斯坦福大学CS231n——assignmentv1——linear_svm

开篇

今天我们聊一下linear_svm线性支持向量机。首先我们需要知道线性svm的损失函数，这也是作业中要求实现的主要部分。

线性支持向量机

svm的损失函数基本上都是hinge损失：loss = max(0,y_pred - y + 1)
这个1表示的是我们损失函数中的margin，意思为我们最多可以接受这么多的距离误差。二者的差值超过了1就会被我们当作错误样本来计算损失了。
作业里面也要求我们使用两种方法，一种是naive版本，遍历每一个样本，计算在每个类别上的得分，然后与真实类别的得分做差并+1，得到hinge损失的后半部分，至于前半部分，无非是与0进行比较。
第二种方法是以向量的形式做差并求解，更新dw。对于预测正确的样本，dw应该减去总体的损失值，因为预测正确的样本的权重不用更新太多。
具体代码中的细节我会以注释的形式说明。

代码

import numpy as np
from random import shuffle


def svm_loss_naive(W, X, y, reg):
    """
  Structured SVM loss function, naive implementation (with loops).

  Inputs have dimension D, there are C classes, and we operate on minibatches
  of N examples.

  Inputs:
  - W: A numpy array of shape (D, C) containing weights.
  - X: A numpy array of shape (N, D) containing a minibatch of data.
  - y: A numpy array of shape (N,) containing training labels; y[i] = c means
    that X[i] has label c, where 0 <= c < C.
  - reg: (float) regularization strength

  Returns a tuple of:
  - loss as single float  损失
  - gradient with respect to weights W; an array of same shape as W  dW
  """
    dW = np.zeros(W.shape)  # initialize the gradient as zero

    # compute the loss and the gradient
    num_classes = W.shape[1]
    num_train = X.shape[0]
    loss = 0.0
    for i in range(num_train):
        scores = X[i].dot(W)
        correct_class_score = scores[y[i]]
        for j in range(num_classes):
            if j == y[i]:
                continue
            margin = scores[j] - correct_class_score + 1  # note delta = 1
            if margin > 0:
                loss += margin
                dW[:, j] += X[i].T
                dW[:, y[i]] -= X[i].T

    # Right now the loss is a sum over all training examples, but we want it
    # to be an average instead so we divide by num_train.
    loss /= num_train
    dW /= num_train

    # Add regularization to the loss.
    # 正则化
    loss += reg * np.sum(W * W)
    # 损失函数求导
    dW += 2 * reg * W
    #############################################################################
    # TODO:                                                                     #
    # Compute the gradient of the loss function and store it dW.                #
    # Rather that first computing the loss and then computing the derivative,   #
    # it may be simpler to compute the derivative at the same time that the     #
    # loss is being computed. As a result you may need to modify some of the    #
    # code above to compute the gradient.                                       #
    #############################################################################

    return loss, dW


def svm_loss_vectorized(W, X, y, reg):
    """
  Structured SVM loss function, vectorized implementation.

  Inputs and outputs are the same as svm_loss_naive.
  """
    loss = 0.0
    dW = np.zeros(W.shape)  # initialize the gradient as zero
    num_classes = W.shape[1]
    num_train = X.shape[0]

    #############################################################################
    # TODO:                                                                     #
    # Implement a vectorized version of the structured SVM loss, storing the    #
    # result in loss.                                                           #
    #############################################################################

    scores = X.dot(W)
    score_y = scores[np.arange(num_train), y]
    score_y = np.tile(score_y.reshape(num_train, 1), (1, num_classes))
    margin = scores - score_y + 1
    margin[np.arange(num_train), y] = 0
    margin[margin < 0] = 0
    loss += np.sum(margin) / num_train
    loss += reg * np.sum(W * W)

    #############################################################################
    #                             END OF YOUR CODE                              #
    #############################################################################

    #############################################################################
    # TODO:                                                                     #
    # Implement a vectorized version of the gradient for the structured SVM     #
    # loss, storing the result in dW.                                           #
    #                                                                           #
    # Hint: Instead of computing the gradient from scratch, it may be easier    #
    # to reuse some of the intermediate values that you used to compute the     #
    # loss.                                                                     #
    #############################################################################

    mask_margin = np.zeros((num_train, num_classes))  # where margin[i,j]>0=1
    # margin > 0的全部设置为1
    mask_margin[margin > 0] = 1
    mask_XW = np.ones((num_train, num_classes))
    # 这里相当于把mask_margin复制了一遍再给mask_XW
    mask_XW = mask_XW * mask_margin
    # 将所有损失,即margin算出来
    y_sum = np.sum(mask_margin, axis=1)
    # 正确的分类的不需要更新太多
    mask_XW[np.arange(num_train), y] -= y_sum
    dW = np.dot(X.T, mask_XW) / num_train
    # 正则化求导
    dW += reg * W

    #############################################################################
    #                             END OF YOUR CODE                              #
    #############################################################################

    return loss, dW

总结

前两次作业都是比较简单的，基本上都是计算损失函数和一些数据的处理，比如大于0的margin全部设为1，这样可以使矩阵中只含有0和1两个值，还有就是计算损失的两种方法，一种就是一一计算样本每个类别的得分，第二种就是向量化的计算，即将所有训练样本与W直接相乘，然后将正确类别的得分也变为同尺度的矩阵，做差算出margin，再接着处理，可以所有训练样本一起计算，不需要单独拿出来一一计算。
好啦这次作业说完了，下次我们说另一种分类器softmax。