I am looking at the biological brain from the machine learning point of view and have formulated next question:
What is the List of the Essential Things for computation with the neuronal network inspired by the biological brain, rather than restrictions and shortcomings of the underlay chemical processes and biophysics?
Input to the neuron is the analog signal
Output of the neuron is the digital signal
Connections between neurons:
b) Recurrent, neuron itself or group of the neurons can form number of the feedback loops
c) Weight, the postsynaptic side signal strength - post synaptic potential (PSP)
d) Transmission speed - dendritic transmission speed rather than synaptic delay
e) Unique profile to be depressing, facilitating or pseudo-linear
f) Input timing matters, for spike time dependent plasticity (STDP)
g) Postsynaptic feedback, spike back-propagation involved in STDP mechanism
h) Place of the synapse matters, where in the dendritic tree it is placed
i) Type of the synapse matters, defining excitatory or inhibitory behavior
Post-synaptic potential propagation - having all above for the connections, one can assume that the every single event in the dendrite can be treated as the propagating wave
I) Does the assumptions above are correct?
II) What of the above should be dynamic and what can be freezed?
II) How does the STDP transformed into long-term memory, caused by Protein Kinase M Zeta^?
III) How does long-term memory transformed into the genetic memory?
IV) What is the motivation to establish new connection between neurons, “Glial Guidance” ^^?
V) What is the motivation to move the neuron, the neuronal locomotion?
VI) What is the motivation to born new neuron?
VII) What is the type of the new born neuron?
VIII) How the new born neuron evolve?
Maybe some of my questions are too naive and some are too abstract, as well as some already solved by neuroscience, but, anyway, let me know that! =)
Thank You for support!
…and one day we will realize, that we have built in-silico the General Purpose Artificial Intelligence