YOLO activation_layer 代码学习

激活层定义

typedef enum{    LOGISTIC, RELU, RELIE, LINEAR, RAMP, TANH, PLSE, LEAKY, ELU, LOGGY}ACTIVATION;大家一开始看这些激活函数看起来很奇怪，下面我们会给出这几个类型的表达式

CPU 端的代码

激活函数对应的代码：static inline float linear_activate(float x){return x;}static inline float logistic_activate(float x){return 1./(1. + exp(-x));}static inline float loggy_activate(float x){return 2./(1. + exp(-x)) - 1;}static inline float relu_activate(float x){return x*(x>0);}static inline float elu_activate(float x){return (x >= 0)*x + (x < 0)*(exp(x)-1);}static inline float relie_activate(float x){return x*(x>0);}static inline float ramp_activate(float x){return x*(x>0)+.1*x;}static inline float leaky_activate(float x){return (x>0) ? x : .1*x;}static inline float tanh_activate(float x){return (exp(2*x)-1)/(exp(2*x)+1);}static inline float plse_activate(float x){    if(x < -4) return .01 * (x + 4);    if(x > 4)  return .01 * (x - 4) + 1;    return .125*x + .5;}激活函数的导数：static inline float linear_gradient(float x){return 1;}static inline float logistic_gradient(float x){return (1-x)*x;}static inline float loggy_gradient(float x){    float y = (x+1.)/2.;    return 2*(1-y)*y;}static inline float relu_gradient(float x){return (x>0);}static inline float elu_gradient(float x){return (x >= 0) + (x < 0)*(x + 1);}static inline float relie_gradient(float x){return (x>0) ? 1 : .01;}static inline float ramp_gradient(float x){return (x>0)+.1;}static inline float leaky_gradient(float x){return (x>0) ? 1 : .1;}static inline float tanh_gradient(float x){return 1-x*x;}static inline float plse_gradient(float x){return (x < 0 || x > 1) ? .01 : .125;}

