scholarly journals Generating stable molecules using imitation and reinforcement learning

2021 ◽  
Author(s):  
Søren Ager Meldgaard ◽  
Jonas Köhler ◽  
Henrik Lund Mortensen ◽  
Mads-Peter Verner Christiansen ◽  
Frank Noé ◽  
...  
Keyword(s):  
Reinforcement Learning ◽  
Chemical Space ◽  
Training Data ◽  
Graph Representation ◽  
Imitation Learning ◽  
Training Set ◽  
Machine Learning Methods ◽  
Multiple Copies ◽  
The Stability ◽  
3D Information

Abstract Chemical space is routinely explored by machine learning methods to discover interesting molecules, before time-consuming experimental synthesizing is attempted. However, these methods often rely on a graph representation, ignoring 3D information necessary for determining the stability of the molecules. We propose a reinforcement learning approach for generating molecules in cartesian coordinates allowing for quantum chemical prediction of the stability. To improve sample-efficiency we learn basic chemical rules from imitation learning on the GDB-11 database to create an initial model applicable for all stoichiometries. We then deploy multiple copies of the model conditioned on a specific stoichiometry in a reinforcement learning setting. The models correctly identify low energy molecules in the database and produce novel isomers not found in the training set. Finally, we apply the model to larger molecules to show how reinforcement learning further refines the imitation learning model in domains far from the training data.

scholarly journals Catastrophic Interference in Reinforcement Learning: A Solution Based on Context Division and Knowledge Distillation

2021 ◽  
Author(s):  
Tiantian Zhang ◽  
Xueqian Wang ◽  
Bin Liang ◽  
Bo Yuan
Keyword(s):  
Neural Networks ◽  
Reinforcement Learning ◽  
Training Data ◽  
Learning Ability ◽  
High Dimensional ◽  
Online Clustering ◽  
Computational Overhead ◽  
Knowledge Distillation ◽  
The Stability ◽  
Catastrophic Interference

The powerful learning ability of deep neural networks enables reinforcement learning (RL) agents to learn competent control policies directly from high-dimensional and continuous environments. In theory, to achieve stable performance, neural networks assume i.i.d. inputs, which unfortunately does no hold in the general RL paradigm where the training data is temporally correlated and non-stationary. This issue may lead to the phenomenon of "catastrophic interference" (a.k.a. "catastrophic forgetting") and the collapse in performance as later training is likely to overwrite and interfer with previously learned good policies. In this paper, we introduce the concept of "context" into the single-task RL and develop a novel scheme, termed as Context Division and Knowledge Distillation (CDaKD) driven RL, to divide all states experienced during training into a series of contexts. Its motivation is to mitigate the challenge of aforementioned catastrophic interference in deep RL, thereby improving the stability and plasticity of RL models. At the heart of CDaKD is a value function, parameterized by a neural network feature extractor shared across all contexts, and a set of output heads, each specializing on an individual context. In CDaKD, we exploit online clustering to achieve context division, and interference is further alleviated by a knowledge distillation regularization term on the output layers for learned contexts. In addition, to effectively obtain the context division in high-dimensional state spaces (e.g., image inputs), we perform clustering in the lower-dimensional representation space of a randomly initialized convolutional encoder, which is fixed throughout training. Our results show that, with various replay memory capacities, CDaKD can consistently improve the performance of existing RL algorithms on classic OpenAI Gym tasks and the more complex high-dimensional Atari tasks, incurring only moderate computational overhead.

scholarly journals DrugEx v3: Scaffold-Constrained Drug Design with Graph Transformer-based Reinforcement Learning

2021 ◽  
Author(s):  
Xuhan Liu ◽  
Kai Ye ◽  
Herman W. T. van Vlijmen ◽  
Adriaan P. IJzerman ◽  
Gerard J. P. van Westen
Keyword(s):  
Deep Learning ◽  
Reinforcement Learning ◽  
Drug Design ◽  
De Novo ◽  
Chemical Space ◽  
Molecular Structures ◽  
Rational Drug Design ◽  
Graph Representation ◽  
General Applicability ◽  
De Novo Drug Design

