Difference between revisions of "Human brain computations"

From RB Wiki
(Created page with "To estimate the capabilities of algorithms, it was worth comparing computational properties of machines to the human brains'. == Throughput == Most of the human brain's inpu...")
 
m (Add references, but without explanation)
Line 4: Line 4:
  
 
Most of the human brain's input seems to come from the visual cortex, at a rate of ~10<sup>10</sup> bits per second, which are reduced to ~10<sup>4</sup> bits per second in layer IV of region V1 [https://www.sciencedirect.com/science/article/pii/S136466131000029X/pdfft?md5=9156147251d5d7da0d08a0456b24db59&pid=1-s2.0-S136466131000029X-main.pdf Raichle][https://scholar.google.ch/scholar?hl=en&as_sdt=0%2C5&q=Two+views+of+brain+function+raichle&btnG= 10].
 
Most of the human brain's input seems to come from the visual cortex, at a rate of ~10<sup>10</sup> bits per second, which are reduced to ~10<sup>4</sup> bits per second in layer IV of region V1 [https://www.sciencedirect.com/science/article/pii/S136466131000029X/pdfft?md5=9156147251d5d7da0d08a0456b24db59&pid=1-s2.0-S136466131000029X-main.pdf Raichle][https://scholar.google.ch/scholar?hl=en&as_sdt=0%2C5&q=Two+views+of+brain+function+raichle&btnG= 10].
 +
 +
 +
== Bayesian brain ==
 +
 +
Tenebaum
 +
 +
== Reward system ==
 +
 +
[https://pubmed.ncbi.nlm.nih.gov/31942076-a-distributional-code-for-value-in-dopamine-based-reinforcement-learning/ DKUSH+][https://scholar.google.ch/scholar?hl=en&as_sdt=0%2C5&q=A+Distributional+Code+for+Value+in+Dopamine-Based+Reinforcement+Learning&btnG= 20]

Revision as of 12:37, 21 January 2020

To estimate the capabilities of algorithms, it was worth comparing computational properties of machines to the human brains'.

Throughput

Most of the human brain's input seems to come from the visual cortex, at a rate of ~1010 bits per second, which are reduced to ~104 bits per second in layer IV of region V1 Raichle10.


Bayesian brain

Tenebaum

Reward system

DKUSH+20