services

Abstract: A technology is provided for using a multi-factor authentication process to access services in a computing service environment. One or more policies can be defined for allowing access to one or more services and/or resources associated with a service provider environment according to an authenticated identity. A device, detected by a voice-capturing endpoint within a defined geographical location, may be authenticated according to a unique identification (ID). Voice data received from the voice-capturing endpoint can be authenticated. The authenticated identity can be established according to the authenticated device and the authenticated voice data. A command, received via a voice command from the voice-capturing endpoint, may be issued with the authenticated identity to access the one or more services and/or resources associated with the service provider environment according to the plurality of policies.

Providing payments automatically in accordance with predefined instructions

Abstract: Techniques are described for facilitating interactions between computing systems, such as by using an authorization system to automatically authorize financial payments between parties in accordance with previously specified private authorization instructions of at least one of the parties. In some situations, some or all of the payments are associated with commerce-related or other transactions, such as transactions initiated by a consumer via the Web to acquire items from a retailer. The authorization instructions may include predefined instruction sets that regulate conditions under which a potential payment can be authorized, with the instruction sets each associated in some situations with a reference. After one or more parties each supply one or more such references or otherwise indicate one or more such instruction sets for use with a potential payment, the authorization system can determine whether to authorize the payment based on whether the instruction sets are compatible or otherwise satisfied.

Bridging the Structural Gap Between Encoding and Decoding for Data-To-Text Generation

Abstract: Generating sequential natural language descriptions from graph-structured data (e.g., knowledge graph) is challenging, partly because of the structural differences between the input graph and the output text. Hence, popular sequence-to-sequence models, which require serialized input, are not a natural fit for this task. Graph neural networks, on the other hand, can better encode the input graph but broaden the structural gap between the encoder and decoder, making faithful generation difficult. To narrow this gap, we propose DualEnc, a dual encoding model that can not only incorporate the graph structure, but can also cater to the linear structure of the output text. Empirical comparisons with strong single-encoder baselines demonstrate that dual encoding can significantly improve the quality of the generated text.