東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁 到查詢結果 [ null ]

切換: 標籤 | MARC模式 | ISBD

Improving Machine Learning Framework...

Musielewicz, Joseph,

FindBook

Google Book

Amazon

博客來

Improving Machine Learning Frameworks for Catalyst Simulations /

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Improving Machine Learning Frameworks for Catalyst Simulations // Joseph Musielewicz.
作者:	Musielewicz, Joseph,
面頁冊數:	1 electronic resource (145 pages)
附註:	Source: Dissertations Abstracts International, Volume: 86-08, Section: B.
Contained By:	Dissertations Abstracts International86-08B.
標題:	Computational chemistry. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31840906
ISBN:	9798304993647

Improving Machine Learning Frameworks for Catalyst Simulations /
Musielewicz, Joseph,

Improving Machine Learning Frameworks for Catalyst Simulations /Joseph Musielewicz. - 1 electronic resource (145 pages)

Source: Dissertations Abstracts International, Volume: 86-08, Section: B.

The looming threat of climate change is driving an urgent need to redesign industrial processes worldwide with new and more efficient catalysts. Machine learning methods have the potential to massively accelerate the discovery of novel catalysts, which are extremely costly and therefore slow to discover by traditional methods. Quantum mechanical simulators like Density Functional Theory (DFT) allow computational simulations to be used to discover new catalysts at great computational expense, which prevents large scale DFT efforts from meeting this demand. Graph Neural Networks (GNNs) have emerged as promising machine learned surrogates for DFT, with the capacity to achieve high accuracy when learning from DFT generated data and improve upon computational costs by up to O(105). However, GNNs can suffer from challenges with consistency, performance, and generalization, and require improvements to be sufficient for the task at hand.This dissertation explores the development of improved machine learning frameworks for catalyst discovery in the areas of trustability, efficiency, and adaptability. First I present Finetuna, an open source package for accelerating catalyst simulations and efficiently acquiring new data with active learning and fine tuning. Second I demonstrate the successes and pitfalls of using Finetuna as a transfer learning strategy to adapt pre-trained GNNs to out-of-distribution structures. Third I benchmark and develop novel improvements to uncertainty quantification methods for GNNs to intelligently identify when to trust the model and when it must be adapted. Finally I explore the usefulness of GNNs for determining the potential energy Hessian, an extremely high value task which is related to, but not explicitly represented in, its training data set.

English

ISBN: 9798304993647Subjects--Topical Terms:

3350019
Computational chemistry.
Subjects--Index Terms:

Active learning

Improving Machine Learning Frameworks for Catalyst Simulations /
LDR:03200nmm a22004213i 4500 001 2435649
005 20251009122859.5
006 m o d
007 cr|nu||||||||
008 261011s2025 miu||||||m |||||||eng d
020 $a 9798304993647
035 $a (MiAaPQD)AAI31840906
035 $a AAI31840906
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Musielewicz, Joseph, $e author. $0 (orcid)0000-0002-5296-9177 $3 3826022
245 1 0 $a Improving Machine Learning Frameworks for Catalyst Simulations / $c Joseph Musielewicz.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2025
300 $a 1 electronic resource (145 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 86-08, Section: B.
500 $a Advisors: Kitchin, John R. Committee members: Ulissi, Zachary; Laird, Carl; Schneider, Jeff.
502 $b Ph.D. $c Carnegie Mellon University $d 2025.
520 $a The looming threat of climate change is driving an urgent need to redesign industrial processes worldwide with new and more efficient catalysts. Machine learning methods have the potential to massively accelerate the discovery of novel catalysts, which are extremely costly and therefore slow to discover by traditional methods. Quantum mechanical simulators like Density Functional Theory (DFT) allow computational simulations to be used to discover new catalysts at great computational expense, which prevents large scale DFT efforts from meeting this demand. Graph Neural Networks (GNNs) have emerged as promising machine learned surrogates for DFT, with the capacity to achieve high accuracy when learning from DFT generated data and improve upon computational costs by up to O(105). However, GNNs can suffer from challenges with consistency, performance, and generalization, and require improvements to be sufficient for the task at hand.This dissertation explores the development of improved machine learning frameworks for catalyst discovery in the areas of trustability, efficiency, and adaptability. First I present Finetuna, an open source package for accelerating catalyst simulations and efficiently acquiring new data with active learning and fine tuning. Second I demonstrate the successes and pitfalls of using Finetuna as a transfer learning strategy to adapt pre-trained GNNs to out-of-distribution structures. Third I benchmark and develop novel improvements to uncertainty quantification methods for GNNs to intelligently identify when to trust the model and when it must be adapted. Finally I explore the usefulness of GNNs for determining the potential energy Hessian, an extremely high value task which is related to, but not explicitly represented in, its training data set.
546 $a English
590 $a School code: 0041
650 4 $a Computational chemistry. $3 3350019
650 4 $a Chemical engineering. $3 560457
653 $a Active learning
653 $a Catalysis
653 $a Graph neural networks
653 $a Transfer learning
690 $a 0800
690 $a 0219
690 $a 0542
710 2 $a Carnegie Mellon University. $b Chemical Engineering. $e degree granting institution. $3 3817251
720 1 $a Kitchin, John R. $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 86-08B.
790 $a 0041
791 $a Ph.D.
792 $a 2025
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31840906