![](https://img-blog.csdnimg.cn/0623a187def1413388ff9ba249f55905.png?x-oss-process=image/resize,m_fixed,h_224,w_224)
物理驱动深度学习论文分享
文章平均质量分 89
主要分享物理驱动深度学习相关论文
pinn山里娃
交流请加vx:worker_harder
展开
-
DEEPXDE
近年来在深度学习在计算机视觉,自然语言等方面得到了广泛应用,尽管在这些和相关领域取得了显著的成功,但深度学习还没有广泛应用于科学计算领域,然而,最近,通过深度学习求解偏微分方程(PDEs),如标准微分形式或积分形式,已经成为科学机器学习下一个潜在的新的子领域;特别的,用一种近似于PDE解的神经网络来代替传统的数值离散方法原创 2020-08-27 11:10:32 · 6881 阅读 · 4 评论 -
hp-VPINNs: Variational Physics-Informed Neural Networks With Domain Decomposition论文笔记
hp-VPINNs: Variational Physics-Informed Neural Networks With Domain原创 2020-08-27 11:21:33 · 631 阅读 · 3 评论 -
Physics-informed neural networks: A deep learning framework论文笔记
论文信息题目:Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations期刊会议:Journal of Computational Physics年份:18论文地址:论文链接代码:代码链接基础补充Runge-KuttaRunge-Kutta法是用于非线原创 2020-08-27 19:11:21 · 1912 阅读 · 0 评论 -
The neural particle method – An updated Lagrangian physics informed neural network for computational
论文信息题目:XX作者及单位:XX期刊、会议:XX时间:xx论文地址:论文链接代码:代码链接基础摘要论文动机Main contributions:Related Work问题背景与定义本文方法Relevant literature数值实验Neural Network Approximation Theorem for PDEs总结:...原创 2020-09-30 14:42:45 · 295 阅读 · 0 评论 -
Physics-informed neural networks for high-speed flows
论文信息题目:作者及单位:期刊、会议:时间:论文地址:论文链接代码:代码链接基础摘要论文动机Main contributions:Related Work问题背景与定义本文方法Relevant literature数值实验Neural Network Approximation Theorem for PDEs总结:...2020-09-30 14:43:16 · 768 阅读 · 1 评论 -
Neural Network Technique in Some Inverse Problems of Mathematical Physics_Vladimir
提出了两种方法求PDE逆问题, 一种是利用Morozov’s condition求解PDE方程中的参数,一种是结合RProp and the cloud methods原创 2020-12-29 14:47:36 · 1007 阅读 · 0 评论 -
Physics-informed neural networks for inverse problems in nano-optics and metamaterials
论文信息题目:hysics-informed neural networks for inverse problems in nano-optics and metamaterials作者:Yuyao Chen, Lu Lu, George Em Karniadakis, and Luca Dal Negro期刊会议:Computational Physics年份:19论文地址:代码:内容动机动机:在强多光散射条件下,复杂多粒子几何中物理驱动的光散射微分模型的逆问题成为一个本质上不定的原创 2020-12-18 10:46:58 · 1951 阅读 · 7 评论 -
Using Hybrid Physics-Informed Neural Networks for Digital Twins in Prognosis and Health Management
Using Hybrid Physics-Informed Neural Networks for Digital Twins in Prognosis and Health ManagementMotivationSimulations are pervasive in every domain of science and engineering, but they often have constraints such as large computational times, limited原创 2022-01-13 09:05:07 · 291 阅读 · 0 评论 -
Physics-informed neural network for ultrasound nondestructive quantification of surface breaking
论文信息题目:Physics-informed neural network for ultrasound nondestructive quantification of surface breaking cracks作者及单位:期刊、会议:时间:论文地址:论文链接代码:代码链接基础摘要论文动机Main contributions:Related Work问题背景与定义本文方法Relevant literature数值实验Neural Network Approximati原创 2020-09-23 15:42:17 · 824 阅读 · 6 评论 -
Transfer learning based multi-fidelity physics informed deep neural network论文笔记
论文信息题目:Transfer learning based multi-fidelity physics informed deep neural network作者:Souvik Chakraborty期刊会议: Machine Learning (cs.LG); Computational Physics (physics.comp-ph)年份:2020论文地址:代码:基础补充内容动机动机:基于某些假设和近似,一些控制方程可以得到,但是建立的这个模型属于low-fidelity原创 2020-08-26 14:59:04 · 716 阅读 · 0 评论 -
SMBO:Sequential Model-Based Optimization for General Algorithm Configuration
