Geometrization Approaches toward Understanding Deep Learning

理解深度学习的几何化方法

基本信息

  • 批准号:
    2310679
  • 负责人:
  • 金额:
    $ 27.5万
  • 依托单位:
  • 依托单位国家:
    美国
  • 项目类别:
    Standard Grant
  • 财政年份:
    2023
  • 资助国家:
    美国
  • 起止时间:
    2023-09-01 至 2026-08-31
  • 项目状态:
    未结题

项目摘要

The project will study the theoretical underpinnings of deep learning, a widely successful approach to various data-extensive applications. A comprehensive understanding of deep learning is crucial for the development of principled design and training of deep learning models, ultimately reducing computational burden and human costs. The research will span three foundational and complementary directions: understanding well-trained deep neural networks, examining deep learning training dynamics, and exploring data processing in interior layers of deep learning models. By bridging the gap between the complex training paradigms of modern neural networks and existing theories, the project will demystify these black-box models, making them more interpretable and efficient for a wide range of scientific and engineering applications. The project will offer multiple interdisciplinary opportunities for boosting the professional development of the next generation of statisticians and data scientists. The research activities will focus on three research projects to develop a comprehensive understanding of deep learning from a statistical and mathematical perspective. The first project will analyze symmetric geometries in the final stages of deep learning training, developing novel optimization techniques and statistical insights. The second project will provide a detailed understanding of the dynamics of modern deep learning training, integrating statistical and inferential ideas into the active field of deep learning dynamics. The third project will investigate how deep learning separates data according to class membership across all layers of the neural network, using techniques from random matrix theory, non-convex optimization, and learning theory. The successful completion of these research projects will result in a geometrization of deep learning methodologies.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
该项目将研究深度学习的理论基础,深度学习是一种广泛成功的方法,用于各种数据广泛的应用。对深度学习的全面理解对于开发深度学习模型的原则性设计和训练至关重要,最终减少计算负担和人力成本。该研究将涵盖三个基础和互补的方向:理解训练有素的深度神经网络,研究深度学习训练动态,以及探索深度学习模型内部层的数据处理。通过弥合现代神经网络的复杂训练范式与现有理论之间的差距,该项目将揭开这些黑箱模型的神秘面纱,使其在广泛的科学和工程应用中更具可解释性和效率。该项目将提供多个跨学科的机会,以促进下一代统计学家和数据科学家的专业发展。研究活动将集中在三个研究项目上,从统计和数学的角度全面了解深度学习。第一个项目将在深度学习训练的最后阶段分析对称几何,开发新的优化技术和统计见解。第二个项目将详细了解现代深度学习训练的动态,将统计和推理思想融入深度学习动态的活跃领域。第三个项目将研究深度学习如何使用随机矩阵理论、非凸优化和学习理论的技术,根据神经网络所有层的类成员关系来分离数据。该奖项反映了NSF的法定使命,并通过使用基金会的知识价值和更广泛的影响审查标准进行评估,被认为值得支持。

项目成果

期刊论文数量(0)
专著数量(0)
科研奖励数量(0)
会议论文数量(0)
专利数量(0)

数据更新时间:{{ journalArticles.updateTime }}

{{ item.title }}
{{ item.translation_title }}
  • DOI:
    {{ item.doi }}
  • 发表时间:
    {{ item.publish_year }}
  • 期刊:
  • 影响因子:
    {{ item.factor }}
  • 作者:
    {{ item.authors }}
  • 通讯作者:
    {{ item.author }}

数据更新时间:{{ journalArticles.updateTime }}

{{ item.title }}
  • 作者:
    {{ item.author }}

数据更新时间:{{ monograph.updateTime }}

{{ item.title }}
  • 作者:
    {{ item.author }}

数据更新时间:{{ sciAawards.updateTime }}

{{ item.title }}
  • 作者:
    {{ item.author }}

数据更新时间:{{ conferencePapers.updateTime }}

{{ item.title }}
  • 作者:
    {{ item.author }}

数据更新时间:{{ patent.updateTime }}

Weijie Su其他文献

RNF123 inhibits cell viability, cell cycle and colony formation of breast cancer by inhibiting glycolysis via ubiquitination of PFKP
Objective Breast Volume, Shape and Surface Area Assessment: A Systematic Review of Breast Measurement Methods
  • DOI:
    10.1007/s00266-014-0412-5
  • 发表时间:
    2014-10-23
  • 期刊:
  • 影响因子:
    2.800
  • 作者:
    Wenjing Xi;Aurelia Trisliana Perdanasari;Yeesiang Ong;Sheng Han;Peiru Min;Weijie Su;Shaoqing Feng;Lucrezia Pacchioni;Yi Xin Zhang;Davide Lazzeri
  • 通讯作者:
    Davide Lazzeri
Ghost in the Minecraft: Generally Capable Agents for Open-World Environments via Large Language Models with Text-based Knowledge and Memory
《我的世界》中的幽灵:通过具有基于文本的知识和记忆的大型语言模型,为开放世界环境提供通用代理
  • DOI:
    10.48550/arxiv.2305.17144
  • 发表时间:
    2023
  • 期刊:
  • 影响因子:
    0
  • 作者:
    Xizhou Zhu;Yuntao Chen;Hao Tian;Chenxin Tao;Weijie Su;Chenyu Yang;Gao Huang;Bin Li;Lewei Lu;Xiaogang Wang;Y. Qiao;Zhaoxiang Zhang;Jifeng Dai
  • 通讯作者:
    Jifeng Dai
You Are the Best Reviewer of Your Own Papers: An Owner-Assisted Scoring Mechanism
Endogenous Electric Field‐Coupled PD@BP Biomimetic Periosteum Promotes Bone Regeneration through Sensory Nerve via Fanconi Anemia Signaling Pathway
内源电场耦合 PD@BP 仿生骨膜通过范可尼贫血信号通路通过感觉神经促进骨再生
  • DOI:
    10.1002/adhm.202203027
  • 发表时间:
    2023-01
  • 期刊:
  • 影响因子:
    10
  • 作者:
    Yanlin Su;Lian Zeng;Rongli Deng;Bing Ye;Shuo Tang;Zekang Xiong;Tingfang Sun;Qiuyue Ding;Weijie Su;Xirui Jing;Qing Gao;Xiumei Wang;Zhiye Qiu;Kaifang Chen;Daping Quan;Xiaodong Guo
  • 通讯作者:
    Xiaodong Guo

