Review of deep learning: concepts, CNN architectures, challenges, applications, future directions

Journal Of Big Data, Journal Year: 2021, Volume and Issue: 8(1)

Published: March 31, 2021

In the last few years, deep learning (DL) computing paradigm has been deemed Gold Standard in machine (ML) community. Moreover, it gradually become most widely used computational approach field of ML, thus achieving outstanding results on several complex cognitive tasks, matching or even beating those provided by human performance. One benefits DL is ability to learn massive amounts data. The grown fast years and extensively successfully address a wide range traditional applications. More importantly, outperformed well-known ML techniques many domains, e.g., cybersecurity, natural language processing, bioinformatics, robotics control, medical information among others. Despite contributed works reviewing State-of-the-Art DL, all them only tackled one aspect which leads an overall lack knowledge about it. Therefore, this contribution, we propose using more holistic order provide suitable starting point from develop full understanding DL. Specifically, review attempts comprehensive survey important aspects including enhancements recently added field. particular, paper outlines importance presents types networks. It then convolutional neural networks (CNNs) utilized network type describes development CNNs architectures together with their main features, AlexNet closing High-Resolution (HR.Net). Finally, further present challenges suggested solutions help researchers understand existing research gaps. followed list major Computational tools FPGA, GPU, CPU are summarized along description influence ends evolution matrix, benchmark datasets, summary conclusion.

Language: Английский

---

Applications of machine learning in drug discovery and development

Nature Reviews Drug Discovery, Journal Year: 2019, Volume and Issue: 18(6), P. 463 - 477

Published: April 11, 2019

Language: Английский

---

Scientific Machine Learning Through Physics–Informed Neural Networks: Where we are and What’s Next

Journal of Scientific Computing, Journal Year: 2022, Volume and Issue: 92(3)

Published: July 26, 2022

Physics-Informed Neural Networks (PINN) are neural networks (NNs) that encode model equations, like Partial Differential Equations (PDE), as a component of the network itself. PINNs nowadays used to solve PDEs, fractional integral-differential and stochastic PDEs. This novel methodology has arisen multi-task learning framework in which NN must fit observed data while reducing PDE residual. article provides comprehensive review literature on PINNs: primary goal study was characterize these their related advantages disadvantages. The also attempts incorporate publications broader range collocation-based physics informed networks, stars form vanilla PINN, well many other variants, such physics-constrained (PCNN), variational hp-VPINN, conservative PINN (CPINN). indicates most research focused customizing through different activation functions, gradient optimization techniques, structures, loss function structures. Despite wide applications for have been used, by demonstrating ability be more feasible some contexts than classical numerical techniques Finite Element Method (FEM), advancements still possible, notably theoretical issues remain unresolved.

Language: Английский

---

Artificial intelligence to deep learning: machine intelligence approach for drug discovery

Molecular Diversity, Journal Year: 2021, Volume and Issue: 25(3), P. 1315 - 1360

Published: April 12, 2021

Drug designing and development is an important area of research for pharmaceutical companies chemical scientists. However, low efficacy, off-target delivery, time consumption, high cost impose a hurdle challenges that impact drug design discovery. Further, complex big data from genomics, proteomics, microarray data, clinical trials also obstacle in the discovery pipeline. Artificial intelligence machine learning technology play crucial role development. In other words, artificial neural networks deep algorithms have modernized area. Machine been implemented several processes such as peptide synthesis, structure-based virtual screening, ligand-based toxicity prediction, monitoring release, pharmacophore modeling, quantitative structure-activity relationship, repositioning, polypharmacology, physiochemical activity. Evidence past strengthens implementation this field. Moreover, novel mining, curation, management techniques provided critical support to recently developed modeling algorithms. summary, advancements provide excellent opportunity rational process, which will eventually mankind. The primary concern associated with consumption production cost. inefficiency, inaccurate target inappropriate dosage are hurdles inhibit process delivery With technology, computer-aided integrating can eliminate traditional referred superset comprising learning, whereas comprises supervised unsupervised reinforcement learning. subset has extensively network, vector machines, classification regression, generative adversarial networks, symbolic meta-learning examples applied process. different areas synthesis molecule design, screening molecular docking, relationship protein misfolding protein-protein interactions, pathway identification polypharmacology. principles active inactive, pre-clinical development, secondary biomarker manufacturing, bioactivity properties, prediction toxicity, mode action.

