scholarly journals Hard-threshold-Neural-Network based Prediction of Organic Synthetic Outcomes

2020 ◽  
Author(s):  
Haoyang Hu ◽  
Zhihong Yuan
Keyword(s):  
Neural Network ◽  
Prediction Accuracy ◽  
Organic Molecules ◽  
Backpropagation Algorithm ◽  
Forward Reaction ◽  
Comparative Tests ◽  
Drug Discovery And Development ◽  
Reaction Prediction ◽  
Propagation Algorithm ◽  
Retrosynthetic Analysis

Abstract Retrosynthetic analysis is the canonical technique to plan the synthesis route of organic molecules in drug discovery and development. In this technique, the screening of synthetic tree branches requires accurate forward reaction prediction, but existing software is still far from completing this step independently. Previous studies have attempted to apply neural network in the forward reaction prediction, but the accuracy is not satisfying. Through using the Edit Vector based description and Extended-Connectivity Fingerprints to transform reaction into vector, the presented work focuses on the update of neural network to improve the template-based forward reaction prediction. Hard-threshold activation and target propagation algorithm are implemented by introducing the mixed-convex combinatorial optimization. Comparative tests are conducted to explore the optimal hyperparameter set. Using 15, 000 experimental reaction extracted from granted United States patents, the proposed hard-threshold neural network is systematically trained and tested. The results demonstrate that a higher prediction accuracy is obtained when compared to the traditional neural network with backpropagation algorithm. Some successfully predicted reaction examples are also briefly illustrated.

scholarly journals Hard-threshold-Neural-Network based Prediction of Organic Synthetic Outcomes

2019 ◽  
Author(s):  
Haoyang Hu ◽  
Zhihong Yuan
Keyword(s):  
Neural Network ◽  
Prediction Accuracy ◽  
Organic Molecules ◽  
Backpropagation Algorithm ◽  
Forward Reaction ◽  
Comparative Tests ◽  
Reaction Prediction ◽  
Medicine Development ◽  
Propagation Algorithm ◽  
Retrosynthetic Analysis

Abstract Retrosynthetic analysis is the canonical technique to plan the synthesis route of organic molecules in medicine development. In this technique, the screening of synthetic tree branches requires accurate forward reaction prediction, but existing software is still far from completing this step independently. Previous studies have attempted to apply neural network in the forward reaction prediction, but the accuracy is not satisfying. Through using the Edit-based Description and Extended-Connectivity Fingerprints to transform reaction into vector, the presented work focuses on the update of neural network to improve the template-based forward reaction prediction. Hard-threshold activation and target propagation algorithm are implemented by introducing the mixed-convex combinatorial optimization. Comparative tests are conducted to explore the optimal hyperparameter set. Using 15 000 experimental reaction records from granted United States patents, the proposed hard-threshold neural network is systematically trained and tested. The results demonstrate that a higher prediction accuracy is obtained when compared to the traditional neural network with backpropagation algorithm. Indeed, the prediction accuracy of the proposed hard-threshold neural network can reach 73.9% which is higher than Coley’s result with 71.8% ( Coley et al. ACS Cent. Sci, 2017 ). Some successfully predicted reaction examples are also briefly discussed.

scholarly journals Hard-threshold neural network-based prediction of organic synthetic outcomes

2020 ◽  
Author(s):  
Haoyang Hu ◽  
Zhihong Yuan
Keyword(s):  
Neural Network ◽  
Prediction Accuracy ◽  
Organic Molecules ◽  
Backpropagation Algorithm ◽  
Forward Reaction ◽  
Drug Discovery And Development ◽  
Reaction Prediction ◽  
The Neural Network ◽  
Reaction Data ◽  
Retrosynthetic Analysis

Abstract Retrosynthetic analysis is a canonical technique for planning the synthesis route of organic molecules in drug discovery and development. In this technique, the screening of synthetic tree branches requires accurate forward reaction prediction, but existing software is far from completing this step independently. Previous studies attempted to apply a neural network to forward reaction prediction, but the accuracy was not satisfying. Through using the Edit Vector-based description and extended-connectivity fingerprints to transform the reaction into a vector, this study focuses on the update of the neural network to improve the template-based forward reaction prediction. Hard-threshold activation and the target propagation algorithm are implemented by introducing mixed convex-combinatorial optimization. Comparative tests were conducted to explore the optimal hyperparameter set. Using 15,000 experimental reaction data extracted from granted United States patents, the proposed hard-threshold neural network was systematically trained and tested. The results demonstrated that a higher prediction accuracy was obtained than that for the traditional neural network with backpropagation algorithm. Some successfully predicted reaction examples are also briefly illustrated.

scholarly journals Hard-threshold-Neural-Network based Prediction of Organic Synthetic Outcomes

2020 ◽  
Author(s):  
Haoyang Hu ◽  
Zhihong Yuan
Keyword(s):  
Neural Network ◽  
Prediction Accuracy ◽  
Organic Molecules ◽  
Backpropagation Algorithm ◽  
Forward Reaction ◽  
Drug Discovery And Development ◽  
Reaction Prediction ◽  
The Neural Network ◽  
Reaction Data ◽  
Retrosynthetic Analysis

