Machine Learning and AI Discussion Resources

This is a collection of the resources mentioned during the Machine Learning and AI discussion group at the HDRUK UK Fellow's day (2018-11-15).

The rough minutes of the discussion are located at https://alastair-droop.github.io/HDRUK-fellows-day-2018/minutes.md.

Online Resources

• https://blackswans.io/

  Online statistics community

• https://aws.amazon.com/blogs/machine-learning/classifying-high-resolution-chest-x-ray-medical-images-with-amazon-sagemaker/

  An Amazon Web Services blog about classification of chest x-ray images using Amazon SageMaker

• https://www.kaggle.com/goelrajat/prediciting-pneumonia-from-chest-xray

  Tutorial on using TensorFlow for chest x-ray classification (Data are here.)

• https://datascience.stackexchange.com

  Stack exchange site for data science-related questions.

• https://codelabs.developers.google.com/codelabs/cloud-tensorflow-mnist/#0

  Tutorial on using TensorFLow with the MNIST data.

• https://www.tensorflow.org/tutorials/

  A good set of initial tutorials for getting up to speed with TensorFlow. Also uses the MNIST data.

• https://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-034-artificial-intelligence-fall-2010/

  MIT Open Course on Artificial Intelligence.

Tools

• Python A very useful language that underpins many AI frameworks

  https://www.python.org

• R

  An extremely useful numerical analysis platform; also used for many AI applications

  https://www.r-project.org

• Tensorflow

  A very popular and powerful framework for Neural Network applications

  https://www.tensorflow.org

• Keras

  High-level neural network library in Python that makes starting with NNs very easy

  https://keras.io

• XGBoost

  Scalable gradient boosting library

  https://xgboost.ai

• eMedLab

  MRC effort to coordinate and store medical and biological data

  https://wiki.emedlab.ac.uk/display/EPW/Welcome+to+MRC+eMedLab

• Microsoft Azure

  Microsoft's cloud computing platform

  https://azure.microsoft.com/en-gb/

• Google Cloud

  Google's cloud computing platform

  https://cloud.google.com

• Amazon AWS

  Amazon's cloud computing platform

  https://aws.amazon.com

• European Open Science Cloud

  Cloud for European research data

  https://ec.europa.eu/research/openscience/index.cfm?pg=open-science-cloud

Notes

• (Thanks to Waty for these)
• It is often unnecessary for you to learn all of the details of the tools you need to use.
• Rather, you should find help and collaboration with domain experts in Computer Science or mathematics.
• With their help, you should understand:
  • Intuitively, what the algorithms are doing;
  • What input data is required;
  • What output data is generated; and
  • Under what conditions will the algorithm fail.