Spike Computing
をテンプレートにして作成
[
トップ
] [
新規
|
一覧
|
単語検索
|
最終更新
|
ヘルプ
|
ログイン
]
開始行:
[[Neuromorphic]]
CENTER:SIZE(40){COLOR(green){Computing with Deep Neural Spikes}}
*Neuron Coding [#n4a603ff]
There are three known types of neuron coding:
-1. Rate Code - neuroscience have been using this coding quite successful (80th)
The rate code- the average number of action potentials/spikes per unit time
Rate code alone is insufficient for account for many sofisticated spike trains especially from a population of neurons [1].
In motor neurons, the strength of an innervated muscle flexion depends solely on the firing rate.
--Stochastic means nondeterministic or unpredictable. Random generally means unrecognizable, not adhering to a pattern.
-2. Inter-spike interval: Frequency, Correlation, Modeling (earlier 90th)
-3. Temporal Code (Price timing of spikes)
RIGHT:SIZE(10){COLOR(green){Last update: 1/11/2018}}
-References:
-1.[[Jennie Si: "Computing with Neural Spikes">https://www.youtube.com/watch?v=DSr1RjUHpXs]]
-2. [[Spike-timing based neuronal information processing: applications to vision and speech>https://www.youtube.com/watch?v=f0nMdWxzXkc]]
*CMOS Neuron [#uff3a8a5]
-http://www.lumerink.com/pages/neuromorphic.html
*Synapse [#v3d5e1aa]
Memristors have reawakened the neuromorphic community with the promise of providing the missing piece to the construction of low-power neuromorphic hardware. A memristor is a nanodevice that exhibits a resistance value which changes based on the past electrical charge that has passed through it. This means that the memristor is able to store information based on the past history of its activation, similarly to brain synapses.
An added bonus is that they have the potential to allow
designers to reach synaptic densities much closer to biological levels with respect to competing approaches without taking up too much space or using too much power. Because of their small size and favorable dynamics, memristors also have been used to model synapses in analog circuits that are able to learn in real time [[[Ref>http://nl.bu.edu/wp-content/uploads/2012/02/Ames_et_al_Memristors_Mind_2012.pdf]]].
終了行:
[[Neuromorphic]]
CENTER:SIZE(40){COLOR(green){Computing with Deep Neural Spikes}}
*Neuron Coding [#n4a603ff]
There are three known types of neuron coding:
-1. Rate Code - neuroscience have been using this coding quite successful (80th)
The rate code- the average number of action potentials/spikes per unit time
Rate code alone is insufficient for account for many sofisticated spike trains especially from a population of neurons [1].
In motor neurons, the strength of an innervated muscle flexion depends solely on the firing rate.
--Stochastic means nondeterministic or unpredictable. Random generally means unrecognizable, not adhering to a pattern.
-2. Inter-spike interval: Frequency, Correlation, Modeling (earlier 90th)
-3. Temporal Code (Price timing of spikes)
RIGHT:SIZE(10){COLOR(green){Last update: 1/11/2018}}
-References:
-1.[[Jennie Si: "Computing with Neural Spikes">https://www.youtube.com/watch?v=DSr1RjUHpXs]]
-2. [[Spike-timing based neuronal information processing: applications to vision and speech>https://www.youtube.com/watch?v=f0nMdWxzXkc]]
*CMOS Neuron [#uff3a8a5]
-http://www.lumerink.com/pages/neuromorphic.html
*Synapse [#v3d5e1aa]
Memristors have reawakened the neuromorphic community with the promise of providing the missing piece to the construction of low-power neuromorphic hardware. A memristor is a nanodevice that exhibits a resistance value which changes based on the past electrical charge that has passed through it. This means that the memristor is able to store information based on the past history of its activation, similarly to brain synapses.
An added bonus is that they have the potential to allow
designers to reach synaptic densities much closer to biological levels with respect to competing approaches without taking up too much space or using too much power. Because of their small size and favorable dynamics, memristors also have been used to model synapses in analog circuits that are able to learn in real time [[[Ref>http://nl.bu.edu/wp-content/uploads/2012/02/Ames_et_al_Memristors_Mind_2012.pdf]]].
ページ名: