【COMP305 LEC 3 LEC 4】

LEC 3 A basic abstract model for a biological neuron

1. Weights of connections

Neuron gets fired if it has received from the presynaptic neurons 突触前神经元 a summary impulse 脉冲, which is above a certain threshold.

Signal from a single synapse突触 may sometime overcome the threshold and push a neuron to fire an action potential, but other synapses can achieve this only by simultaneously delivering their signals: Some inputs are more important!

定义:

Therefore, input from every synapse, or "connection", to the neuron in the abstract model must be assigned with some value w, called connection strength or weight of connection, to describe the importance of a connection.

2. Model

  1. The abstract neuron is excited when weighted sum is above the threshold 0

vs.

The biological neuron is excited when the signal density (spatial or temporal summation) is above the excitation potential threshold.

  1. Output is either 1 or 0.

vs.

Only the spikes(峰值)are remembered

LEC 4

Topic 2. The McCulloch-Pitts Neuron (1943)

1. McCulloch and Pitts demonstrated that

"...because of the all-or-none character of nervous activity, neural events and the relations among them can be treated by means of the propositional logic".

  1. The authors modelled the neuron as

a. a binary, discrete-time input

b. discrete-time:

The basic idea was to divide time into units, i.e., steps, and in each time period at most one spike can be initiated in the axon of a given neuron

将时间分成单位和步骤,每个时间一个神经元的轴突最多产生一次峰值

uniform velocity 脉冲基本都以匀速传播

Thus, the McCulloch-Pitts neuron operates on a discrete time scale,

t = 0,1,2,3, ...

c. binary input:

The types of the input and the output of a MP neuron are thus unified.

d. with excitatory and inhibitory connections 有着兴奋和抑制之间的联系 and an excitation threshold. 兴奋阙值

The network of such elements was the first model to tie the study of neural networks to the idea of computation in its modern sense.

将神经网络和现代意义上的计算思想联系起来

e. with excitatory and inhibitory connections 有着兴奋和抑制之间的联系 and an excitation threshold. 兴奋阙值

f. The network of such elements was the first model to tie the study of neural networks to the idea of computation in its modern sense. 将神经网络和现代意义上的计算思想联系起来

g. excitatory and inhibitory connections :

The weight of connection wi are:

+1 for excitatory type connection and 加一促进

Cerebral pyramidal cell:

-1 for inhibitory type connection. 减一抑制

h. Threshold

I. MP Neuron

In the MP neuron, we call the instant total input

St-1: instant stateof the neuron

j. Actication Function

相关推荐
m0_7382065443 分钟前
嵌入式学习的第二十二天-数据结构-栈+队列
数据结构·学习
向上的车轮6 小时前
MATLAB学习笔记(七):MATLAB建模城市的雨季防洪排污的问题
笔记·学习·matlab
前端小崔7 小时前
从零开始学习three.js(18):一文详解three.js中的着色器Shader
前端·javascript·学习·3d·webgl·数据可视化·着色器
龙湾开发8 小时前
计算机图形学编程(使用OpenGL和C++)(第2版)学习笔记 10.增强表面细节(二)法线贴图
c++·笔记·学习·图形渲染·贴图
liang_20268 小时前
【HT周赛】T3.二维平面 题解(分块:矩形chkmax,求矩形和)
数据结构·笔记·学习·算法·平面·总结
虾球xz8 小时前
游戏引擎学习第290天:完成分离渲染
c++·人工智能·学习·游戏引擎
虾球xz8 小时前
游戏引擎学习第285天:“Traversables 的事务性占用”
c++·学习·游戏引擎
虾球xz9 小时前
游戏引擎学习第280天:精简化的流式实体sim
数据库·c++·学习·游戏引擎
深度学习入门9 小时前
学习深度学习是否要先学习机器学习?
人工智能·深度学习·神经网络·学习·机器学习·ai·深度学习入门
FAREWELL000759 小时前
Unity基础学习(十五)核心系统——音效系统
学习·unity·c#·游戏引擎