Steve Brunton
Steve Brunton
  • 486
  • 21 621 706
Residual Networks (ResNet) [Physics Informed Machine Learning]
This video discusses Residual Networks, one of the most popular machine learning architectures that has enabled considerably deeper neural networks through jump/skip connections. This architecture mimics many of the aspects of a numerical integrator.
This video was produced at the University of Washington, and we acknowledge funding support from the Boeing Company
%%% CHAPTERS %%%
00:00 Intro
01:09 Concept: Modeling the Residual
03:26 Building Blocks
05:59 Motivation: Deep Network Signal Loss
07:43 Extending to Classification
09:00 Extending to DiffEqs
10:16 Impact of CVPR and Resnet
12:17 Resnets and Euler Integrators
13:34 Neural ODEs and Improved Integrators
16:07 Outro
zhlédnutí: 28 702

Video

Neural ODEs (NODEs) [Physics Informed Machine Learning]24:37Neural ODEs (NODEs) [Physics Informed Machine Learning]
Neural ODEs (NODEs) [Physics Informed Machine Learning]
zhlédnutí 46KPřed 14 dny
This video describes Neural ODEs, a powerful machine learning approach to learn ODEs from data. This video was produced at the University of Washington, and we acknowledge funding support from the Boeing Company %%% CHAPTERS %%% 00:00 Intro 02:09 Background: ResNet 05:05 From ResNet to ODE 07:59 ODE Essential Insight/ Why ODE outperforms ResNet // 09:05 ODE Essential Insight Rephrase 1 // 09:54...
Physics Informed Neural Networks (PINNs) [Physics Informed Machine Learning]34:32Physics Informed Neural Networks (PINNs) [Physics Informed Machine Learning]
Physics Informed Neural Networks (PINNs) [Physics Informed Machine Learning]
zhlédnutí 39KPřed 21 dnem
This video introduces PINNs, or Physics Informed Neural Networks. PINNs are a simple modification of a neural network that adds a PDE in the loss function to promote solutions that satisfy known physics. For example, if we wish to model a fluid flow field and we know it is incompressible, we can add the divergence of the field in the loss function to drive it towards zero. This approach relies ...
SINDy-RL: Interpretable and Efficient Model-Based Reinforcement Learning21:28SINDy-RL: Interpretable and Efficient Model-Based Reinforcement Learning
SINDy-RL: Interpretable and Efficient Model-Based Reinforcement Learning
zhlédnutí 14KPřed měsícem
SINDy-RL: Interpretable and Efficient Model-Based Reinforcement Learning by Nicholas Zolman, Urban Fasel, J. Nathan Kutz, Steven L. Brunton arxiv paper: arxiv.org/abs/2403.09110 github code: github.com/nzolman/sindy-rl Deep reinforcement learning (DRL) has shown significant promise for uncovering sophisticated control policies that interact in environments with complicated dynamics, such as sta...
AI/ML+Physics: Preview of Upcoming Modules and Bootcamps [Physics Informed Machine Learning]19:01AI/ML+Physics: Preview of Upcoming Modules and Bootcamps [Physics Informed Machine Learning]
AI/ML+Physics: Preview of Upcoming Modules and Bootcamps [Physics Informed Machine Learning]
zhlédnutí 16KPřed měsícem
This video provides a brief preview of the upcoming modules and bootcamps in this series on Physics Informed Machine Learning. Topics include: (1) Parsimonious modeling and SINDy; (2) Physics informed neural networks (PINNs); (3) Operator methods, like DeepONets and Fourier Neural Operators; (4) Symmetries in physics and machine learning; (5) Digital Twin technology; and (6) Case studies in eng...
