Introduction to Machine Learning techniques for Geoscience

Slide 1

Slide 1 text

Machine Learning Techniques for Geoscience Applications Jesper Dramsch

Slide 2

Slide 2 text

Where are we going today? ● My Background ● What’s Machine Learning? ● Geo to ML?! ● What’s important in ML?

Slide 3

Slide 3 text

What is Machine Learning? And why should I care?

Slide 4

Slide 4 text

No content

Slide 5

Slide 5 text

if x > 5 Classical rule-based and expert systems

Slide 6

Slide 6 text

Computers ﬁgure it out based on the data

Slide 7

Slide 7 text

Tossing Bot! (http://tossingbot.cs.princeton.edu/)

Slide 8

Slide 8 text

Models that do not have external information Blackbox models

Slide 9

Slide 9 text

What is AI, ML, and Deep Learning?

Slide 10

Slide 10 text

Style Transfer for Multiple Tasks

Slide 11

Slide 11 text

Everybody dance now! (https://arxiv.org/pdf/1808.07371.pdf)

Slide 12

Slide 12 text

What is the connection Why do we care in geoscience?!

Slide 13

Slide 13 text

Take seismic data

Slide 14

Slide 14 text

Get some “interpretation”

Slide 15

Slide 15 text

Or get some “inversion”

Slide 16

Slide 16 text

Neural Networks The new hype that is older than all of us

Slide 17

Slide 17 text

A Simple Neuron 1 1 + 2

Slide 18

Slide 18 text

A Small Neural Network

Slide 19

Slide 19 text

A Deep Neural Network

Slide 20

Slide 20 text

Diﬀerent Combinations In Neural Networks

Slide 21

Slide 21 text

Is that all? What else is there to make Artiﬁcial Intelligence?!

Slide 22

Slide 22 text

Convolutional Neural Networks The machine learning that sees

Slide 23

Slide 23 text

A neural network as it is used today

Slide 24

Slide 24 text

The learned filters with diﬀerent abstractions

Slide 25

Slide 25 text

A Geoscience Application

Slide 26

Slide 26 text

Matching of seismic data

Slide 27

Slide 27 text

The diﬀerences between unmatched and matched

Slide 28

Slide 28 text

Compared to a baseline method

Slide 29

Slide 29 text

Take it to the next level

Slide 30

Slide 30 text

Real Data (CC-BY A. Çuğun) 30 How do we generate data like this with a Neural Network?

Slide 31

Slide 31 text

Real Data (CC-BY A. Çuğun) Network Data Legend 31 Generative Adversarial Networks

Slide 32

Slide 32 text

Latent Space (CC-BY Hudson) Real Data (CC-BY A. Çuğun) Network Data Legend 32 Start from latent space

Slide 33

Slide 33 text

Latent Space (CC-BY Hudson) Generator (CC-BY-SA T. Vaughn) Real Data (CC-BY A. Çuğun) Network Data Legend 33 Feed it to a Generator

Slide 34

Slide 34 text

Latent Space (CC-BY Hudson) Generator (CC-BY-SA T. Vaughn) Fake Data (CC-BY Tony A.) Real Data (CC-BY A. Çuğun) Network Data Legend 34 That generates some Fake Data

Slide 35

Slide 35 text

Latent Space (CC-BY Hudson) Generator (CC-BY-SA T. Vaughn) Fake Data (CC-BY Tony A.) Real Data (CC-BY A. Çuğun) Discriminator(CC-BY Brickset) Network Data Legend 35 Discriminator judges if sample is real

Slide 36

Slide 36 text

Latent Space (CC-BY Hudson) Generator (CC-BY-SA T. Vaughn) Fake Data (CC-BY Tony A.) Real Data (CC-BY A. Çuğun) Loss (CC-BY S. MacEntee) Discriminator(CC-BY Brickset) Network Data Legend (CC-BY Dramsch) 36 Both networks learn regardless if D is correct

Slide 37

Slide 37 text

What’s the application?

Slide 38

Slide 38 text

Some seismic data

Slide 39

Slide 39 text

An “inversion” Generator (CC-BY-SA T. Vaughn) generated by a neural network

Slide 40

Slide 40 text

● Generating Synthetic Data ● Earthquake localisation ● Drill Core Processing ● Clustering of Well Data ● Prediction of Volcanic activity ● Satellite Data What else is possible?

Slide 41

Slide 41 text

Many Other Machine Learning Algorithms

Slide 42

Slide 42 text

Seismogram Tagging (https://www.nature.com/articles/s41467-020-17591-w)

Slide 43

Slide 43 text

Core Image Segmentation (https://joss.theoj.org/papers/10.21105/joss.01969)

Slide 44

Slide 44 text

Well Log Fascies Prediction (https://doi.org/10.1190/tle35100906.1)

Slide 45

Slide 45 text

Remote Sensing Data (https://doi.org/10.1117/1.JRS.11.036028)

Slide 46

Slide 46 text

Some Tips to Get Started?

Slide 47

Slide 47 text

Start Projects Application as important as Theory ● Participate in Hackathons ● Build Apps ● Start Small ● Compete on Kaggle ● Get familiar with the Math

Slide 48

Slide 48 text

70 Years of Machine Learning in Geoscience https://authors.elsevier.com/b/1bpiNEroEut0M

Slide 49

Slide 49 text

No content

Slide 50

Slide 50 text

Available Data For Projects and Experiments ● Open Data: https://wiki.seg.org/wiki/Open_d ata ● Full Reservoir with Wells: https://wiki.seg.org/wiki/The_No rth_Sea_Volve_Data_Village ● Seismic Interpretation: https://github.com/yalaudah/faci es_classification_benchmark ● Salt Identification in Seismic: http://bit.ly/kaggle-salt