EAGER: Creating an Unsupervised Interpretable Representation of the World Through Concept Disentanglement
EAGER:通过概念解开创建一个无监督的、可解释的世界表征
基本信息
- 批准号:2130250
- 负责人:
- 金额:$ 16.93万
- 依托单位:
- 依托单位国家:美国
- 项目类别:Standard Grant
- 财政年份:2021
- 资助国家:美国
- 起止时间:2021-11-01 至 2024-10-31
- 项目状态:已结题
- 来源:
- 关键词:
项目摘要
Humans are able to break down a large entity into smaller and simpler concepts, just from having seen many objects and their relationships. Reproducing this type of behavior in a machine learning model has several benefits. In particular, it could lead to computational ways of representing the world that are interpretable yet powerful. These new representations could be used within machine learning algorithms, allowing the algorithms to be more robust and more likely to generalize when the underlying situations change. For instance, if an algorithm has found a collection of parts that an object is typically comprised of, then it can use those parts to identify this type of object even when it is in an unusual setting, or when the object itself is unusual. This new way of representing the world will allow more robust and generalizable machine learning models. This will be particularly helpful for difficult challenges in computer vision, including problems related to vision systems in automated vehicles, analysis of medical time-series, and materials science problems related to the understanding of material properties and discovery of new materials.Specifically, the main goal of this project is learning with interpretable learned concepts using a disentangled neural network. The approach breaks the problem down into three steps that each could be manageable, and each step can be checked and improved independently of the other steps. The steps are to decompose each observation into local parts, identify possible concepts by looking at common relationships between the local parts, and align the proposed concepts, based on their semantic meaning, within a disentangled neural network. The discovered concepts will be interpretable and can be used as features for many downstream tasks. The disentangled neural networks built from these concepts could potentially generalize more easily to new situations than other approaches.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.
人类能够将一个大的实体分解成更小更简单的概念,仅仅是因为看到了许多物体和它们之间的关系。在机器学习模型中重现这种行为有几个好处。特别是,它可能会导致可解释但功能强大的表示世界的计算方法。这些新的表示可以在机器学习算法中使用,使算法更加鲁棒,并且在底层情况发生变化时更有可能泛化。例如,如果一个算法发现了一个对象通常由部分组成的集合,那么它可以使用这些部分来识别这种类型的对象,即使它处于不寻常的设置中,或者对象本身是不寻常的。这种表示世界的新方式将使机器学习模型更加健壮和通用。这将特别有助于解决计算机视觉中的困难挑战,包括与自动驾驶车辆的视觉系统相关的问题,医疗时间序列的分析,以及与理解材料特性和发现新材料相关的材料科学问题。具体来说,这个项目的主要目标是使用一个解纠缠的神经网络来学习可解释的学习概念。该方法将问题分解为三个步骤,每个步骤都可以管理,并且每个步骤都可以独立于其他步骤进行检查和改进。步骤是将每个观察结果分解为局部部分,通过查看局部部分之间的共同关系来识别可能的概念,并在解缠的神经网络中根据其语义对齐所提出的概念。发现的概念将是可解释的,并且可以用作许多下游任务的特征。与其他方法相比,从这些概念构建的解纠缠神经网络可能更容易泛化到新情况。该奖项反映了美国国家科学基金会的法定使命,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(6)