Abstract Retrosynthetic analysis is a canonical technique for planning the synthesis route of organic molecules in drug discovery and development. In this technique, the screening of synthetic tree branches requires accurate forward reaction prediction, but existing software is far from completing this step independently. Previous studies attempted to apply a neural network to forward reaction prediction, but the accuracy was not satisfying. Through using the Edit Vector-based description and extended-connectivity fingerprints to transform the reaction into a vector, this study focuses on the update of the neural network to improve the template-based forward reaction prediction. Hard-threshold activation and the target propagation algorithm are implemented by introducing mixed convex-combinatorial optimization. Comparative tests were conducted to explore the optimal hyperparameter set. Using 15,000 experimental reaction data extracted from granted United States patents, the proposed hard-threshold neural network was systematically trained and tested. The results demonstrated that a higher prediction accuracy was obtained than that for the traditional neural network with backpropagation algorithm. Some successfully predicted reaction examples are also briefly illustrated.

scholarly journals Completion of Partial Reaction Equations

2020 ◽  
Author(s):  
Alain C. Vaucher ◽  
Philippe Schwaller ◽  
Teodoro Laino
Keyword(s):  
Deep Learning ◽  
Chemical Transformations ◽  
Learning Models ◽  
Forward Reaction ◽  
Partial Reaction ◽  
Reaction Prediction ◽  
Retrosynthetic Analysis ◽  
Incomplete Reaction ◽  
Reaction Equations ◽  
Deep Learning Model

We present a deep-learning model for inferring missing molecules in reaction equations. Such an algorithm features multiple interesting behaviors. First, it can infer the necessary reagents and solvents in chemical transformations specified only in terms of main compounds, as often resulting from retrosynthetic analyses. The completion with necessary reagents ensures that reaction equations are compatible with deep-learning models relying on a complete reaction specification. Second, it can cure existing datasets by detecting missing compounds, such as reagents that are essential for given classes of reactions. Finally, this model is a generalization of models for forward reaction prediction and retrosynthetic analysis, as both can be formulated in terms of incomplete reaction equations. We illustrate that a single trained model, based on the transformer architecture and acting on reaction SMILES strings, can address all three points.<br><br>Workshop paper at the Machine Learning for Molecules Workshop at NeurIPS 2020.<br>

scholarly journals Completion of Partial Reaction Equations

2020 ◽  
Author(s):  
Alain C. Vaucher ◽  
Philippe Schwaller ◽  
Teodoro Laino
Keyword(s):  
Deep Learning ◽  
Chemical Transformations ◽  
Learning Models ◽  
Forward Reaction ◽  
Partial Reaction ◽  
Reaction Prediction ◽  
Retrosynthetic Analysis ◽  
Incomplete Reaction ◽  
Reaction Equations ◽  
Deep Learning Model

We present a deep-learning model for inferring missing molecules in reaction equations. Such an algorithm features multiple interesting behaviors. First, it can infer the necessary reagents and solvents in chemical transformations specified only in terms of main compounds, as often resulting from retrosynthetic analyses. The completion with necessary reagents ensures that reaction equations are compatible with deep-learning models relying on a complete reaction specification. Second, it can cure existing datasets by detecting missing compounds, such as reagents that are essential for given classes of reactions. Finally, this model is a generalization of models for forward reaction prediction and retrosynthetic analysis, as both can be formulated in terms of incomplete reaction equations. We illustrate that a single trained model, based on the transformer architecture and acting on reaction SMILES strings, can address all three points.<br><br>Workshop paper at the Machine Learning for Molecules Workshop at NeurIPS 2020.<br>

scholarly journals IMPLEMENTASI ALGORITMA BACKPROPAGATION UNTUK MEMPREDIKSI KELULUSAN MAHASISWA

2018 ◽  
Vol 5 (2) ◽  
pp. 169
Author(s):  
Muhammad Dedek Yalidhan
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Information System ◽  
Prediction Accuracy ◽  
Confusion Matrix ◽  
The Other ◽  
Training Data ◽  
Backpropagation Algorithm ◽  
Testing Data ◽  
Artificial Neural