Language: Английский

---

Machine learning in chemoinformatics and drug discovery

Drug Discovery Today, Journal Year: 2018, Volume and Issue: 23(8), P. 1538 - 1546

Published: May 8, 2018

• Chemical graph theory and descriptors in drug discovery. fingerprint similarity analysis. Machine learning models for virtual screening. Future challenges direction machine-learning-based Chemoinformatics is an established discipline focusing on extracting, processing extrapolating meaningful data from chemical structures. With the rapid explosion of 'big' HTS combinatorial synthesis, machine has become indispensable tool designers to mine information large compound databases design drugs with important biological properties. To process data, we first reviewed multiple layers chemoinformatics pipeline followed by introduction commonly used discovery QSAR Here, present basic principles recent case studies demonstrate utility techniques analyses; discuss limitations future directions guide further development this evolving field.

Language: Английский

---

Concepts of Artificial Intelligence for Computer-Assisted Drug Discovery

Chemical Reviews, Journal Year: 2019, Volume and Issue: 119(18), P. 10520 - 10594

Published: July 11, 2019

Artificial intelligence (AI), and, in particular, deep learning as a subcategory of AI, provides opportunities for the discovery and development innovative drugs. Various machine approaches have recently (re)emerged, some which may be considered instances domain-specific AI been successfully employed drug design. This review comprehensive portrayal these techniques their applications medicinal chemistry. After introducing basic principles, alongside application notes, various algorithms, current state-of-the art AI-assisted pharmaceutical is discussed, including structure- ligand-based virtual screening, de novo design, physicochemical pharmacokinetic property prediction, repurposing, related aspects. Finally, several challenges limitations methods are summarized, with view to potential future directions

Language: Английский

---

Artificial intelligence in drug development: present status and future prospects

Drug Discovery Today, Journal Year: 2018, Volume and Issue: 24(3), P. 773 - 780

Published: Nov. 23, 2018

Language: Английский

---

Expanding the medicinal chemistry synthetic toolbox

Nature Reviews Drug Discovery, Journal Year: 2018, Volume and Issue: 17(10), P. 709 - 727

Published: Aug. 24, 2018

Language: Английский

---

Prediction of higher-selectivity catalysts by computer-driven workflow and machine learning

Science, Journal Year: 2019, Volume and Issue: 363(6424)

Published: Jan. 18, 2019

Catalyst design in asymmetric reaction development has traditionally been driven by empiricism, wherein experimentalists attempt to qualitatively recognize structural patterns improve selectivity. Machine learning algorithms and chemoinformatics can potentially accelerate this process recognizing otherwise inscrutable large datasets. Herein we report a computationally guided workflow for chiral catalyst selection using at every stage of development. Robust molecular descriptors that are agnostic the scaffold allow universal training set on basis steric electronic properties. This be used train machine methods make highly accurate predictive models over broad range selectivity space. Using support vector machines deep feed-forward neural networks, demonstrate modeling phosphoric acid-catalyzed thiol addition N-acylimines.

Language: Английский

---

Deep Learning in Chemistry

Journal of Chemical Information and Modeling, Journal Year: 2019, Volume and Issue: 59(6), P. 2545 - 2559

Published: June 13, 2019

Machine learning enables computers to address problems by from data. Deep is a type of machine that uses hierarchical recombination features extract pertinent information and then learn the patterns represented in Over last eight years, its abilities have increasingly been applied wide variety chemical challenges, improving computational chemistry drug materials design even synthesis planning. This review aims explain concepts deep chemists any background follows this with an overview diverse applications demonstrated literature. We hope will empower broader community engage burgeoning field foster growing movement accelerated chemistry.

Language: Английский

---