专著数量(0)
科研奖励数量(0)
会议论文数量(0)
专利数量(0)
This Looks Like Those: Illuminating Prototypical Concepts Using Multiple Visualizations
- DOI:10.48550/arxiv.2310.18589
- 发表时间:2023-10
- 期刊:
- 影响因子:0
- 作者:Chiyu Ma;Brandon Zhao;Chaofan Chen;Cynthia Rudin
- 通讯作者:Chiyu Ma;Brandon Zhao;Chaofan Chen;Cynthia Rudin
OKRidge: Scalable Optimal k-Sparse Ridge Regression
OKRidge:可扩展的最优 k-稀疏岭回归
- DOI:
- 发表时间:2023
- 期刊:
- 影响因子:0
- 作者:Liu, Jiachang Liu;Rosen, Sam;Zhong, Chudi;Rudin, Cynthia
- 通讯作者:Rudin, Cynthia
A Path to Simpler Models Starts With Noise
通往更简单模型的道路从噪声开始
- DOI:
- 发表时间:2023
- 期刊:
- 影响因子:0
- 作者:Semenova, Lesia;Chen, Harry;Parr, Ronald;Rudin, Cynthia
- 通讯作者:Rudin, Cynthia
Exploring and Interacting with the Set of Good Sparse Generalized Additive Models
探索一组良好的稀疏广义可加模型并与之交互
- DOI:
- 发表时间:2023
- 期刊:
- 影响因子:0
- 作者:Zhong, Chudi;Chen, Zhi;Liu, Jiachang;Seltzer, Margo;Rudin, Cynthia
- 通讯作者:Rudin, Cynthia
The Rashomon Importance Distribution: Getting RID of Unstable, Single Model-based Variable Importance
- DOI:10.48550/arxiv.2309.13775
- 发表时间:2023-09
- 期刊:
- 影响因子:0
- 作者:J. Donnelly;Srikar Katta;C. Rudin;E. Browne
- 通讯作者:J. Donnelly;Srikar Katta;C. Rudin;E. Browne
{{
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 }}
Cynthia Rudin其他文献
Fast and Interpretable Mortality Risk Scores for Critical Care Patients
重症监护患者快速且可解释的死亡风险评分
- DOI:
- 发表时间:
2023 - 期刊:
- 影响因子:0
- 作者:
Chloe Qinyu Zhu;Muhang Tian;Lesia Semenova;Jiachang Liu;Jack Xu;Joseph Scarpa;Cynthia Rudin - 通讯作者:
Cynthia Rudin
Exploring the Whole Rashomon Set of Sparse Decision Trees
探索整个罗生门稀疏决策树集
- DOI:
- 发表时间:
2022 - 期刊:
- 影响因子:0
- 作者:
Rui Xin;Chudi Zhong;Zhi Chen;Takuya Takagi;Margo Seltzer;Cynthia Rudin - 通讯作者:
Cynthia Rudin
Graph-based design of irregular metamaterials
基于图的不规则超材料设计
- DOI:
10.1016/j.ijmecsci.2025.110203 - 发表时间:
2025-06-01 - 期刊:
- 影响因子:9.400
- 作者:
Rayehe Karimi Mahabadi;Zhi Chen;Alexander C. Ogren;Han Zhang;Chiara Daraio;Cynthia Rudin;L. Catherine Brinson - 通讯作者:
L. Catherine Brinson
Understanding and Exploring the Whole Set of Good Sparse Generalized Additive Models
理解和探索一整套良好的稀疏广义可加模型
- DOI:
- 发表时间:
2023 - 期刊:
- 影响因子:0
- 作者:
Zhi Chen;Chudi Zhong;Margo I. Seltzer;Cynthia Rudin - 通讯作者:
Cynthia Rudin
Machine learning for science and society
- DOI:
10.1007/s10994-013-5425-9 - 发表时间:
2013-11-28 - 期刊:
- 影响因子:2.900
- 作者:
Cynthia Rudin;Kiri L. Wagstaff - 通讯作者:
Kiri L. Wagstaff
Cynthia Rudin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('Cynthia Rudin', 18)}}的其他基金
FAI: An Interpretable AI Framework for Care of Critically Ill Patients Involving Matching and Decision Trees
FAI:用于危重患者护理的可解释人工智能框架,涉及匹配和决策树
- 批准号:
2147061 - 财政年份:2022
- 资助金额:
$ 16.93万 - 项目类别:
Standard Grant
