國立虎尾科技大學 |

Automated Machine Learning for Intelligent Systems.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Automated Machine Learning for Intelligent Systems./
Author:	Li, Zelong.
Description:	1 online resource (146 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 85-11, Section: B.
Contained By:	Dissertations Abstracts International85-11B.
Subject:	Computer science. -
Online resource:	click for full text (PQDT)
ISBN:	9798382758145

Automated Machine Learning for Intelligent Systems.
Li, Zelong.

Automated Machine Learning for Intelligent Systems. - 1 online resource (146 pages)

Source: Dissertations Abstracts International, Volume: 85-11, Section: B.

Thesis (Ph.D.)--Rutgers The State University of New Jersey, School of Graduate Studies, 2024.

Includes bibliographical references

The fast progress in Machine Learning techniques has led to a growing impact of Artificial Intelligence (AI) on various aspects of people's lives. AI model learning comprises three vital components: data inputs, model design, and loss functions. Each of these components makes a significant contribution to the AI system's performance. Traditionally, these components needed skilled domain experts to meticulously design them, creating a challenging barrier to entry into AI. Besides, manual approaches are relatively inefficient and often fail to achieve optimal results. Recently, automated machine learning (AutoML), such as neural architecture search (NAS), has tried to tackle the challenge by automating the design and parameters of deep models. However, traditional AutoML research mainly focuses on automated model design instead of the inputs and loss functions, and AutoML's research on recent large language models (LLMs) is still in its infancy due to the enormous computations required. Consequently, this thesis aims at answering the following questions:• How can we design AutoML techniques for the other two essential parts of AI systems, i.e., inputs and loss functions?• What is the role of AutoML under the background of the rapid development of large-scale models such LLMs? How can we leverage the insights of AutoML to improve LLMs?Given these questions, this thesis introduces three research works. • We propose AutoLossGen, which can generate the most effective loss function using basic mathematical operations for various recommendation tasks and models. The generated loss gives better recommendation performance than commonly used baseline losses and is transferable to another model or dataset on the same recommendation task.• As for the data input aspect, we create the Personalized Automatic Prompt for Recommendation Language Model (PAP-REC) framework. This framework aims to teach language models the optimal automatic input instructions, i.e., prompts, for recommendation tasks. We develop surrogate metrics and leverage an alternative updating schedule to address the inflated search space derived from personalized prompts. Our PAP-REC framework manages to generate personalized prompts, and the automatically generated prompts outperform manually constructed prompts.• For LLM-based agents, our Formal-LLM framework utilizes formal language through automaton to automate the workflow of LLM-based agents when solving complex tasks under constraints. Our framework enables the agents to automatically adhere to the planning policy with automata guidance, resulting in more controllable LLM-based agents.In these three studies, through the development of automated machine learning techniques at various stages of an AI pipeline, we further advance both small-scale and large-scale intelligent systems, making them more precise, effective and user-friendly.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798382758145Subjects--Topical Terms:

573171
Computer science.
Subjects--Index Terms:

Automated machine learningIndex Terms--Genre/Form:

554714
Electronic books.

Automated Machine Learning for Intelligent Systems.
LDR:04328ntm a22003977 4500 001 1144984
005 20240617111734.5
006 m o d
007 cr mn ---uuuuu
008 250605s2024 xx obm 000 0 eng d
020 $a 9798382758145
035 $a (MiAaPQ)AAI31143120
035 $a AAI31143120
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Li, Zelong. $3 1470179
245 1 0 $a Automated Machine Learning for Intelligent Systems.
264 0 $c 2024
300 $a 1 online resource (146 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: 85-11, Section: B.
500 $a Advisor: Zhang, Yongfeng.
502 $a Thesis (Ph.D.)--Rutgers The State University of New Jersey, School of Graduate Studies, 2024.
504 $a Includes bibliographical references
520 $a The fast progress in Machine Learning techniques has led to a growing impact of Artificial Intelligence (AI) on various aspects of people's lives. AI model learning comprises three vital components: data inputs, model design, and loss functions. Each of these components makes a significant contribution to the AI system's performance. Traditionally, these components needed skilled domain experts to meticulously design them, creating a challenging barrier to entry into AI. Besides, manual approaches are relatively inefficient and often fail to achieve optimal results. Recently, automated machine learning (AutoML), such as neural architecture search (NAS), has tried to tackle the challenge by automating the design and parameters of deep models. However, traditional AutoML research mainly focuses on automated model design instead of the inputs and loss functions, and AutoML's research on recent large language models (LLMs) is still in its infancy due to the enormous computations required. Consequently, this thesis aims at answering the following questions:• How can we design AutoML techniques for the other two essential parts of AI systems, i.e., inputs and loss functions?• What is the role of AutoML under the background of the rapid development of large-scale models such LLMs? How can we leverage the insights of AutoML to improve LLMs?Given these questions, this thesis introduces three research works. • We propose AutoLossGen, which can generate the most effective loss function using basic mathematical operations for various recommendation tasks and models. The generated loss gives better recommendation performance than commonly used baseline losses and is transferable to another model or dataset on the same recommendation task.• As for the data input aspect, we create the Personalized Automatic Prompt for Recommendation Language Model (PAP-REC) framework. This framework aims to teach language models the optimal automatic input instructions, i.e., prompts, for recommendation tasks. We develop surrogate metrics and leverage an alternative updating schedule to address the inflated search space derived from personalized prompts. Our PAP-REC framework manages to generate personalized prompts, and the automatically generated prompts outperform manually constructed prompts.• For LLM-based agents, our Formal-LLM framework utilizes formal language through automaton to automate the workflow of LLM-based agents when solving complex tasks under constraints. Our framework enables the agents to automatically adhere to the planning policy with automata guidance, resulting in more controllable LLM-based agents.In these three studies, through the development of automated machine learning techniques at various stages of an AI pipeline, we further advance both small-scale and large-scale intelligent systems, making them more precise, effective and user-friendly.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Computer science. $3 573171
650 4 $a Information technology. $3 559429
653 $a Automated machine learning
653 $a Intelligent systems
653 $a Large language models
653 $a Data inputs
653 $a Neural architecture search
655 7 $a Electronic books. $2 local $3 554714
690 $a 0984
690 $a 0800
690 $a 0489
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a Rutgers The State University of New Jersey, School of Graduate Studies. $b Computer Science. $3 1241263
773 0 $t Dissertations Abstracts International $g 85-11B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31143120 $z click for full text (PQDT)