I am getting confused with the literature. Is a perceptron simply a network of MCP neurons? From what I understand, in 1957 Rosenblatt developed the perceptron based on relaxed constraints from the MCP neurons (by McCulloch and Pitts). Here are some statements I have come up with:
- MCP neurons treated every input equally (i.e. all weights set to 1). Perceptron introduced variable weights, which could therefore be trained.
- BOTH MCP and perceptron used a bias that was set to a single value.
- BOTH MCP and perceptron have Boolean inputs (at least originally).
- BOTH MCP and perceptron apply a threshold activation function (i.e. these networks can tell you is something is A or B, nothing more).
Are these true statements? I am very confused.