FW-HTF-R: Interpretable Machine Learning for Human-Machine Collaboration in High Stakes Decisions in Mammography
FW-HTF-R:用于乳腺 X 线摄影高风险决策中人机协作的可解释机器学习
- 批准号:
2222336 - 财政年份:2022
- 资助金额:
$ 16.93万 - 项目类别:
Standard Grant
NSF Workshop on Seamless/Seamful Human-Technology Interaction
NSF 无缝/无缝人类技术交互研讨会
- 批准号:
2131355 - 财政年份:2021
- 资助金额:
$ 16.93万 - 项目类别:
Standard Grant
CAREER: New Approaches for Ranking in Machine Learning
职业:机器学习排名的新方法
- 批准号:
1658794 - 财政年份:2016
- 资助金额:
$ 16.93万 - 项目类别:
Continuing Grant
CAREER: New Approaches for Ranking in Machine Learning
职业:机器学习排名的新方法
- 批准号:
1053407 - 财政年份:2011
- 资助金额:
$ 16.93万 - 项目类别:
Continuing Grant
Postdoctoral Research Fellowship in Biological Informatics for FY 2005
2005财年生物信息学博士后研究奖学金
- 批准号:
0434636 - 财政年份:2005
- 资助金额:
$ 16.93万 - 项目类别:
Fellowship Award
相似海外基金
Creating pH-sensitive self-healing concrete using sludge waste for sewers
利用下水道污泥废物制造 pH 敏感的自修复混凝土
- 批准号:
DP230100688 - 财政年份:2024
- 资助金额:
$ 16.93万 - 项目类别:
Discovery Projects
Observing, Creating and Addressing Topological Spin Textures in a Monolayer XY Magnet
观察、创建和解决单层 XY 磁体中的拓扑自旋纹理
- 批准号:
EP/Y023250/1 - 财政年份:2024
- 资助金额:
$ 16.93万 - 项目类别:
Research Grant
Creating a Path to Achieving Success and Sense of Belonging in Computer Science
创造一条在计算机科学领域取得成功和归属感的道路
- 批准号:
2322665 - 财政年份:2024
- 资助金额:
$ 16.93万 - 项目类别:
Standard Grant
NSF Engines Development Award: Creating climate-resilient opportunities for plant systems (NC)
NSF 发动机开发奖:为工厂系统创造气候适应机会 (NC)
- 批准号:
2315399 - 财政年份:2024
- 资助金额:
$ 16.93万 - 项目类别:
Cooperative Agreement
EAGER: Creating a Composite EL Nino Record from the Lowland Neotropics
EAGER:创造低地新热带区综合厄尔尼诺记录
- 批准号:
2417794 - 财政年份:2024
- 资助金额:
$ 16.93万 - 项目类别:
Standard Grant
Creating cosmopolitan rural communities through Japanese crafts: a comparative perspective with Portugal and Brazil
通过日本手工艺创建国际化农村社区:与葡萄牙和巴西的比较视角
- 批准号:
24K21000 - 财政年份:2024
- 资助金额:
$ 16.93万 - 项目类别:
Grant-in-Aid for Early-Career Scientists
Creating a reflective, assessment workbook for University teachers to enhance teaching techniques and improve student engagement, by incorporating International Baccalaureate (IB) teaching practices
通过纳入国际文凭 (IB) 教学实践,为大学教师创建反思性评估工作簿,以提高教学技巧并提高学生参与度
- 批准号:
24K06129 - 财政年份:2024
- 资助金额:
$ 16.93万 - 项目类别:
Grant-in-Aid for Scientific Research (C)
CAREER: Creating Tough, Sustainable Materials Using Fracture Size-Effects and Architecture
职业:利用断裂尺寸效应和架构创造坚韧、可持续的材料
- 批准号:
2339197 - 财政年份:2024
- 资助金额:
$ 16.93万 - 项目类别:
Standard Grant
Creating a Grow-Your-Own Program for Recruiting and Supporting Computer Science Teacher Candidates in Rural Georgia
创建一个自己成长的计划,用于招募和支持佐治亚州农村地区的计算机科学教师候选人
- 批准号:
2344678 - 财政年份:2024
- 资助金额:
$ 16.93万 - 项目类别:
Standard Grant
Creating a Teen Science Cafe Movement
创建青少年科学咖啡馆运动
- 批准号:
2424451 - 财政年份:2024
- 资助金额:
$ 16.93万 - 项目类别:
Continuing Grant














{{item.name}}会员




