Improving Deep Neural Networks學(xué)習(xí)筆記(二)

文章作者:Tyan
博客:noahsnail.com ?|? CSDN ?|? 簡書

4. Optimization algorithms

4.1 Mini-batch gradient descent

$x{\{t\}}$,$y{\{t\}}$ is used to index into different mini batches. $x{[t]}$,$y{[t]}$ is used to index into different layer. $x{(t)}$,$y{(t)}$ is used to index into different examples.

Batch gradient descent is to process entire training set at the same time. Mini-batch gradient descent is to process single mini batch $x{\{t\}}$,$y{\{t\}}$ at the same time.

Run forward propagation and back propagation once on mini batch is called one iteration.

Mini-batch gradient descent runs much faster than batch gradient descent.

4.2 Understanding mini-batch gradient descent

If mini-batch size = m, it's batch gradient descend.
If mini-batch size = 1, it's stochastic gradient descend.
In pracice, mini-batch size between 1 and m.

Batch gradient descend: too long per iteration.
Stochastic gradient descend: lose speed up from vectorization.
Mini-batch gradient descend: Faster learning, 1. vectorization 2. Make progress without needing to wait.

Choosing mini-batch size:

If small training set(m <= 2000), use batch gradient descend.
Typical mini-batch size: 64, 128, 256, 512, 1024(rare).

4.3 Exponentially weighted averages

$$V_t = \beta V_{t-1} + (1-\beta)\theta_t$$

View $V_t$ as approximately averaging over $\frac {1} {1 - \beta}$.

It's called moving average in the statistics literature.

$\beta = 0.9$:

Figure 1

$\beta = 0.9(red)$,$\beta = 0.98(green)$,$\beta = 0.5(yellow)$:

Figure 2

4.4 Understanding exponentially weighted averages

$\theta$ is the temperature of the day.

$$v_{100} = 0.9v_{99} + 0.1 \theta_{100}$$$$v_{99} = 0.9v_{98} + 0.1 \theta_{99}$$$$...$$

So $$v_{100} = 0.1 * \theta _{100} + 0.1 * 0.9 * \theta _{99} + ... + 0.1 * 0.9^{i} * \theta _{100-i} + ...$$

Th coefficients is $$0.1 + 0.1 * 0.9 + 0.1 * 0.9^2 + ...$$

All of these coefficients, add up to one or add up to very close to one. It is called bias correction.

$$(1 - \epsilon)^{\frac {1} {\epsilon}} \approx \frac {1} {e}$$ $$\frac {1} {e} \approx 0.3679$$

Implement exponentially weighted average:

$$v_0 = 0$$$$v_1 = \beta v_0 + (1- \beta) \theta _1$$$$v_2 = \beta v_1 + (1- \beta) \theta _2$$$$...$$

Exponentially weighted average takes very low memory.

4.5 Bias correction in exponentially weighted averages

It's not a very good estimate of the first several day's temperature. Bias correction is used to mofity this estimate that makes it much better. The formula is: $$\frac {v_t} {1 - \beta^t} = \beta v_{t-1} + (1- \beta) \theta _t.$$

4.6 Gradient descent with momentum

Gradient descent with momentum almost always works faster than the standard gradient descent algorithm. The basic idea is to compute an exponentially weighted average of gradients, and then use that gradient to update weights instead.

On iteration t:

  1. compute $dw$, db on current mini-batch.
  2. compute $v_{dw}$, $v_{db}$
    $$v_{dw} = \beta v_{dw} + (1 - \beta)dw$$$$v_{db} = \beta v_{db} + (1 - \beta)db$$
  3. update dw, db
    $$w = w - \alpha v_{dw}$$$$b = b - \alpha v_{db}$$

There are two hyperparameters, the most common value for $\beta$ is 0.9.

Another formula is $v_{dw} = \beta v_{dw} + dw$, you need to modify corresponding $\alpha$.

4.7 RMSprop

RMSprop stands for root mean square prop, that can also speed up gradient descent.