Due to the large drug-like chemical space available to search for feasible drug-like molecules, rational drug design often starts from specific scaffolds to which side chains/substituents are added or modified. With the rapid growth of the application of deep learning in drug discovery, a variety of effective approaches have been developed for de novo drug design. In previous work, we proposed a method named DrugEx, which can be applied in polypharmacology based on multi-objective deep reinforcement learning. However, the previous version is trained under fixed objectives similar to other known methods and does not allow users to input any prior information (i.e. a desired scaffold). In order to improve the general applicability, we updated DrugEx to design drug molecules based on scaffolds which consist of multiple fragments provided by users. In this work, the Transformer model was employed to generate molecular structures. The Transformer is a multi-head self-attention deep learning model containing an encoder to receive scaffolds as input and a decoder to generate molecules as output. In order to deal with the graph representation of molecules we proposed a novel positional encoding for each atom and bond based on an adjacency matrix to extend the architecture of the Transformer. Each molecule was generated by growing and connecting procedures for the fragments in the given scaffold that were unified into one model. Moreover, we trained this generator under a reinforcement learning framework to increase the number of desired ligands. As a proof of concept, our proposed method was applied to design ligands for the adenosine A2A receptor (A2AAR) and compared with SMILES-based methods. The results demonstrated the effectiveness of our method in that 100% of the generated molecules are valid and most of them had a high predicted affinity value towards A2AAR with given scaffolds.

scholarly journals On-the-Fly Machine Learning for Improving Image Resolution in Tomography

2019 ◽  
Vol 9 (12) ◽  
pp. 2445 ◽  
Author(s):  
Allard A. Hendriksen ◽  
Daniël M. Pelt ◽  
Willem Jan Palenstijn ◽  
Sophia B. Coban ◽  
Kees Joost Batenburg
Keyword(s):  
Machine Learning ◽  
Super Resolution ◽  
Image Resolution ◽  
Training Data ◽  
Training Set ◽  
Pixel Resolution ◽  
Machine Learning Methods ◽  
Scanning Procedure ◽  
Resolution Imaging ◽  
3D Volume

In tomography, the resolution of the reconstructed 3D volume is inherently limited by the pixel resolution of the detector and optical phenomena. Machine learning has demonstrated powerful capabilities for super-resolution in several imaging applications. Such methods typically rely on the availability of high-quality training data for a series of similar objects. In many applications of tomography, existing machine learning methods cannot be used because scanning such a series of similar objects is either impossible or infeasible. In this paper, we propose a novel technique for improving the resolution of tomographic volumes that is based on the assumption that the local structure is similar throughout the object. Therefore, our approach does not require a training set of similar objects. The technique combines a specially designed scanning procedure with a machine learning method for super-resolution imaging. We demonstrate the effectiveness of our approach using both simulated and experimental data. The results show that the proposed method is able to significantly improve resolution of tomographic reconstructions.

scholarly journals Catastrophic Interference in Reinforcement Learning: A Solution Based on Context Division and Knowledge Distillation

2021 ◽  
Author(s):  
Tiantian Zhang ◽  
Xueqian Wang ◽  
Bin Liang ◽  
Bo Yuan
Keyword(s):  
Neural Networks ◽  
Reinforcement Learning ◽  
Training Data ◽  
Learning Ability ◽  
High Dimensional ◽  
Online Clustering ◽  
Computational Overhead ◽  
Knowledge Distillation ◽  
The Stability ◽  
Catastrophic Interference

The powerful learning ability of deep neural networks enables reinforcement learning (RL) agents to learn competent control policies directly from high-dimensional and continuous environments. In theory, to achieve stable performance, neural networks assume i.i.d. inputs, which unfortunately does no hold in the general RL paradigm where the training data is temporally correlated and non-stationary. This issue may lead to the phenomenon of "catastrophic interference" (a.k.a. "catastrophic forgetting") and the collapse in performance as later training is likely to overwrite and interfer with previously learned good policies. In this paper, we introduce the concept of "context" into the single-task RL and develop a novel scheme, termed as Context Division and Knowledge Distillation (CDaKD) driven RL, to divide all states experienced during training into a series of contexts. Its motivation is to mitigate the challenge of aforementioned catastrophic interference in deep RL, thereby improving the stability and plasticity of RL models. At the heart of CDaKD is a value function, parameterized by a neural network feature extractor shared across all contexts, and a set of output heads, each specializing on an individual context. In CDaKD, we exploit online clustering to achieve context division, and interference is further alleviated by a knowledge distillation regularization term on the output layers for learned contexts. In addition, to effectively obtain the context division in high-dimensional state spaces (e.g., image inputs), we perform clustering in the lower-dimensional representation space of a randomly initialized convolutional encoder, which is fixed throughout training. Our results show that, with various replay memory capacities, CDaKD can consistently improve the performance of existing RL algorithms on classic OpenAI Gym tasks and the more complex high-dimensional Atari tasks, incurring only moderate computational overhead.

