GIszter Flashcards
simplest possible model
input weighted
function
output
cascade
layered, acyclic, one direction
has hidden layers
recurrent/recursive
cyclic
some go back
can be self exciting as is the case for CPGs
interconnected
like everything is cyclic
hidden layers
layers of interneurons between input and output
perceptrons
edeg detection movinge directions inhibitory layer step functions cant take derivative so difficult to analyze but easy to program
edge detection
edges have less activation so they activate less inhibitors so a set of outputs get inhibited less and are higher than middle ones which are higher than outter ones which dont even have much activation
McCulloch-Pitts
inhibition trumps al; discrete thrshould neurons
turing machine with fixed weights and binary output
hopfield content addressable memory CAM
fully interconnected
matching with something it already knows
very recursive
plastic weights so updatable
fan in
converging
averages population
fan out
diverging for accuracty
weight strength
all networks incorporate weights
like how strong is the EPSP/IPSP
spiking
digital code
longe range
fast
discrete
non spiking
analog short range slow no action potetnial in retina and olfactory bulb
lateral inhibition
like edge detection again
output is better than input because modified
mauthner cells
turn
inhibition of other
singl cell drives decision and controlls lots of motor neurons to evade
DCMD
decending contralateral motion detection
used to guid flight and direction
intermediate proccestion of info to uide motor output
AXON CROSSES MIDLINE
from eye to opposite side to turn away from looming object
just mediates, not a binary response
contralateral
excitatory toward stim
inhibitory away from stim
ipsilateral
excitatorr away from stim
inhibitory toward
plasticity
local - Hebb fire and wire, so use or lose
external teacher - reinforce as it gets close to correct
occams razor
as simple as possible but as complex as necessary to perform a function/describe a mechanism
NMJ
1:1 firing ratio so no integration or info processiing at this point
motor neuron populaitons
firing of single neurone can recruit others in relay fashion
recruitment makes smoother
motor pools as integration layer in spinal network
Size (henneman) principle
slow fibers recruited first for fine controlled tasks
slow motor neurons have smaller somas and therefore have higher resistence so a lower threshold
