MACHINE-LEARNING IN CHEMISTRY Yashasvi S. Ranawat Filippo - - PowerPoint PPT Presentation

▶
machine learning in chemistry
SMART_READER_LITE
LIVE PREVIEW

MACHINE-LEARNING IN CHEMISTRY Yashasvi S. Ranawat Filippo - - PowerPoint PPT Presentation

Oct 12, 2023 520 likes •627 views

MACHINE-LEARNING IN CHEMISTRY Yashasvi S. Ranawat Filippo Federici UNSUPERVISED LEARNING finds similarities in complex data records does not require knowledge of properties/outputs, only descriptors/inputs sensitive to the

slide-1
SLIDE 1

MACHINE-LEARNING IN CHEMISTRY

Yashasvi S. Ranawat Filippo Federici

slide-2
SLIDE 2

UNSUPERVISED LEARNING

  • finds similarities in complex data records
  • does not require knowledge of properties/outputs, only descriptors/inputs
  • sensitive to the similarity measure
  • requires the user to know how many classes to expect
  • useful to reduce data dimensionality
slide-3
SLIDE 3

SUPERVISED LEARNING

  • learns input → output relation from examples
  • training data is the limit
  • useful for fast screening and classification
slide-4
SLIDE 4

DESCRIPTORS FOR CHEMISTRY

ML methods need a computer-friendly way to input the atomistic system: 010110101010001011100100010001111110

easy for us easy for cpu

Issues for ML:

  • arbitrary size
  • arbitrary order

Ideal features:

  • general
  • compact
  • unique
  • invariant *
  • smooth
  • fast

* invariants are determined by the physics of the quantity to predict from the descriptor!

slide-5
SLIDE 5

DESCRIPTORS FOR CHEMISTRY

ML methods need a computer-friendly way to input the atomistic system: 010110101010001011100100010001111110

global descriptor local/atomic descriptor 010110101010001011100001111110 110100011110001011100001111110 110100011110000110010111111110

slide-6
SLIDE 6

DESCRIPTORS FOR CHEMISTRY

1. ACSF.ipynb 2. SOAP.ipynb 3. MBTR.ipynb 4. LMBTR.ipynb

slide-7
SLIDE 7

SUPERVISED ML METHODS

  • KRR - TotalEnergy.ipynb

KERNEL RIDGE REGRESSION

slide-8
SLIDE 8

SUPERVISED ML METHODS

1. NeuralNetwork - Intro.ipynb 2. ACSF-Dimer.ipynb 3. NeuralNetwork - TotalEnergy.ipynb 4. NeuralNetwork - AtomicCharges.ipynb

NEURAL NETWORKS

slide-9
SLIDE 9

TAKEHOME

  • all notebooks and data is in github: https://github.com/fullmetalfelix/ML-CSC-tutorial

○ notebooks require Jupyter python module ○ data in numpy array form

  • useful goodies:

○ describe package: https://github.com/SINGROUP/describe ■ python package for creating machine learning descriptors for atomistic systems