introduction to deep learning

general resources

there are many great resources to introduce the concepts and principles behind deep learning. some of the best online resources include: machine learning mastery, kaggle and youtube. there are also paid courses (eg data camp, fast.ai) and tutorials at sites such as udemy.

• deep learning with python
• deep learning with R
• francois chollet on twitter
• ian goodfellow on twitter

languages and libraries

• R
• python
• keras
• tensorflow

recommended IDEs

• R studio: best for R
• atom: best for python

workflow

we use teams for communication, trello for project organisation and google docs for writing collaboratively. this ensures that there is only ever a single copy of any document. we use paperpile for reference management - greg can give you an access code.

my github repository, where most code should be available.

previous output/documents

• 2017 Ruth Muir MSc Project
• Machine-Learning (neural network) driven algorithmic classification of Type 1 or Type 2 diabetes at the time of presentation significantly outperforms experienced clinician classification
• Parallel time-series analysis of HbA1c and Systolic BP using a recurrent neural network (RNN) stratifies for 1-year mortality in Type 2 Diabetes independent of age and parameter variability. DUK. London
• Parallel time-series analysis of HbA1c, Systolic Blood Pressure and BMI using Recurrent Neural Networks stratifies for 1-year mortality in T2DM, independent of age and parameter variability. ATTD. Vienna
• Time-series analysis of HbA1c using Recurrent Neural Networks stratifies for 1-year mortality in Type 1 and Type 2 Diabetes, independent of age and variability. ATTD. Vienna
• MyDiabetesIQ. machine learning for decision support & outcome prediction in diabetes. Automated Clinical Epidemiology Studies workshop, University of Birmingham. 141217