如何更新漏洞集成和激活神经元

时间:2016-01-27 16:33:40

标签: c++ neural-network

我正在尝试用C ++实现泄漏的集成和激发神经元,但是当新电流到来时,无法找到更新神经元电压的好方法。

我现在正在使用此代码编写,但我是该领域的新手,并不认为这是一个最佳解决方案。这是更新神经元的正确方法吗?

void LIF_neuron::update(double current, size_t time){
    //update current
    //formula: (V_current - V_reset)*(1-exp((t_last_fired - t_now)/RC)) + R*I
    mCurrentVolt += -1*(mCurrentVolt-mResetVolt)*(1-exp((mLastTimeFired-time)/mRC)) + mR*current;

    //all leakage since last activation was substracted, give back what was already substracted in earlier updates
    mCurrentVolt += (mCurrentVolt-mResetVolt)*(1-exp((mLastTimeFired-mLastTimeUpdated)/mRC));

    if(mCurrentVolt >= mThreshold){
        mFires = true;
        mLastTimeFired = time;
    }else{
        mFires = false;
    }

    mLastTimeUpdated = time;
}

mRCmR是神经元的常数。

0 个答案:

没有答案