目录文章目录题目The basic definitionThe motivationPrevious methods and advantages and disadvantagesThe proposed methodPaper detailed introductioncomment题目SMBO:Sequential Model-Based Optimization forGener...原创 2020-03-29 23:18:12 · 817 阅读 · 2 评论 -
Towards Evaluating the Robustness of Neural Networks详解
关于JSMA中选择像素对来进行修改的方法,作者做出了不一样的解释:假设softmax层最小的输入为-100,那么softmax层的对于该输入的输出为0,即使增加了10,由-100变成了-90,仍然为0,显然这对输出没有什么影响。而如果softmax层最大的输入为10,将其更改为了0,显然这会使得输出产生巨大的变化。而JSMA的攻击方式并不会考虑从0到10和从-100到-90的不同,他是一视同仁的...原创 2020-09-01 21:30:26 · 800 阅读 · 0 评论 -
Sequential Classification-Based Optimization for Direct Policy Search
论文信息题目:Sequential Classification-BasedOptimization for Direct Policy Search期刊会议:the Thirty-First AAAI Conference on Artificial Intelligence年份:17论文地址:论文链接代码:内容动机策略优化任务通常非常复杂,涉及许多局部优化,在这种情况下,无导...原创 2020-09-01 21:29:20 · 566 阅读 · 0 评论 -
Physics-informed neural networks for inverse problems in nano-optics and metamaterials论文笔记
论文信息题目:Physics-informed neural networks for inverse problems in nano-optics and metamaterials作者:Yuyao Chen, Lu Lu, George Em Karniadakis, and Luca DPal Negro期刊会议:Computational Physics年份:19论文地址:代码:内容动机动机:- 在强多光散射条件下,复杂多粒子几何中物理驱动的光散射微分模型的逆问题成为一个本质上不原创 2020-09-01 21:25:04 · 555 阅读 · 0 评论 -
Towards Gradient Free and Projection Free Stochastic Optimization
论文信息题目:Towards Gradient Free and Projection Free Stochastic Optimization期刊会议:AISTATS 2019年份:18代码:内容动机随机性优化问题:minx∈Cf(x)=minx∈CEy∼P[F(x;y)]\min _{\mathbf{x} \in \mathcal{C}} f(\mathbf{x})=\mi...原创 2020-09-01 21:29:57 · 179 阅读 · 0 评论 -
A physics-aware learning architecture with input transfer networks for predictive modeling
论文信息题目:A physics-aware learning architecture with input transfer networksfor predictive modeling作者及单位:Amir Behjat a,∗, Chen Zeng a, Rahul Rai a, Ion Matei b, David Doermann a,Souma Chowdhury a,∗a University at Buffalo, Buffalo, NY, 14260, USAb Palo原创 2020-12-18 10:45:52 · 199 阅读 · 0 评论 -
Physics-informed neural networks for high-speed flows
论文信息题目:Physics-informed neural networks for high-speed flows作者:Zhiping Mao, Ameya D. Jagtap, George Em Karniadakis∗Division of Applied Mathematics, 182 George Street, Brown University, Providence, RI 02912, USA期刊会议:年份:19论文地址:代码:摘要在这项工作中,原创 2020-10-21 16:47:00 · 3088 阅读 · 8 评论 -
Physics-informed semantic inpainting: Application to geostatistical modeling
论文信息题目:Physics-informed semantic inpainting: Application to geostatistical modeling作者:Qiang Zhenga,b, Lingzao Zenga,∗, Zhendan Caoc, George Em Karniadakisb,d,a Institute of Soil and Water Resource and Environmental Science, Zhejiang Provincial Key Labo原创 2020-10-16 10:17:33 · 908 阅读 · 5 评论 -
Prevention is Better than Cure: Handling Basis Collapse and Transparency in Dense Networks
论文信息题目作者会议:单位:论文动机本文方法问题描述本文方法方法对比实验结果总结:下步工作原创 2020-12-18 10:46:04 · 320 阅读 · 0 评论 -
Distributed physics informed neural network for data-efficient solution to PDE
论文信息题目:Distributed physics informed neural network for data-efficient solution to partial differential equations作者:Vikas Dwivedia,∗, Nishant Parasharb, Balaji Srinivasana期刊会议:年份:论文地址:代码:基础补充内容动机动机:传统it is a well-known fact that mesh generation原创 2020-10-14 15:29:01 · 1055 阅读 · 1 评论 -
Understanding and mitigating gradient pathologies in physics-informed neural networks.
