Physics Informed Deep Learning in 2 Minutes.
As I dive into my Masters Degree in Civil Engineering and developing the future living experience called "AI Smart Homes".
Invest in AI Smart Homes: Wefunder.com/aismarthomes
I came across a sector of Deep Learning that really fascinated me. For those that do not know Deep Learning is a Higher form of Machine Learning.
Machine Learning is the technology behind Artificial Intelligence. Within Machine Learning you have 3 Main Sections:
Machine Learning typically utilizes 1 Terabyte of Data or Less.
Small Data is considered 1 Terabyte or less.
Deep Learning typically utilizes 1 Terabyte of Data or More.
Big Data is considered 1 Terabyte or more.
For Machine Learning we would use Physics Informed Machine Learning.
Physics Informed machine learning combines the power of data-driven machine learning with the knowledge of physics with Small Data.
For Deep Learning we would use Physics Informed Deep Learning.
领英推荐
Physics Informed Deep learning combines the power of data-driven Deep Learning with the knowledge of physics for Big Data.
Today I will be talking about using Physic Informed Deep Learning with Big Data in Robotics.
Physics Informed Deep Learning (PIDL) is like a turbo engine for Robotics.
PIDE can accelerate reinforcement learning by incorporating prior knowledge of robot dynamics, reducing the number of required training episodes.
PIDE can be used to learn accurate models of robot dynamics for MPC, enabling more precise and robust control.
PIDE helps robots adapt quicker in real time by learning and updating their models based on new data.
PIDE can help merge data from multiple sensors in real time. An example of this would be sensors from different angles capturing pictures being able to comprehend immediately what is being captured.
PIDE can often learn quicker and more effectively with less data than traditional or older deep learning methods.