Introduction to XOR Gate with AI using Keras
Starting with AI can be tricky. So, I've looked for a simple AI example to share, and I found one I really like called the XOR Hello World. In this, we're going to use Keras, a Python library that sits on TensorFlow or LibTorch.
Along with Keras, we use NumPy, another Python library that manipulates sets of data and reduces our need to write extra codes. Keras offers two types of layers for our model: dense and activation. Dense is a unique matrix with random numbers, while activation acts as a safeguard to prevent numbers from skyrocketing to infinity or dropping to zero during training.
Both these functions help us train our AI without having to write for loops or range loops. Besides these, we have an optimizer that tells us how we're going to train our model. In our case, it's random, - ‘stochastic gradient descent’ or SGD. The real magic comes from the training data.
We have features and labels, where features are inputs and labels are the outputs we train the model to repeat. For the XOR operator, when comparing two Boolean values, if both values are the same, we get a false. But if either of them is different, we get a true.
This model can also be trained for different operators like OR or AND. The beauty of AI is once you train it to do a task, it can keep performing that task.