<p><em>Student’s graduation is one kind of the college accreditation elements by BAN-PT. Because of that. Information System is one of the department in STMIK Banjarbaru, there is no application has been implemented to predict imprecisely of student’s graduation time so far, which causes on time graduation percentage tend low every year. Therefore the accurate student’s graduation prediction can help the committe to choose the correct decisions in order to prevent the imprecisely of student’s graduation time. In this research, the backpropagation algorithm of artificial neural network will be implemented into the application with the output result as delayed and on time graduation. This reseach is using 318 data samples which the 70 % of it will be used as the training data and the other 30 % will be used as testing data. From the calculation of confusion matrix table’s the percentage of the prediction accuracy is 98.97 %.</em></p><p><em></em><em><strong>Keywords</strong>: student’s graduation, artificial neural network, backpropagation, confusion matrix</em></p><p><em></em><em>Kelulusan mahasiswa merupakan salah satu elemen dalam standar akreditasi perguruan tinggi oleh BAN-PT. Sistem Informasi adalah salah satu program studi yang ada di STMIK Banjarbaru, selama ini belum ada aplikasi yang diimplementasikan untuk memprediksi ketidaktepatan waktu kelulusan mahasiswanya yang menyebabkan angka kelulusan tepat waktu cenderung rendah setiap tahunnya. Oleh sebab itu, prediksi kelulusan mahasiswa yang akurat dapat membantu pihak Program Studi dalam mengambil keputusan-keputusan yang tepat untuk mencegah ketidaktepatan waktu kelulusan mahasiswanya. Pada penelitian ini, artificial neural network algoritma backpropagation diimplementasikan pada aplikasi yang dibuat dengan output lulus terlambat dan lulus tepat waktu. Penelitian ini menggunakan sebanyak 318 sampel data yang mana 70 % data digunakan sebagai data training dan 30 % data digunakan sebagai data testing. Dari hasil perhitungan tabel confusion matrix diperoleh persentase akurasi prediksi sebesar 98.97 %.</em></p><p><em></em><em><strong>Kata kunci</strong>: kelulusan mahasiswa, artificial neural network, backpropagation, confusion matrix</em></p>

Rancang Bangun Sistem Pakar Prediksi Stres Belajar Dengan Neural Network Algoritma Backpropagation

2016 ◽  
Vol 7 (2) ◽  
pp. 105-112
Author(s):  
Adhi Kusnadi ◽  
Idul Putra
Keyword(s):  
Neural Network ◽  
Expert System ◽  
Stress Level ◽  
Network Structure ◽  
Human Being ◽  
Accuracy Rate ◽  
Backpropagation Algorithm ◽  
Stress Prediction ◽  
Index Terms ◽  
Hidden Nodes

Stress will definitely be experienced by every human being and the level of stress experienced by each individual is different. Stress experienced by students certainly will disturb their study if it is not handled quickly and appropriately. Therefore we have created an expert system using a neural network backpropagation algorithm to help counselors to predict the stress level of students. The network structure of the experiment consists of 26 input nodes, 5 hidden nodes, and 2 the output nodes, learning rate of 0.1, momentum of 0.1, and epoch of 5000, with a 100% accuracy rate. Index Terms - Stress on study, expert system, neural network, Stress Prediction

The Convolution Neural Network Combined with the HT Person Fit Statistic to Develop an APP for Detecting Dengue Fever in Children: Development and Usability Study (Preprint)

2019 ◽  
Author(s):  
CHIEN WEI ◽  
Chi Chow Julie ◽  
Chou Willy
Keyword(s):  
Neural Network ◽  
Dengue Fever ◽  
Prediction Accuracy ◽  
Early Stage ◽  
Family Members ◽  
Convolution Neural Network ◽  
Public Health Issue ◽  
Person Fit ◽  
Model Accuracy ◽  
Comparison Results

UNSTRUCTURED Backgrounds: Dengue fever (DF) is an important public health issue in Asia. However, the disease is extremely hard to detect using traditional dichotomous (i.e., absent vs. present) evaluations of symptoms. Convolution neural network (CNN), a well-established deep learning method, can improve prediction accuracy on account of its usage of a large number of parameters for modeling. Whether the HT person fit statistic can be combined with CNN to increase the prediction accuracy of the model and develop an application (APP) to detect DF in children remains unknown. Objectives: The aim of this study is to build a model for the automatic detection and classification of DF with symptoms to help patients, family members, and clinicians identify the disease at an early stage. Methods: We extracted 19 feature variables of DF-related symptoms from 177 pediatric patients (69 diagnosed with DF) using CNN to predict DF risk. The accuracy of two sets of characteristics (19 symptoms and four other variables, including person mean, standard deviation, and two HT-related statistics matched to DF+ and DF−) for predicting DF, were then compared. Data were separated into training and testing sets, and the former was used to predict the latter. We calculated the sensitivity (Sens), specificity (Spec), and area under the receiver operating characteristic curve (AUC) across studies for comparison. Results: We observed that (1) the 23-item model yields a higher accuracy rate (0.95) and AUC (0.94) than the 19-item model (accuracy = 0.92, AUC = 0.90) based on the 177-case training set; (2) the Sens values are almost higher than the corresponding Spec values (90% in 10 scenarios) for predicting DF; (3) the Sens and Spec values of the 23-item model are consistently higher than those of the 19-item model. An APP was subsequently designed to detect DF in children. Conclusion: The 23-item model yielded higher accuracy rates (0.95) and AUC (0.94) than the 19-item model (accuracy = 0.92, AUC = 0.90). An APP could be developed to help patients, family members, and clinicians discriminate DF from other febrile illnesses at an early stage.

Sign in / Sign up

Export Citation Format

Share Document