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 statistics community
-
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.
-
MIT Open Course on Artificial Intelligence.
-
Python A very useful language that underpins many AI frameworks
-
R
An extremely useful numerical analysis platform; also used for many AI applications
-
Tensorflow
A very popular and powerful framework for Neural Network applications
-
Keras
High-level neural network library in Python that makes starting with NNs very easy
-
XGBoost
Scalable gradient boosting library
-
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
-
Google Cloud
Google's cloud computing platform
-
Amazon AWS
Amazon's cloud computing platform
-
European Open Science Cloud
Cloud for European research data
https://ec.europa.eu/research/openscience/index.cfm?pg=open-science-cloud
- (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.