AI Learn Physics | Projectile Motion with Neural Network 

If you like the video, please subscribe to the channel by using the below link
tinyurl.com/1w5i9nnu
An AI neural network learns how to throw balls using projectile motion generated data. I have used 3d simulation to show how a neural network can learns throwing a ball using projectile motion.
I have generated data for training a neural network to learn ball motion with different angles.
The training process shown in this video took about 15-20 min on google colab while training with tensor flow 2.3 utilizing multiple shape of neural network.
you may have seen rocket motion or fountain motion or ball motion all motion all are governed by some physics equation. I have used simple neural network to learn the
projectile path which is hightly nonliner in nature and governed by complex mathematical equation. with 2 leyer neural network I was able to complete the task.
The AI starts off with random behavior, i.e. the Neural Network is initialized with random weights. It then gradually learns to understanding the behavior of curve and trains up.
The AI consists of a deep Neural Network with 2 hidden layers of 10 neurons each.
I have shown here how different type of activation function can affect the training process. Projectile motion is governed by non linear equation. In this video you will under how activation function can affect the optimization and how simple and strong neural network behaves with the data given.
Most of physics equation are governed by complex mathematical equation here in this example I have shown we can train a neural network to learn complex physics and mathematical equations.
#ArtificialIntelligence #MachineLearning #ReinforcementLearning #AI #NeuralNetworks