论文信息题目:Understanding and mitigating gradient pathologies in physics-informed neural networks.作者及单位:Sifan Wang(Graduate Group in Applied Mathematics and Computational Science University of Pennsylvania Philadelphia)Yujun Teng(Department of Mechanichal E原创 2020-12-18 10:45:20 · 1153 阅读 · 0 评论 -
Online Meta-Learning详解
论文信息题目:Online Meta-Learning作者:Chelsea Finn * 1 Aravind Rajeswaran * 2 Sham Kakade 2 Sergey Levine 11 University of California, Berkeley2University of Washington. Correspondence to: Chelsea Finn cbfinn@cs.stanford.edu, Aravind Rajeswaran aravraj@cs原创 2020-10-28 16:43:18 · 1363 阅读 · 1 评论 -
DGM: A deep learning algorithm for solving partial differential equations
论文信息题目:DGM: A deep learning algorithm for solving partial differential equations作者:Justin Sirignano∗ and Konstantinos Spiliopoulos期刊、会议:单位:代码:基础论文动机High-dimensional partial differential equations (PDEs) are used in physics, engineering, and financ原创 2020-12-18 10:46:20 · 2594 阅读 · 0 评论 -
Learning and Meta-Learning of Stochastic Advection-Diffusion-Reaction Systems from Sparse Measuremen
论文信息题目:Learning and Meta-Learning of Stochastic Advection-Diffusion-Reaction Systems from Sparse Measurements(对流扩散响应)作者:Xiaoli Chen(a,b), Jinqiao Duan©, George Em Karniadakis(b,d)期刊、会议:computational physics单位:a Center for Mathematical Sciences & S原创 2020-12-18 10:46:44 · 584 阅读 · 0 评论 -
Rapid learning or feature reuse ? Toward understanding the effectiveness of MAML
论文信息题目:Rapid learning or feature reuse ? Toward understanding the effectiveness of MAML作者及单位:Aniruddh Raghu,MITMaithra Raghu,Cornell University & Google BrainSamy Bengio,Google BrainOriol Vinyals,DeepMind期刊、会议:ICLR时间:2020论文地址:论文链接原创 2020-12-18 10:45:36 · 1517 阅读 · 5 评论 -
极限学习机(Extreme Learning Machine,ELM)详解
极限学习机的提出目的是为了解决人工神经网络训练时耗费的时间和高成本。对此,2004年由南洋理工大学的黄广斌提出了ELM学习理论。原创 2021-12-23 10:13:25 · 8820 阅读 · 2 评论 -
Modified physics-informed neural network method based on the conservation law constraint
Modified physics-informed neural network method based on theconservation law constraint and its prediction of optical solitons作者:Gang-Zhou Wu, Yin Fang, Yue-Yue Wang and Chao-Qing Dai地址:2021arxiv文章目录Modified physics-informed neural network method bas原创 2021-09-04 11:44:27 · 877 阅读 · 0 评论 -
Local extreme learning machines and domain decomposition for solving linear and nonlinear PDE论文阅读笔记
Local extreme learning machines and domain decomposition for solving linear and nonlinear partial diSuchuan Dong, Zongwei LibCenter for Computational and Applied Mathematics, Department ofMathematics, Purdue University West Lafayette, IN, USAComput. Me原创 2021-12-23 20:11:22 · 301 阅读 · 2 评论 -
Physics-informed Neural Network for Nonlinear Dynamics in Fiber Optics论文笔记
