🎴feedback to neurons & actions to get profit

    🎴profit into neural network & actions to get profit

 

 

 

 

    what is neural network ?

    ① its system in human's brain,  and system of Artifact intelligence

    ② neural network, is system of our brain.

 

    brain & neural network, compose with  neurons.

 

    how neuron moves ?

 

    ① neuron, exist in brain.

    ② neuron's input    = electric flow.

    ③ neuron's output = electric flow.

    ④ neuron get several input from some newrons

    ⑤ if input is powerful enough, neuron output electric flow to other neuron.

    

 

    how neuron study ?

 

    ① neuron, get input

    ② neuron, output electric flow.

    ③ if neuron input & output many times, neuron power up .

    ④ when neuron power up, output increase

 

    usually, input & output of neuron, treat as 0 ~ 1 number, in program.

 

    flash of neuron

    ① neuron, output electric flow., call as flash.

    ② when neuron flashes, = when neuron power up

    ③ how neuron power up?   ... neuron's string's Cross-sectional area, become broaden.

                                                ... writer heard length between neuron, become more closer.

 

    how neural network moves ?

 

    ① neurons, get electric flow

    ② neurons, stand in row.

    ③ neurons, create network.

    ④ when neurons get image of something, particular neurons in brain, get electric flow input

    ⑤ when neurons in brain get reword , they return OK feedback.

    ⑥ when OK feedback happens, connection within that neurons, power up 

 

    neural network example recognize number program

    ① prepair neurons, connect & row in 3 layers

    ② convert number image into input into 1st layer

    ③ 2nd layer, get inpur from 1st layer

    ④ 3rd layer, output answer, what was imput image, with  10 neurons, 

    ⑤ if answer correct, neuron where flash, power up

          => its called reward, if answer right.

    ⑥ if answer not correct, neuron where flash, become weak.

          => ⑤ & ⑥ called feedback, from answer.

 

    each neuron, create output, with particular function

    ① if input from 4 neuron, a, b, c, d exist

    ② if output  of neuron = f ( a,b,c,d )

    ③ f ( a,b,c,d ) = a × ? + b × ? + c × ? + D × ?  

 

    each neuron, study from feedback, with particular function

    ① if answer right 

    ② if particular neuron flash

    ③ f ( a,b,c,d ) = a × ? + b × ? + c × ? + D × ? 

    ④ coefficient, of a, b, c, d increase.

         (  in program, programmer adjust how much increase  )

 

    ① if answer wrong 

    ② if particular neuron flash

    ③ f ( a,b,c,d ) = a × ? + b × ? + c × ? + D × ? 

    ④ coefficient, of a, b, c, d reduce.

         (  in program, programmer adjust how much increase )

 

    neural network, sometime overstudy.

    ① even neural network study, answer ration not increase

    ② reason 1⃣ study with same text 

    ③ reason 2⃣ too much reward, into neuron

    ④ reason 3⃣ study text, is biased

                          to think example within number recognaize program,

                              picture of 1, 2, 3 is too many, 

                              and picture of 7 ~ 0 is small, etc.

                          to think example with exam,

                              math text & review test is too many, 

                              and social study text & review test is small ,

                              but at exam, we have to take 100 point in each exam, etc.

 

   info blackhole, in neural network 

 

   if we think society = neural network, 

   mechanism of info blackhole

    research from writer