The University of Auckland
Browse
Using_Artificial_Intelligence_to_forecast_the_location_of_earthquake_and_post-earthquake-induced_landslides.pdf (5.01 MB)

Using Artificial Intelligence to forecast the location of earthquake and post-earthquake-induced landslides

Download (5.01 MB)
presentation
posted on 2019-09-24, 02:57 authored by Rand Huso

This presentation will be on the progress we've made trying to forecast landslides caused by earthquakes and precipitation. Good quality input datasets are used to train a Tensorflow / Keras Sequential neural network whose parameters are determined by using Bayesian hyperparameter optimization (BayesianOptimization), and whose training details are established by using KFold (sklearn) iterations with the EarlyStopping monitor. The resulting trained neural network is then applied to all the available data, and some graphs are created to show the results.


Funding

New Zealand eScience Infrastructure

Ministry of Business, Innovation and Employment

Find out more...

History

Publisher

New Zealand eScience Infrastructure

Spatial coverage

New Zealand

Temporal coverage: start

2019-09-05

Temporal coverage: end

2019-09-06

Usage metrics

    NeSI

    Categories

    No categories selected

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC