Bayesian Deep Learning for Spatial Interpolation in the Presence of Auxiliary Information

نویسندگان

چکیده

Abstract Earth scientists increasingly deal with ‘big data’. For spatial interpolation tasks, variants of kriging have long been regarded as the established geostatistical methods. However, and its (such regression kriging, in which auxiliary variables or derivatives these are included covariates) relatively restrictive models lack capabilities provided by deep neural networks. Principal among is feature learning: ability to learn filters recognise task-relevant patterns gridded data such images. Here, we demonstrate power learning a context showing how networks can automatically complex high-order point-sampled target relate those remote sensing) doing so produce detailed maps. In order cater for needs decision makers who require well-calibrated probabilities, also both aleatoric epistemic uncertainty be quantified our approach via Bayesian approximation known Monte Carlo dropout. example, national-scale probabilistic geochemical map from observations terrain elevation grid. By combining location information learned derivatives, achieves an excellent coefficient determination ( $$R^{2} = 0.74$$ R 2 = 0.74 ) near-perfect calibration on held-out test data. Our results indicate suitability feature-learning large-scale applications where matters. Graphic

برای دانلود باید عضویت طلایی داشته باشید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

study of cohesive devices in the textbook of english for the students of apsychology by rastegarpour

this study investigates the cohesive devices used in the textbook of english for the students of psychology. the research questions and hypotheses in the present study are based on what frequency and distribution of grammatical and lexical cohesive devices are. then, to answer the questions all grammatical and lexical cohesive devices in reading comprehension passages from 6 units of 21units th...

A hierarchical finite mixture model for Bayesian classification in the presence of auxiliary information

where y is a d-dimensional metrical (continuous) response random vector and x is a p-dimensional vector of metrical and/or categorical auxiliary variables (from now on simply referred to as covariates). Assume that the underlying population appears to be clustered into a fixed known number of say k groups and no training information is available regarding the group membership of observation uni...

متن کامل

the effect of learning strategies on the speaking ability of iranian students in the context of language institutes

the effect of learning strategies on the speaking ability of iranian students in the context of language institutes abstract language learning strategies are of the most important factors that help language learners to learn a foreign language and how they can deal with the four language skills specifically speaking skill effectively. acknowledging the great impact of learning strategies...

learners’ attitudes toward the effectiveness of mobile-assisted language learning (mall) in vocabulary acquisition in the iranian efl context: the case of word lists, audiobooks and dictionary use

رشد انفجاری تکنولوژی فرصت های آموزشی مهیج و جدیدی را پیش روی فراگیران و آموزش دهندگان گذاشته است. امروزه معلمان برای اینکه در امر آموزش زبان بروز باشند باید روش هایی را اتخاذ نمایند که درآن ها از تکنولوژی جهت کمک در یادگیری زبان دوم و چندم استفاده شده باشد. با در نظر گرفتن تحولاتی که رشته ی آموزش زبان در حال رخ دادن است هم اکنون زمان مناسبی برای ارزشیابی نگرش های موجود نسبت به تکنولوژی های جدید...

15 صفحه اول

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

ژورنال

عنوان ژورنال: Mathematical Geosciences

سال: 2022

ISSN: ['1874-8961', '1874-8953']

DOI: https://doi.org/10.1007/s11004-021-09988-0