Datasetについて
• USPTO Datasetができてから機械学習の適用トレンドが生まれた。
• Lowe が集めたものを Jin et al. などがそれぞれ自身のタスクようにFilterし
独自のSub setのデータセットを作成・評価している
7
https://rxn.res.ibm.com/wp-content/uploads/2019/05/acs_orlando_presentation.pdf
8.
研究の流れ
Molecular Transormer
Molecular Transformerfor Chemical
Reaction Prediction and Uncertainty
Estimation
seq2seq, Schwaller IBM
“Found in Translation": Predicting Outcomes
of Complex Organic Chemistry Reactions
using Neural Sequence-to-Sequence Models
2017
2016
2019
NLP系
8
Neural Symbolic ML
- Neural‐Symbolic Machine Learning for
Retrosynthesis and Reaction Prediction
AlphaChem
- Towards "AlphaChem": Chemical Synthesis Planning
with Tree Search and Deep Neural Network Policies
- AlphaChem: Planning chemical syntheses with deep
neural networks and symbolic AI
- Learning to Plan Chemical Syntheses
Computer Assisted Retrosynthesis
Computer-Assisted Retrosynthesis Based on
Molecular Similarity
WLDN / WLDN5
- Predicting Organic Reaction Outcomes with
Weisfeiler-Lehman Network
- A graph-convolutional neural network model for
the prediction of chemical reactivity
ELECTRO
A GENERATIVE MODEL FOR ELECTRON
PATHS
seq2seq, Liu Stanford
Retrosynthetic reaction prediction using
neural sequence-to-sequence models
seq2seq, Nam&Kim
Linking the Neural Machine Translation and
the Prediction of Organic Chemistry Reactions
Molecule CHEF
Generating Molecules via Chemical Reactions
Graph conv特徴量抽出+NN
GTPN
GRAPH TRANSFORMATION POLICY
NETWORK FOR CHEMICAL REACTION
PREDICTION
2018