This posting referenced
https://github.com/youbeebee/deeplearning_from_scratch and the book

My environment

this book uses anaconda, python(3.5.2)

• python3: 3.9.6
• venv: name is 'yo'
• m1 Mac air
• numpy(1.24.0), matplotlib(3.6.2)

Perceptron

Chapter 2
perceptron algorithm

• multi input, single output
• parameter: weights, bias(threshold)
• output is 0 or 1

AND, OR, NAND vs XOR

limits of simple perceptron

How perceptron works can be figured out.
OR

• triangle: 1, circle: 0
• perceptron seperates space into 2 parts, one is 0-domain and another is 1-domain
• the point is it seperates with straight line, or linearity
• And, simple perceptron makes only linearity

But the linearity cannot seperate XOR
it can be done only with curved line, or non-linearity

then we need multilayer perceptron.