a classification problem

learning curves

learning rates

<->

performance, motivation (consistency, accuracy)

linear learning, nonlinear learning..?

basic reading (linear learning; one book at a time)

complex reading (nonlinear learning, open all the books)

space platinum, space magnesium, space zinc.

plus asteroids as space stations for fuel, water and air.

Planetary Resources (expand humanity, and our economy, to outer space)

1. We have to do this for the sake of humanity (collective subconscious, community psychology)

2. Appealing to the investors by specifying “the economy” as if it stands separately from humanity.

vs

Deep Space Industries (there are minerals to be mined)

1. We have to seize this opportunity SEIZE (time is of the essence children)

million

billion

trillion

quadrillion

quintillion

sextillion

septillion

octillion

nonilion

decillion

undecillion

duodecillion

tredecilion

quattuordecillion... ...

...

1) 0-1 “neurons” (0 if image is categorized as not(XX), 1 if image is categorized as XX)

2) re-assigning images into single line pixels (easier, 98.+% accuracy) OR addition of all subsamples of pixels into “convolutional layers” (99.+% accuracy)

3) “dropout” randomly dropping out 00% of neurons

4) exponentially decay the “learning rate” to smooth out the accuracy curve (learning too fast)

5) watch again.

additional notes :