자료유형  

 학위논문

Control Number  

0017164885

International Standard Book Number  

9798346395614

Dewey Decimal Classification Number  

571.6

Main Entry-Personal Name  

Wu, Kevin Eric.

Publication, Distribution, etc. (Imprint  

[S.l.] : Stanford University., 2024

Publication, Distribution, etc. (Imprint  

Ann Arbor : ProQuest Dissertations & Theses, 2024

Physical Description  

165 p.

General Note  

Source: Dissertations Abstracts International, Volume: 86-05, Section: B.

General Note  

Advisor: Howard , Howard;Zou, James.

Dissertation Note  

Thesis (Ph.D.)--Stanford University, 2024.

Summary, Etc.  

요약Over the past few years, the field of machine learning has undergone a renaissance where state-of-the-art models have made huge advances in their ability to perform a wide range of tasks. For example, large language models like Llama [1] can produce text virtually indistinguishable from that written by a human, generative models can draw coherent, striking images of varying artistic styles given just a few words of prompting [2], and self- driving cars have progressed from a science fiction pipe dream to working prototypes that autonomously shuttle people through city streets. While these applications have already drawn much attention due to their ability to mimic human interactions, behaviors, and cre- ativity, they have only begun to scratch the surface of how machine learning could advance humanity. Namely, if we consider the aforementioned examples, we'll notice that these models have been primarily targeted at mimicking and replicating tasks that are within the realm of human proficiency. This raises the question: what if we instead trained these powerful models to perform tasks that we, as humans, cannotcurrently do or understand?What could machines learn about datasets that are opaque to humans, and what can they then teach us in turn? This body of work is dedicated to exploring the feasibility and implications of this question specifically through the lens of machine learning as applied to biological sciences. Biology is an incredibly exciting field for machine learning as it is critical to understanding and advancing human health, but it is also harbors deep nuance that has only begun to come to light as scientists gather more and more data describing these complex systems. This implies that if machine learning can be applied successfully to biological datasets, it could unlock whole new domains of knowledge previously inaccessible to humans. Imagine if we could teach a computer to read the human genome -- the genetic information that encodes all the pieces and functions necessary for life -- much as it can learn to read human text. Just as we can ask a chatbot questions about the world, one could imagine similarly probing that genetic model for insights on what parts of the genome may be misbehaving to cause certain diseases -- something that even world-class doctors and scientists struggle to understand consistently today.The remainder of this work explores the usage and impact of machine learning in biolog- ical sciences through a series of studies focused on several key biological data modalities. In chapters 2-4, I explore the development of machine learning models for understanding how RNA behaves within human cells, and more importantly, how these models can be interrogated to reveal scientific insights into biological mechanisms, and how these models can be used to generate scientific hypotheses in the context of real epidemiological challenges. I conclude my work on RNA by synthesizing an overview of structural modeling methods for RNA. In chapter 5, I explore a model for translating single-cell datasets and how it could be used to improve how we understand patient tissue samples and disease.

Subject Added Entry-Topical Term  

Endoplasmic reticulum.

Subject Added Entry-Topical Term  

Gene expression.

Subject Added Entry-Topical Term  

Deep learning.

Subject Added Entry-Topical Term  

Ribonucleic acid--RNA.

Subject Added Entry-Topical Term  

Severe acute respiratory syndrome coronavirus 2.

Subject Added Entry-Topical Term  

Binding sites.

Subject Added Entry-Topical Term  

Neural networks.

Subject Added Entry-Topical Term  

Biology.

Subject Added Entry-Topical Term  

Bioinformatics.

Subject Added Entry-Topical Term  

Cellular biology.

Subject Added Entry-Topical Term  

Genetics.

Subject Added Entry-Topical Term  

Virology.

Added Entry-Corporate Name  

Stanford University.

Host Item Entry  

Dissertations Abstracts International. 86-05B.

Electronic Location and Access  

로그인을 한후 보실 수 있는 자료입니다.

Control Number  

joongbu:656187