COG SCI LEC 14, 15 (dissociation and computing) + neural networks

TURNING MACHINE

a machine that by following a fixed set of procedures can give the answer to any mathematical problem

4 COMPONENTS OF THE TURNING MACHINE

1. physical tape = holds physical symbols

2. read - write head = writes and removes physical symbols on tape

3. state register = stores representations of physical tape at current time

4. machine table = contains all possible rules that machine can preform

TURNING MACHINE FUNCTIONAL DEFINITION

• any algorithmically calculable function are exactly the functions that can be computed by this machine

ALGORITHM

• a fixed set of procedures or rules that if followed will lead to a solution

TAPE

• divided into cells

• each cell is capable of holding one symbol

READ - WRITE HEAD

• detecting symbols in a cell

• adding symbols in a cell

• remove symbols from a cell

• move the head to another cell

STATE REGISTER

• a current record of the state the machine is in

• ex of what is in the register

• entries in each cell on the tape

• the location of the head on the tape

• the instructions that are going to be preformed

TABEL OF INSTRUCTIONS

a list of all the possible rules the machine can perform

FINITE STATE MACHINE

there is a limit to one of the following

• length of physical tape

• number of states

• number of instructions

• ability to record states

ARTIFICIAL NEURAL NETWORK

• a model of how neurons work

• a computer simulation of how actual populations of neurons perform tasks

NODE

• each node can be thought of a computing unit

• each computing unit represents a neuron

LINK

• describes the connection between two nodes

• how the output of one node affects the input received by a second node

THRESHOLD OF EXCITATION

• a threshold of excitation is a transitional point in electrical potential in a neuron

• neuron transitions from inactive to active

THRESHOLD OF EXCITATION (computational language)

• if the electrical signal a neuron receives is equal to or greater than the threshold of excitement than the neuron will fire

• if not neuron doesn’t fire

ACTIVATION FUNCTION

• is a computational model of the threshold of excitation

• relationship between the output of a node and the input into the node in computational terms

WEIGHTS IN ANN

• weight is the relative importance of each input link to determine the output of a node

• weights modify an output by a factor ranging from 1 to -1

ACTIVATION FUNCTION (2)

if sum of all weighted inputs a node receives is greater than threshold of excitatio, then node fires. If not, node does not fire