Scalable Approach to High Coverages on Oxides via Iterative Training of a Machine-Learning Algorithm

2019 ◽  
Author(s):  
Andrew Medford ◽  
Shengchun Yang ◽  
Fuzhu Liu
Keyword(s):  
Machine Learning ◽  
Chemical Potential ◽  
Learning Algorithm ◽  
Absolute Error ◽  
Low Energy ◽  
Training Data ◽  
High Coverage ◽  
Metal Compounds ◽  
Adsorption Energies ◽  
The Stability

Understanding the interaction of multiple types of adsorbate molecules on solid surfaces is crucial to establishing the stability of catalysts under various chemical environments. Computational studies on the high coverage and mixed coverages of reaction intermediates are still challenging, especially for transition-metal compounds. In this work, we present a framework to predict differential adsorption energies and identify low-energy structures under high- and mixed-adsorbate coverages on oxide materials. The approach uses Gaussian process machine-learning models with quantified uncertainty in conjunction with an iterative training algorithm to actively identify the training set. The framework is demonstrated for the mixed adsorption of CH<sub>x</sub>, NH<sub>x</sub> and OH<sub>x</sub> species on the oxygen vacancy and pristine rutile TiO<sub>2</sub>(110) surface sites. The results indicate that the proposed algorithm is highly efficient at identifying the most valuable training data, and is able to predict differential adsorption energies with a mean absolute error of ~0.3 eV based on <25% of the total DFT data. The algorithm is also used to identify 76% of the low-energy structures based on <30% of the total DFT data, enabling construction of surface phase diagrams that account for high and mixed coverage as a function of the chemical potential of C, H, O, and N. Furthermore, the computational scaling indicates the algorithm scales nearly linearly (N<sup>1.12</sup>) as the number of adsorbates increases. This framework can be directly extended to metals, metal oxides, and other materials, providing a practical route toward the investigation of the behavior of catalysts under high-coverage conditions.

scholarly journals Goal-driven active learning

2021 ◽  
Vol 35 (2) ◽  
Author(s):  
Nicolas Bougie ◽  
Ryutaro Ichise
Keyword(s):  
Decision Making ◽  
Reinforcement Learning ◽  
Learning Process ◽  
Real World ◽  
Imitation Learning ◽  
Learning Approaches ◽  
Wide Range ◽  
Fixed Set ◽  
Complex Decision Making ◽  
Complex Decision

AbstractDeep reinforcement learning methods have achieved significant successes in complex decision-making problems. In fact, they traditionally rely on well-designed extrinsic rewards, which limits their applicability to many real-world tasks where rewards are naturally sparse. While cloning behaviors provided by an expert is a promising approach to the exploration problem, learning from a fixed set of demonstrations may be impracticable due to lack of state coverage or distribution mismatch—when the learner’s goal deviates from the demonstrated behaviors. Besides, we are interested in learning how to reach a wide range of goals from the same set of demonstrations. In this work we propose a novel goal-conditioned method that leverages very small sets of goal-driven demonstrations to massively accelerate the learning process. Crucially, we introduce the concept of active goal-driven demonstrations to query the demonstrator only in hard-to-learn and uncertain regions of the state space. We further present a strategy for prioritizing sampling of goals where the disagreement between the expert and the policy is maximized. We evaluate our method on a variety of benchmark environments from the Mujoco domain. Experimental results show that our method outperforms prior imitation learning approaches in most of the tasks in terms of exploration efficiency and average scores.

scholarly journals Effects of Training Set Size on Supervised Machine-Learning Land-Cover Classification of Large-Area High-Resolution Remotely Sensed Data

2021 ◽  
Vol 13 (3) ◽  
pp. 368
Author(s):  
Christopher A. Ramezan ◽  
Timothy A. Warner ◽  
Aaron E. Maxwell ◽  
Bradley S. Price
Keyword(s):  
Machine Learning ◽  
Sample Size ◽  
Remotely Sensed ◽  
Training Data ◽  
Supervised Machine Learning ◽  
Sample Sizes ◽  
Remotely Sensed Data ◽  
Large Area ◽  
Training Set ◽  
Set Size