GPU端的代码

//  __device__ 该函数在器件里调用，在器件中执行//  __global__ 该函数在主机里调用，在器件中执行//    __host__ 该函数在主机中调用，在主机中执行激活函数对应的代码：__device__ float linear_activate_kernel(float x){return x;}__device__ float logistic_activate_kernel(float x){return 1./(1. + exp(-x));}__device__ float loggy_activate_kernel(float x){return 2./(1. + exp(-x)) - 1;}__device__ float relu_activate_kernel(float x){return x*(x>0);}__device__ float elu_activate_kernel(float x){return (x >= 0)*x + (x < 0)*(exp(x)-1);}__device__ float relie_activate_kernel(float x){return x*(x>0);}__device__ float ramp_activate_kernel(float x){return x*(x>0)+.1*x;}__device__ float leaky_activate_kernel(float x){return (x>0) ? x : .1*x;}__device__ float tanh_activate_kernel(float x){return (exp(2*x)-1)/(exp(2*x)+1);}__device__ float plse_activate_kernel(float x){    if(x < -4) return .01 * (x + 4);    if(x > 4)  return .01 * (x - 4) + 1;    return .125*x + .5;}激活函数的导数：__device__ float linear_gradient_kernel(float x){return 1;}__device__ float logistic_gradient_kernel(float x){return (1-x)*x;}__device__ float loggy_gradient_kernel(float x){    float y = (x+1.)/2.;    return 2*(1-y)*y;}__device__ float relu_gradient_kernel(float x){return (x>0);}__device__ float elu_gradient_kernel(float x){return (x >= 0) + (x < 0)*(x + 1);}__device__ float relie_gradient_kernel(float x){return (x>0) ? 1 : .01;}__device__ float ramp_gradient_kernel(float x){return (x>0)+.1;}__device__ float leaky_gradient_kernel(float x){return (x>0) ? 1 : .1;}__device__ float tanh_gradient_kernel(float x){return 1-x*x;}__device__ float plse_gradient_kernel(float x){return (x < 0 || x > 1) ? .01 : .125;}__device__ float activate_kernel(float x, ACTIVATION a){    switch(a){        case LINEAR:            return linear_activate_kernel(x);        case LOGISTIC:            return logistic_activate_kernel(x);        case LOGGY:            return loggy_activate_kernel(x);        case RELU:            return relu_activate_kernel(x);        case ELU:            return elu_activate_kernel(x);        case RELIE:            return relie_activate_kernel(x);        case RAMP:            return ramp_activate_kernel(x);        case LEAKY:            return leaky_activate_kernel(x);        case TANH:            return tanh_activate_kernel(x);        case PLSE:            return plse_activate_kernel(x);    }    return 0;}__device__ float gradient_kernel(float x, ACTIVATION a){    switch(a){        case LINEAR:            return linear_gradient_kernel(x);        case LOGISTIC:            return logistic_gradient_kernel(x);        case LOGGY:            return loggy_gradient_kernel(x);        case RELU:            return relu_gradient_kernel(x);        case ELU:            return elu_gradient_kernel(x);        case RELIE:            return relie_gradient_kernel(x);        case RAMP:            return ramp_gradient_kernel(x);        case LEAKY:            return leaky_gradient_kernel(x);        case TANH:            return tanh_gradient_kernel(x);        case PLSE:            return plse_gradient_kernel(x);    }    return 0;}__global__ void activate_array_kernel(float *x, int n, ACTIVATION a){    //这个的计算可以参照下面的示意图，因为每个block的维度是一维的，因此只需要使用threadIdx.x。    int i = (blockIdx.x + blockIdx.y*gridDim.x) * blockDim.x + threadIdx.x;    if(i < n) x[i] = activate_kernel(x[i], a);}__global__ void gradient_array_kernel(float *x, int n, ACTIVATION a, float *delta){    int i = (blockIdx.x + blockIdx.y*gridDim.x) * blockDim.x + threadIdx.x;    if(i < n) delta[i] *= gradient_kernel(x[i], a);}//extern "C"表明了一种编译规约，其中extern是关键字属性，“C”表征了编译器链接规范。对于extern "C"可以理解成在C++/C中的混合编程的编译指令。extern "C" void activate_array_ongpu(float *x, int n, ACTIVATION a) {    activate_array_kernel<<<cuda_gridsize(n), BLOCK>>>(x, n, a);    check_error(cudaPeekAtLastError());}extern "C" void gradient_array_ongpu(float *x, int n, ACTIVATION a, float *delta) {    gradient_array_kernel<<<cuda_gridsize(n), BLOCK>>>(x, n, a, delta);    check_error(cudaPeekAtLastError());}//这个地方已经定义了BLOCK的大小为512，这个函数的功能是使得设置网格的维度，要求是x的大小不超过65535，返回网格的大小。网格的示意图如下图。dim3 cuda_gridsize(size_t n){    size_t k = (n-1) / BLOCK + 1;    size_t x = k;    size_t y = 1;    if(x > 65535){         x = ceil(sqrt(k));         y = (n-1)/(x*BLOCK) + 1;    }    dim3 d = {x, y, 1};    return d;}

对于给定的线程的坐标为（blockIdx.x,blockIdx.y,threadIdx.x,threadIdx.y）对应的线程标号为(blockIdx.y*gridDim.x + blockIdx.x)*BlockDim.x*BlockDim.y + (threadIdx.y*BlockDim.x + threadIdx.y)对应于这个程序代码里面 ，BlockDim.x = BLOCK , BlockDim.y = 1, threadIdx.y = 0对应的线程标号为 (blockIdx.y*gridDim.x + blockIdx.x)*BlockDim.x* + threadIdx.y

LOGISTIC 的表达式

y=1.01.0+e−x

dydx=1

RELU 的表达式

y={x0 if x>0 if x≤0

RELIE 的表达式

y={x0if x>0if x≤0

dydx={10if x>0if x≤0

LINEAR 的表达式

y=x

dydx=1

RAMP 的表达式

y={x+0.1x0.1xif x>0if x≤0

dydx={1+0.10.1if x>0if x≤0

TANH 的表达式

y=e2x−1e2x+1

dydx=1−x2

PLSE 的表达式

y=⎧⎩⎨0.01(x+4)0.01(x−4)0.125x+0.5if x<−4if x>4if x=4

dydx={0.010.125if x<0∥x>1if x=4

LEAKY 的表达式

y={x0.1xif x>0if x≤0

dydx={10.1if x>0if x≤0

ELU 的表达式

y={xex−1if x≥0if x<0

dydx={1x+1if x≥0if x<0

LOGGY 的表达式

y=2.01.0+e−x−1=1.0−e−x1.0+e−x

dydx=21−x1+x