AI/ML+Physics: Recap and Summary [Physics Informed Machine Learning]24:01AI/ML+Physics: Recap and Summary [Physics Informed Machine Learning]
AI/ML+Physics: Recap and Summary [Physics Informed Machine Learning]
zhlédnutí 14KPřed měsícem
This video provides a brief recap of this introductory series on Physics Informed Machine Learning. We revisit the five stages of machine learning, and how physics may be incorporated into these stages. We also discuss architectures, symmetries, the digital twin, applications in engineering, and the importance of dynamical systems and controls benchmarks. This video was produced at the Universi...
Using sparse trajectory data to find Lagrangian Coherent Structures (LCS) in fluid flows20:26Using sparse trajectory data to find Lagrangian Coherent Structures (LCS) in fluid flows
Using sparse trajectory data to find Lagrangian Coherent Structures (LCS) in fluid flows
zhlédnutí 8KPřed 2 měsíci
Video by Tanner Harms, based on "Lagrangian Gradient Regression for the Detection of Coherent Structures from Sparse Trajectory Data" by Tanner D. Harms, Steven L. Brunton, Beverley J. McKeon arxiv.org/abs/2310.10994 The method of Lagrangian Coherent Structures (LCS) uses particle trajectories in fluid flows to identify coherent structures that govern the behavior of the flow. The typical metho...
AI/ML+Physics Part 5: Employing an Optimization Algorithm [Physics Informed Machine Learning]32:35AI/ML+Physics Part 5: Employing an Optimization Algorithm [Physics Informed Machine Learning]
AI/ML+Physics Part 5: Employing an Optimization Algorithm [Physics Informed Machine Learning]
zhlédnutí 14KPřed 2 měsíci
This video discusses the fifth stage of the machine learning process: (5) selecting and implementing an optimization algorithm to train the model. There are opportunities to incorporate physics into this stage of the process, such as using constrained optimization to force a model onto a susbpace or submanifold characterized by a symmetry or other physical constraint. This video was produced at...
The Future of Model Based Engineering: Collimator 2.012:22The Future of Model Based Engineering: Collimator 2.0
The Future of Model Based Engineering: Collimator 2.0
zhlédnutí 18KPřed 2 měsíci
Learn more at www.collimator.ai/ Collimator allows you to model, simulate, optimize, control, and collaborate in the cloud, with the power of Python and JAX New features: * Powered by JAX * Generative AI * Auto-Differentiation * PID Auto-Tune * SINDy model blocks * Model Predictive Control * Real-Time Collaboration * Hardware in the Loop * Hybrid models * State machines * FMU support * Updated ...
AI/ML+Physics Part 4: Crafting a Loss Function [Physics Informed Machine Learning]23:12AI/ML+Physics Part 4: Crafting a Loss Function [Physics Informed Machine Learning]
AI/ML+Physics Part 4: Crafting a Loss Function [Physics Informed Machine Learning]
zhlédnutí 27KPřed 3 měsíci
This video discusses the fourth stage of the machine learning process: (4) designing a loss function to assess the performance of the model. There are opportunities to incorporate physics into this stage of the process, such as adding regularization terms to promote sparsity or extra loss functions to ensure that a partial differential equation is satisfied, as in PINNs. This video was produced...
AI/ML+Physics Part 3: Designing an Architecture [Physics Informed Machine Learning]36:30AI/ML+Physics Part 3: Designing an Architecture [Physics Informed Machine Learning]
AI/ML+Physics Part 3: Designing an Architecture [Physics Informed Machine Learning]
zhlédnutí 34KPřed 3 měsíci
This video discusses the third stage of the machine learning process: (3) choosing an architecture with which to represent the model. This is one of the most exciting stages, including all of the new architectures, such as UNets, ResNets, SINDy, PINNs, Operator networks, and many more. There are opportunities to incorporate physics into this stage of the process, such as incorporating known sym...
