Entity

Time filter

Source Type

Seattle, WA, United States

Newcombe C.,Amazon Inc.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2014

Since 2011, engineers at Amazon have been using TLA+ to help solve difficult design problems in critical systems. This paper describes the reasons why we chose TLA+ instead of other methods, and areas in which we would welcome further progress. © 2014 Springer-Verlag. Source


Liu J.,Xian Jiaotong University | Wang J.,Xian Jiaotong University | Zheng Q.,Xian Jiaotong University | Zhang W.,Amazon Inc. | Jiang L.,Xian Jiaotong University
Knowledge-Based Systems | Year: 2012

A knowledge map can be viewed as a directed graph, in which each node is a knowledge unit (KU), and each edge is a learning-dependency between two KUs. Understanding the topological properties of knowledge map can help us gain better insights into human cognition structure and its mechanism, design better knowledge map construction algorithms, and guide learners' navigational learning through knowledge map. In this paper, we perform topological analysis on 12 knowledge maps from computer science, mathematics, and physics. We discover that they exhibit small-world and scale-free properties like many other networks. Specifically, we show the locality of learning-dependency and hierarchical modular structure in the 12 knowledge maps. In addition, we study how KUs affect the network efficiency by removing KUs based on different centrality measures. We find that the importance of KUs varies greatly. © 2012 Elsevier B.V. All rights reserved. Source


Foster D.,University of Pennsylvania | Foster D.,Amazon Inc.
Obesity | Year: 2016

Objective The objective was twofold: (1) to estimate for each individual the body mass index (BMI) which is associated with the lowest risk of death, and (2) to study variants of the BMI formula to determine which gives the best predictions of death. Methods Treating BMI = mass/height2 as a continuous variable and estimating its interaction effects with several other variables, this study analyzed the NIH-AARP study data set of approximately 566,000 individuals and fit Cox proportional hazards models to the survival times. Results For each individual, a "personalized optimal BMI," the BMI for that individual which, according to the model, is associated with the lowest risk of death, is estimated. The average personalized optimal BMI is approximately 26, which is in the current "overweight" category. In fact, mass/height is a better predictor of death on the data set than BMI itself. Conclusions The model suggests that an individual's "optimal" BMI depends on his or her features; "one-size-fits-all" recommendations may be not best. © 2016 The Obesity Society. Source


Zhang X.,University of Florida | Zhang X.,Amazon Inc. | Bolton J.,University of Florida | Gader P.,University of Florida
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing | Year: 2014

A new learning algorithm based on Gibbs sampling to learn the parameters of continuous Hidden Markov Models (HMMs) with multivariate Gaussian mixtures is presented. The proposed sampling algorithm outperformed the standard expectation-maximization (EM) algorithm and a minimum classification error algorithm when applied to a synthetic data set. The proposed algorithm outperforms the state of the art when applied to landmine detection using ground penetrating radar (GPR) data. © 2008-2012 IEEE. Source


Yan J.,Amazon Inc. | Zhang K.,Florida International University | Zhang C.,University of Alabama at Birmingham | Chen S.-C.,Florida International University | Narasimhan G.,Florida International University
IEEE Transactions on Geoscience and Remote Sensing | Year: 2015

The snake algorithm has been proposed to solve many remote sensing and computer vision problems such as object segmentation, surface reconstruction, and object tracking. This paper introduces a framework for 3-D building model construction from LIDAR data based on the snake algorithm. It consists of nonterrain object identification, building and tree separation, building topology extraction, and adjustment by the snake algorithm. The challenging task in applying the snake algorithm to building topology adjustment is to find the global minima of energy functions derived for 2-D building topology. The traditional snake algorithm uses dynamic programming for computing the global minima of energy functions which is limited to snake problems with 1-D topology (i.e., a contour) and cannot handle problems with 2-D topology. In this paper, we have extended the dynamic programming method to address the snake problems with a 2-D planar topology using a novel graph reduction technique. Given a planar snake, a set of reduction operations is defined and used to simplify the graph of the planar snake into a set of isolated vertices while retaining the minimal energy of the graph. Another challenging task for 3-D building model reconstruction is how to enforce different kinds of geometric constraints during building topology refinement. This framework proposed two energy functions, deviation and direction energy functions, to enforce multiple geometric constraints on 2-D topology refinement naturally and efficiently. To examine the effectiveness of the framework, the framework has been applied on different data sets to construct 3-D building models from airborne LIDAR data. The results demonstrate that the proposed snake algorithm successfully found the global optima in polynomial time for all of the building topologies and generated satisfactory 3-D models for most of the buildings in the study areas. © 2014 IEEE. Source

Discover hidden collaborations