| Autor: |
Tamrakar, Poi, Pathak, Abha, Thorat, Pallavi, Lal, Mily, Goel, Akanksha, Bhende, Manisha, Sharma, Swati |
| Předmět: |
|
| Zdroj: |
AIP Conference Proceedings; 2024, Vol. 3139 Issue 1, p1-7, 7p |
| Abstrakt: |
In recent years, deep learning has become an invaluable resource for chemists, notably in the areas of chemical reaction prediction and synthesis design. This abstract gives a quick summary of how deep learning is used in certain contexts. Predicting the result of chemical reactions is a difficult undertaking that requires much research. Training deep learning models on massive databases of known responses allows them to discover patterns and correlations. Examples of such models are graph neural networks, convolutional neural networks, and recurrent neural networks. Some of the things these models can do is anticipate key products, reaction processes, and yields by using the structural and contextual information of reactants. The goal of synthesis planning is to create a reaction plan that will successfully synthesize a desired molecule. To predict possible reactions for a given target molecule, deep learning algorithms study and learn from large reaction databases. These models think about things like whether or not a reaction can happen, under what circumstances, and with what reagents. Deep learning models may speed up the process of inventing novel chemical synthesis by proposing efficient and practical synthesis pathways using reinforcement learning or optimization approaches. Despite its potential, deep learning faces a number of obstacles and restrictions. The interpretability of model predictions, the ability to generalize to novel reaction types, the integration of safety and feasibility limitations, and the availability of extensive and trustworthy reaction datasets all play a role. Despite these obstacles, deep learning is set to transform chemical reaction prediction and synthesis planning thanks to continuous research and breakthroughs. It is possible for scientists to leverage the power of AI by combining deep learning with curated databases and domain knowledge to speed up the discovery of novel reactions and aid in the creation of more efficient and sustainable synthesis pathways. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