AI/ML+Physics Part 2: Curating Training Data [Physics Informed Machine Learning]36:11AI/ML+Physics Part 2: Curating Training Data [Physics Informed Machine Learning]
AI/ML+Physics Part 2: Curating Training Data [Physics Informed Machine Learning]
zhlédnutí 26KPřed 3 měsíci
This video discusses the second stage of the machine learning process: (2) collecting and curating training data to inform the model. There are opportunities to incorporate physics into this stage of the process, such as data augmentation to incorporate known symmetries. This video was produced at the University of Washington, and we acknowledge funding support from the Boeing Company %%% CHAPT...
AI/ML+Physics Part 1: Choosing what to model [Physics Informed Machine Learning]43:27AI/ML+Physics Part 1: Choosing what to model [Physics Informed Machine Learning]
AI/ML+Physics Part 1: Choosing what to model [Physics Informed Machine Learning]
zhlédnutí 67KPřed 4 měsíci
This video discusses the first stage of the machine learning process: (1) formulating a problem to model. There are lots of opportunities to incorporate physics into this process, and learn new physics by applying ML to the right problem. This video was produced at the University of Washington, and we acknowledge funding support from the Boeing Company %%% CHAPTERS %%% 00:00 Intro 04:51 Decidin...
Physics Informed Machine Learning: High Level Overview of AI and ML in Science and Engineering47:27Physics Informed Machine Learning: High Level Overview of AI and ML in Science and Engineering
Physics Informed Machine Learning: High Level Overview of AI and ML in Science and Engineering
zhlédnutí 206KPřed 4 měsíci
This video describes how to incorporate physics into the machine learning process. The process of machine learning is broken down into five stages: (1) formulating a problem to model, (2) collecting and curating training data to inform the model, (3) choosing an architecture with which to represent the model, (4) designing a loss function to assess the performance of the model, and (5) selectin...
Can we make commercial aircraft faster? Mitigating transonic buffet with porous trailing edges16:10Can we make commercial aircraft faster? Mitigating transonic buffet with porous trailing edges
Can we make commercial aircraft faster? Mitigating transonic buffet with porous trailing edges
zhlédnutí 9KPřed 4 měsíci
Can we make commercial aircraft faster? Mitigating transonic buffet with porous trailing edges
A Neural Network Primer19:14A Neural Network Primer
A Neural Network Primer
zhlédnutí 35KPřed 5 měsíci
A Neural Network Primer
Supervised & Unsupervised Machine Learning11:46Supervised & Unsupervised Machine Learning
Supervised & Unsupervised Machine Learning
zhlédnutí 23KPřed 5 měsíci
Supervised & Unsupervised Machine Learning
A Machine Learning Primer: How to Build an ML Model13:48A Machine Learning Primer: How to Build an ML Model
A Machine Learning Primer: How to Build an ML Model
zhlédnutí 43KPřed 5 měsíci
A Machine Learning Primer: How to Build an ML Model
Arousal as a universal embedding for spatiotemporal brain dynamics25:23Arousal as a universal embedding for spatiotemporal brain dynamics
Arousal as a universal embedding for spatiotemporal brain dynamics
zhlédnutí 25KPřed 6 měsíci
Arousal as a universal embedding for spatiotemporal brain dynamics
New Advances in Artificial Intelligence and Machine Learning28:02New Advances in Artificial Intelligence and Machine Learning
New Advances in Artificial Intelligence and Machine Learning
zhlédnutí 72KPřed 6 měsíci
