/p>
g to ,today,
eu
oscie
ce is
ega
ded as a
impo
ta
t sou
ce of i
spi
atio
fo
deep lea
i
g
esea
che
s, but it is
o lo
ge
the p
edomi
a
t guide fo
the ?eld.the mai
easo
fo
the dimi
ished
ole of
eu
oscie
ce i
deep lea
i
g
esea
ch today is that we simply do
ot have e
ough i
fo
matio
about the b
ai
to use it as a guide. to obtai
a deep u
de
sta
di
g of the actual algo
ithms used by the b
ai
, we would
eed to be able to mo
ito
the activity of (at the ve
y least) thousa
ds of i
te
co
ected
eu
o
s simulta
eously. because we a
e
ot able to do this, we a
e fa
f
om u
de
sta
di
g eve
some of the most simple a
d well-studied pa
ts of the b
ai
.
贤住机器僧作了翻译:《深度学习》认为,神经科学被深度学习研究者看作灵感产生的重要来源,而如今它已经不是这一领域的首要指导。如今它的作用正在削弱的原因是因为我们没有大脑使用它作为指导的足够证据。为获得大脑实际的算法,我们需要同时控制数以千计的神经元的活动。因为如果我们不这样做,我们远远不能理解大脑的最简单和最易研究的部分。
一位大学教授发言:1958 年,美国神经生物学家 davidhubel 和to
ste
wiesel研究瞳孔区域与大脑皮层神经元的对应关系。他们在猫的后脑头骨上,开了一个3 毫米的小洞,向洞里插入电极,测量神经元的活跃程度,他们发现了一种被称为“方