Optimization of Molecules via Deep Reinforcement Learning https://arxiv.org/abs/1810.08678
##Model
1). logP
- reward : penalized logP
- ./models/logp_model/train.py
2). QED
- reward : QED(Quantitative Estimate of Druglikeness)
- ./models/qed_model/train.py
3). logP_constraint
- reward : logP - w*(k-similarity)
- ./models/logP_constraint/train.py
4). multiobjective
- reward : w*similarity + (1-w)*QED
- ./models/multi_logp_qed_model/train.py
##Result