链接: [link]https://www.cnblogs.com/guoyaohua/p/8542554.html
种类
1.梯度下降最常见三种（BGD，SGD，MBGD）
三种形式的区别就是取决于我们用多少数据来计算目标函数的梯度，这样的话自然就涉及到一个 trade－off，即参数更新的准确率和运行时间
①1.Batch Gradient Descent （BGD）-----批量梯度下降法
代价函数
J ( θ ) = 1 2 m ∑ i = 1 m ( y i − h θ ( x i ) ) 2 % MathType!MTEF!2!1!+- % feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn % hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr % 4rNCHbWexLMBbXgBd9gzLbvyNv2CaeHbl7mZLdGeaGqipu0Je9sqqr % pepC0xbbL8F4rqqrFfpeea0xe9Lq-Jc9vqaqpepm0xbba9pwe9Q8fs % 0-yqaqpepae9pg0FirpepeKkFr0xfr-xfr-xb9adbaqaceGaciGaai % aabeqacmaabaabauaakeaacaWGkbGaaiikaiabeI7aXjaacMcacqGH % 9aqpdaWcaaqaaiaaigdaaeaacaaIYaGaamyBaaaadaaeWbqaaiaacI % cacaWG5bWaaWbaaSqabeaacaWGPbaaaOGaeyOeI0IaamiAamaaBaaa % leaacqaH4oqCaeqaaOGaaiikaiaadIhadaahaaWcbeqaaiaadMgaaa % GccaGGPaGaaiykamaaCaaaleqabaGaaGOmaaaaaeaacaWGPbGaeyyp % a0JaaGymaaqaaiaad2gaa0GaeyyeIuoaaaa!57E1! J(\theta ) = \frac{1}{
{2m}}\sum\limits_{i = 1}^m {
{
{({y^i} - {h_\theta }({x^i}))}^2}} J(θ)=2m1i=1∑m(yi−h
最低0.47元/天 解锁文章
1472
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