Weijie Su的其他文献

{{ item.title }}
{{ item.translation_title }}
  • DOI:
    {{ item.doi }}
  • 发表时间:
    {{ item.publish_year }}
  • 期刊:
  • 影响因子:
    {{ item.factor }}
  • 作者:
    {{ item.authors }}
  • 通讯作者:
    {{ item.author }}

{{ truncateString('Weijie Su', 18)}}的其他基金

CAREER: A Statistical Inferential Framework for Online Learning Algorithms
职业:在线学习算法的统计推理框架
  • 批准号:
    1847415
  • 财政年份:
    2019
  • 资助金额:
    $ 27.5万
  • 项目类别:
    Continuing Grant

相似国自然基金

Lagrangian origin of geometric approaches to scattering amplitudes
  • 批准号:
    24ZR1450600
  • 批准年份:
    2024
  • 资助金额:
    0.0 万元
  • 项目类别:
    省市级项目

相似海外基金

Approaches toward External Validity of Randomized Controlled Trials in the Social Sciences
社会科学中随机对照试验的外部有效性方法
  • 批准号:
    2318659
  • 财政年份:
    2023
  • 资助金额:
    $ 27.5万
  • 项目类别:
    Standard Grant
Chemoenzymatic Protein Semisynthesis Approaches Toward Cell Signaling Enzymes
细胞信号酶的化学酶蛋白半合成方法
  • 批准号:
    10713633
  • 财政年份:
    2023
  • 资助金额:
    $ 27.5万
  • 项目类别:
Revisiting Wallach's Rule: Approaches toward singular point interplaying molecular symmetries and electronic properties
重温瓦拉赫法则:研究奇点相互作用的分子对称性和电子特性的方法
  • 批准号:
    22H00314
  • 财政年份:
    2022
  • 资助金额:
    $ 27.5万
  • 项目类别:
    Grant-in-Aid for Scientific Research (A)
Converging biomechanical and biophysical approaches toward ligament and tendon regeneration
融合生物力学和生物物理方法实现韧带和肌腱再生
  • 批准号:
    RGPIN-2022-04233
  • 财政年份:
    2022
  • 资助金额:
    $ 27.5万
  • 项目类别:
    Discovery Grants Program - Individual
Converging biomechanical and biophysical approaches toward ligament and tendon regeneration
融合生物力学和生物物理方法实现韧带和肌腱再生
  • 批准号:
    DGECR-2022-00204
  • 财政年份:
    2022
  • 资助金额:
    $ 27.5万
  • 项目类别:
    Discovery Launch Supplement
Toward a deeper understanding of allostery and allotargeting by computational approaches
通过计算方法更深入地理解变构和异体靶向
  • 批准号:
    10462594
  • 财政年份:
    2021
  • 资助金额:
    $ 27.5万
  • 项目类别:
Toward a deeper understanding of allostery and allotargeting by computational approaches
通过计算方法更深入地理解变构和异体靶向
  • 批准号:
    10231654
  • 财政年份:
    2021
  • 资助金额:
    $ 27.5万
  • 项目类别:
Energy-Efficient and Sustainable Approaches toward Greening the Next Generation of Wireless Communication Systems
实现下一代无线通信系统绿色化的节能且可持续的方法
  • 批准号:
    RGPIN-2016-05243
  • 财政年份:
    2021
  • 资助金额:
    $ 27.5万
  • 项目类别:
    Discovery Grants Program - Individual
Bioinformatic and molecular approaches toward molecular surveillance of insecticide resistance in African Aedes aegypti
非洲埃及伊蚊杀虫剂抗性分子监测的生物信息和分子方法
  • 批准号:
    2665223
  • 财政年份:
    2021
  • 资助金额:
    $ 27.5万
  • 项目类别:
    Studentship
Toward a deeper understanding of allostery and allotargeting by computational approaches
通过计算方法更深入地理解变构和异体靶向
  • 批准号:
    10887238
  • 财政年份:
    2021
  • 资助金额:
    $ 27.5万
  • 项目类别:
{{ showInfoDetail.title }}

作者:{{ showInfoDetail.author }}

知道了