The size of the training data set is a major determinant of classification accuracy. Nevertheless, the collection of a large training data set for supervised classifiers can be a challenge, especially for studies covering a large area, which may be typical of many real-world applied projects. This work investigates how variations in training set size, ranging from a large sample size (n = 10,000) to a very small sample size (n = 40), affect the performance of six supervised machine-learning algorithms applied to classify large-area high-spatial-resolution (HR) (1–5 m) remotely sensed data within the context of a geographic object-based image analysis (GEOBIA) approach. GEOBIA, in which adjacent similar pixels are grouped into image-objects that form the unit of the classification, offers the potential benefit of allowing multiple additional variables, such as measures of object geometry and texture, thus increasing the dimensionality of the classification input data. The six supervised machine-learning algorithms are support vector machines (SVM), random forests (RF), k-nearest neighbors (k-NN), single-layer perceptron neural networks (NEU), learning vector quantization (LVQ), and gradient-boosted trees (GBM). RF, the algorithm with the highest overall accuracy, was notable for its negligible decrease in overall accuracy, 1.0%, when training sample size decreased from 10,000 to 315 samples. GBM provided similar overall accuracy to RF; however, the algorithm was very expensive in terms of training time and computational resources, especially with large training sets. In contrast to RF and GBM, NEU, and SVM were particularly sensitive to decreasing sample size, with NEU classifications generally producing overall accuracies that were on average slightly higher than SVM classifications for larger sample sizes, but lower than SVM for the smallest sample sizes. NEU however required a longer processing time. The k-NN classifier saw less of a drop in overall accuracy than NEU and SVM as training set size decreased; however, the overall accuracies of k-NN were typically less than RF, NEU, and SVM classifiers. LVQ generally had the lowest overall accuracy of all six methods, but was relatively insensitive to sample size, down to the smallest sample sizes. Overall, due to its relatively high accuracy with small training sample sets, and minimal variations in overall accuracy between very large and small sample sets, as well as relatively short processing time, RF was a good classifier for large-area land-cover classifications of HR remotely sensed data, especially when training data are scarce. However, as performance of different supervised classifiers varies in response to training set size, investigating multiple classification algorithms is recommended to achieve optimal accuracy for a project.

scholarly journals Possibility of Autonomous Estimation of Shiba Goat’s Estrus and Non-Estrus Behavior by Machine Learning Methods

Animals ◽  
2020 ◽  
Vol 10 (5) ◽  
pp. 771
Author(s):  
Toshiya Arakawa
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Random Forest ◽  
Markov Models ◽  
Tracking System ◽  
Video Tracking ◽  
Training Data ◽  
Support Vector ◽  
Learning Methods ◽  
Machine Learning Methods

Mammalian behavior is typically monitored by observation. However, direct observation requires a substantial amount of effort and time, if the number of mammals to be observed is sufficiently large or if the observation is conducted for a prolonged period. In this study, machine learning methods as hidden Markov models (HMMs), random forests, support vector machines (SVMs), and neural networks, were applied to detect and estimate whether a goat is in estrus based on the goat’s behavior; thus, the adequacy of the method was verified. Goat’s tracking data was obtained using a video tracking system and used to estimate whether they, which are in “estrus” or “non-estrus”, were in either states: “approaching the male”, or “standing near the male”. Totally, the PC of random forest seems to be the highest. However, The percentage concordance (PC) value besides the goats whose data were used for training data sets is relatively low. It is suggested that random forest tend to over-fit to training data. Besides random forest, the PC of HMMs and SVMs is high. However, considering the calculation time and HMM’s advantage in that it is a time series model, HMM is better method. The PC of neural network is totally low, however, if the more goat’s data were acquired, neural network would be an adequate method for estimation.

scholarly journals Diversity oriented Deep Reinforcement Learning for targeted molecule generation

2021 ◽  
Vol 13 (1) ◽  
Author(s):  
Tiago Pereira ◽  
Maryam Abbasi ◽  
Bernardete Ribeiro ◽  
Joel P. Arrais
Keyword(s):  
Neural Networks ◽  
Deep Learning ◽  
Reinforcement Learning ◽  
Deep Neural Networks ◽  
Chemical Space ◽  
Biological Properties ◽  
Training Process ◽  
Training Strategy ◽  
Inhibitory Power ◽  
Exploratory Strategy

AbstractIn this work, we explore the potential of deep learning to streamline the process of identifying new potential drugs through the computational generation of molecules with interesting biological properties. Two deep neural networks compose our targeted generation framework: the Generator, which is trained to learn the building rules of valid molecules employing SMILES strings notation, and the Predictor which evaluates the newly generated compounds by predicting their affinity for the desired target. Then, the Generator is optimized through Reinforcement Learning to produce molecules with bespoken properties. The innovation of this approach is the exploratory strategy applied during the reinforcement training process that seeks to add novelty to the generated compounds. This training strategy employs two Generators interchangeably to sample new SMILES: the initially trained model that will remain fixed and a copy of the previous one that will be updated during the training to uncover the most promising molecules. The evolution of the reward assigned by the Predictor determines how often each one is employed to select the next token of the molecule. This strategy establishes a compromise between the need to acquire more information about the chemical space and the need to sample new molecules, with the experience gained so far. To demonstrate the effectiveness of the method, the Generator is trained to design molecules with an optimized coefficient of partition and also high inhibitory power against the Adenosine $$A_{2A}$$ A 2 A and $$\kappa$$ κ opioid receptors. The results reveal that the model can effectively adjust the newly generated molecules towards the wanted direction. More importantly, it was possible to find promising sets of unique and diverse molecules, which was the main purpose of the newly implemented strategy.

Sign in / Sign up

Export Citation Format

Share Document