Physics-informed Neural Network for Nonlinear Dynamics in Fiber Optics作者:Xiaotian Jiang1, Danshi Wang1,∗, Qirui Fan2, Min Zhang1, Chao Lu3, and Alan Pak Tao Lau2时间:2021期刊:arixiv摘要研究PINN求解用于学习光纤非线性动力学的Schrödinger方程,对多种物理效应进行深入研究,包括dispersion, self-原创 2021-09-07 22:36:48 · 1058 阅读 · 2 评论 -
A Physics-Data-Driven Bayesian Method for Heat Conduction Problems
A Physics-Data-Driven Bayesian Method for Heat Conduction ProblemsXinchao Jiang*, Hu Wang†, Yu liState2021arxivFirst author. E-mail address: jiangxinchao@hnu.edu.cn (X.C. Jiang) and Corresponding author. E-mail address: wanghu@hnu.edu.cn (H. Wang)摘原创 2021-09-08 11:05:45 · 1042 阅读 · 4 评论 -
Physics-informed dynamic mode decomposition
In this work, we demonstrate how physical principles – such as symmetries, invariances, and conservation laws – can be integrated into the dynamic mode decomposition (DMD). - However, **DMD frequently produces models that are sensitive to noise, fail to g原创 2021-12-20 17:24:52 · 1886 阅读 · 0 评论 -
Learn bifurcations of nonlinear parametric systems via equation-driven neural networks
we develop a new machine learning approach to compute the bifurcations via so-called equation-driven neural networks (EDNNs). The EDNNs consist of a two-step optimization: the first step is to approximate the solution function of the parameter by training原创 2021-12-21 11:46:41 · 178 阅读 · 0 评论 -
A novel meta-learning initialization method for physics-informed neural networks
Physics-informed neural networks (PINN)可以看作是通用的PDE求解器,但是,PINN存在两点缺陷:1.在特定问题上训练效率慢;2.真实样本少的情况下,即使迭代多次,仍得不到好的结果。为了克服PINN的两个缺陷,在这篇报告中,拓展了PINN,将元学习的Reptile算法应用于PINN中,针对两类先验信息,提出了两种算法。在我们的方法中,通过利用先验信息,经过预训练后,能够实现在少量真实样本甚至没有真实样本的新的问题上更快收敛。最后,在一维和二维的泊松方程的算例上演示原创 2022-01-13 09:22:44 · 831 阅读 · 3 评论 -
Self-supervised learning method to solve the inverse problem of Fokker-Planck Equation
Fokker-Planckequation方程在很多应用中是未知的,使得直接求解PDE这样的正问题充满挑战,通过监督学习从观测数据中恢复FPE亟待解决创新点提出FPE-net,直接将FPE-netterm作为网络权重,能够从观测数据中恢复FPEterm,而且对于数据降噪也有效果。...原创 2022-07-22 10:06:05 · 562 阅读 · 0 评论 -
Extended Physics-InformedNeural Networks论文详解
提出了更灵活分解域的XPINN方法,比cPINN区域分解更灵活,而且使用与所有方程。原创 2022-07-22 10:09:17 · 1433 阅读 · 0 评论 -
Systems biology informed deep learning for inferring parameters and hidden dynamics
利用网络训练拟合模型参数,提出了system-biology -informed 神经算法,能够可靠而准确的推断hidden dynamics,能够用很少的观测数据对dynamics和模型参数进行推断。原创 2022-10-28 10:55:23 · 382 阅读 · 0 评论 -
重型机床热误差补偿温度敏感测点的识别与优化选择
传感器测点优化原创 2022-10-28 10:30:33 · 1096 阅读 · 0 评论 -
Extended Physics-InformedNeural Networks (XPINNs)
cPINN是通过区域分解,每个区域使用小的网络进行训练,使得求解时不同区域能够并行计算。原创 2022-10-28 11:00:47 · 918 阅读 · 0 评论 -
PhyGeoNet一种可用于不规则区域的物理信息极限学习机
主要就是解决了CNN求解域为非规则形状这样问题,同时将物理信息嵌入CNN中,实现了物理数据双驱动。原创 2022-10-28 10:50:36 · 1190 阅读 · 0 评论 -
Solving Inverse Problems With Deep Neural Networks
本篇论文对深度神经网络解决欠定(underdetermined)的逆问题的鲁棒性(robustness)进行了研究。研究内容包括用Guassian measurement的压感(compressed sening)以及通过Founier and Radon measurement的图像恢复原创 2022-10-28 10:57:29 · 1361 阅读 · 0 评论