From eda51bd539c94167aac399e277732317151881a8 Mon Sep 17 00:00:00 2001 From: George Batchkala Date: Wed, 16 Sep 2020 13:47:32 +0100 Subject: [PATCH] update Data section --- README.md | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index c0930e2..c981e8b 100644 --- a/README.md +++ b/README.md @@ -13,11 +13,14 @@ In this work, I explore ways of quantifying the confidence of machine learning m ## Data -Physical Chemistry Datasets from [MoleculeNet Benchmark Dataset Collection](http://moleculenet.ai/datasets-1). +I used the [MoleculeNet dataset](http://moleculenet.ai/datasets-1) which accompanies the [MoleculeNet benchmarking paper](https://pubs.rsc.org/en/content/articlelanding/2018/sc/c7sc02664a#!divAbstract), and in particular, I focused on the Physical Chemistry datasets: [ESOL](https://pubs.acs.org/doi/10.1021/ci034243x), [FreeSolv](https://link.springer.com/article/10.1007/s10822-014-9747-x), and [Lipophilicity](https://onlinelibrary.wiley.com/doi/abs/10.1002/cem.2718). The MoleculeNet datasets are widely used to validate machine learning models used to estimate a particular property directly from small molecules including drug-like compounds. + +The Physical Chemistry datasets can be downloaded from [MoleculeNet benchmark dataset collection](http://moleculenet.ai/datasets-1). ## Models + ## Obtaining Confidence Intervals