New Advances in Artificial Intelligence and Machine Learning
Nonlinear parametric models of viscoelastic fluid flows with SINDy14:51Nonlinear parametric models of viscoelastic fluid flows with SINDy
Nonlinear parametric models of viscoelastic fluid flows with SINDy
zhlédnutí 6KPřed 6 měsíci
Nonlinear parametric models of viscoelastic fluid flows with SINDy
[5/8] Control for Societal-Scale Challenges: Road Map 2030 [Technology, Validation, and Transition]14:16[5/8] Control for Societal-Scale Challenges: Road Map 2030 [Technology, Validation, and Transition]
[5/8] Control for Societal-Scale Challenges: Road Map 2030 [Technology, Validation, and Transition]
zhlédnutí 10KPřed 11 měsíci
[5/8] Control for Societal-Scale Challenges: Road Map 2030 [Technology, Validation, and Transition]
[8/8] Control for Societal-Scale Challenges: Road Map 2030 [Recommendations]10:19[8/8] Control for Societal-Scale Challenges: Road Map 2030 [Recommendations]
[8/8] Control for Societal-Scale Challenges: Road Map 2030 [Recommendations]
zhlédnutí 6KPřed 11 měsíci
[8/8] Control for Societal-Scale Challenges: Road Map 2030 [Recommendations]
[2/8] Control for Societal-Scale Challenges: Road Map 2030 [Societal Drivers]11:39[2/8] Control for Societal-Scale Challenges: Road Map 2030 [Societal Drivers]
[2/8] Control for Societal-Scale Challenges: Road Map 2030 [Societal Drivers]
zhlédnutí 7KPřed 11 měsíci
[2/8] Control for Societal-Scale Challenges: Road Map 2030 [Societal Drivers]
[1/8] Control for Societal-Scale Challenges: Road Map 2030 [Introduction]9:16[1/8] Control for Societal-Scale Challenges: Road Map 2030 [Introduction]
[1/8] Control for Societal-Scale Challenges: Road Map 2030 [Introduction]
zhlédnutí 15KPřed 11 měsíci
[1/8] Control for Societal-Scale Challenges: Road Map 2030 [Introduction]
[6/8] Control for Societal-Scale Challenges: Road Map 2030 [Education]11:20[6/8] Control for Societal-Scale Challenges: Road Map 2030 [Education]
[6/8] Control for Societal-Scale Challenges: Road Map 2030 [Education]
zhlédnutí 3,4KPřed 11 měsíci
[6/8] Control for Societal-Scale Challenges: Road Map 2030 [Education]
[3/8] Control for Societal-Scale Challenges: Road Map 2030 [Technological Trends]10:46[3/8] Control for Societal-Scale Challenges: Road Map 2030 [Technological Trends]
[3/8] Control for Societal-Scale Challenges: Road Map 2030 [Technological Trends]
zhlédnutí 5KPřed 11 měsíci
[3/8] Control for Societal-Scale Challenges: Road Map 2030 [Technological Trends]
[7/8] Control for Societal-Scale Challenges: Road Map 2030 [Ethics, Fairness, & Regulatory Issues]8:08[7/8] Control for Societal-Scale Challenges: Road Map 2030 [Ethics, Fairness, & Regulatory Issues]
[7/8] Control for Societal-Scale Challenges: Road Map 2030 [Ethics, Fairness, & Regulatory Issues]
zhlédnutí 2,1KPřed 11 měsíci
[7/8] Control for Societal-Scale Challenges: Road Map 2030 [Ethics, Fairness, & Regulatory Issues]
[4/8] Control for Societal-Scale Challenges: Road Map 2030 [Emerging Methodologies]11:37[4/8] Control for Societal-Scale Challenges: Road Map 2030 [Emerging Methodologies]
[4/8] Control for Societal-Scale Challenges: Road Map 2030 [Emerging Methodologies]
zhlédnutí 4,2KPřed 11 měsíci
[4/8] Control for Societal-Scale Challenges: Road Map 2030 [Emerging Methodologies]
🤯🤯🤯 Integrating GPT-4 into Collimator.ai for Symbolic Modeling and Control** 🔥🔥🔥2:38🤯🤯🤯 Integrating GPT-4 into Collimator.ai for Symbolic Modeling and Control** 🔥🔥🔥
🤯🤯🤯 Integrating GPT-4 into Collimator.ai for Symbolic Modeling and Control** 🔥🔥🔥
zhlédnutí 19KPřed rokem
🤯🤯🤯 Integrating GPT-4 into Collimator.ai for Symbolic Modeling and Control 🔥🔥🔥