On iteration t:

  1. compute $dw$, db on current mini-batch.
  2. compute $s_{dw}$, $s_{db}$
    $$s_{dw} = \beta s_{dw} + (1 - \beta){dw}^2$$$$s_{db} = \beta s_{db} + (1 - \beta){db}^2$$
  3. update dw, db
    $$w = w - \alpha \frac {dw} {\sqrt {s_{dw}}}$$$$b = b - \alpha \frac {db} {\sqrt {s_{db}}}$$

In practice, in order to avoid $\sqrt {s_{dw}}$ being very close zero:

$$w = w - \alpha \frac {dw} {\sqrt {s_{dw}} + \epsilon}$$$$b = b - \alpha \frac {db} {\sqrt {s_{db}} + \epsilon}$$

Usually $$\epsilon = 10^{-8}$$

4.8 Adam optimization algorithm

$$v_{dw}=0, s_{dw}=0,v_{db},s_{db}=0$$

On iteration t:

$$v_{dw} = \beta_1 v_{dw} + (1 - \beta_1)dw$$$$v_{db} = \beta_1 v_{db} + (1 - \beta_1)db$$

$$s_{dw} = \beta_2 s_{dw} + (1 - \beta_2){dw}^2$$$$s_{db} = \beta_2 s_{db} + (1 - \beta_2){db}^2$$

Bias correction:

$$v_{dw}^{bc} = \frac {v_{dw}} {1 - \beta_1^t}, v_{db}^{bc} = \frac {v_{db}} {1 - \beta_1t}$$$$s_{dw}{bc} = \frac {s_{dw}} {1 - \beta_2^t}, s_{db}^{bc} = \frac {s_{db}} {1 - \beta_2^t}$$

Update weight:

$$w = w - \alpha \frac {v_{dw}^{bc}} {\sqrt {s_{dw}^{bc}} + \epsilon}$$$$b = b - \alpha \frac {v_{db}^{bc}} {\sqrt {s_{db}^{bc}} + \epsilon}$$

Adam combines the effect of gradient descent with momentum together with gradient descent with RMSprop. It's a commonly used learning algorithm that is proven to be very effective for many different neural networks of a very wide variety of architectures.\

$\alpha$ needs to be tuned. $\beta_1 = 0.9$, $\beta_2 = 0.999$, $\epsilon = 10^{-8}$.

Adam stands for Adaptive Moment Estimation.

4.9 Learning rate decay

Learning rate decay is slowly reduce the learning rate.

$$\alpha = \frac {1} {1 + {decay rate} * epochs} \alpha_0$$

$\alpha_0$ is the initial learning rate.

Other learning rate decay methods:

$\alpha = 0.95^{epochs}\alpha_0$, this is called exponentially decay.

$\alpha = \frac {k} {\sqrt {epochs} } \alpha_0$, $\alpha = \frac {k} {\sqrt t} \alpha_0$.

$\alpha = {\frac {1} {2}}^{epochs} \alpha _0$, this is called a discrete staircase.

4.10 The problem of local optima

In very high-dimensional spaces you're actually much more likely to run into a saddle point, rather than local optimum.

  • Unlikely to get stuck in a bad local optima.
  • Plateaus can make learning slow.
最后編輯于
?著作權(quán)歸作者所有,轉(zhuǎn)載或內(nèi)容合作請(qǐng)聯(lián)系作者
平臺(tái)聲明:文章內(nèi)容(如有圖片或視頻亦包括在內(nèi))由作者上傳并發(fā)布,文章內(nèi)容僅代表作者本人觀點(diǎn),簡書系信息發(fā)布平臺(tái),僅提供信息存儲(chǔ)服務(wù)。
  • 序言:七十年代末,一起剝皮案震驚了整個(gè)濱河市,隨后出現(xiàn)的幾起案子,更是在濱河造成了極大的恐慌,老刑警劉巖,帶你破解...
    沈念sama閱讀 230,247評(píng)論 6 543
  • 序言:濱河連續(xù)發(fā)生了三起死亡事件,死亡現(xiàn)場離奇詭異,居然都是意外死亡,警方通過查閱死者的電腦和手機(jī),發(fā)現(xiàn)死者居然都...
    沈念sama閱讀 99,520評(píng)論 3 429
  • 文/潘曉璐 我一進(jìn)店門,熙熙樓的掌柜王于貴愁眉苦臉地迎上來,“玉大人,你說我怎么就攤上這事。” “怎么了?”我有些...
    開封第一講書人閱讀 178,362評(píng)論 0 383
  • 文/不壞的土叔 我叫張陵,是天一觀的道長。 經(jīng)常有香客問我,道長,這世上最難降的妖魔是什么? 我笑而不...
    開封第一講書人閱讀 63,805評(píng)論 1 317
  • 正文 為了忘掉前任,我火速辦了婚禮,結(jié)果婚禮上,老公的妹妹穿的比我還像新娘。我一直安慰自己,他們只是感情好,可當(dāng)我...
    茶點(diǎn)故事閱讀 72,541評(píng)論 6 412
  • 文/花漫 我一把揭開白布。 她就那樣靜靜地躺著,像睡著了一般。 火紅的嫁衣襯著肌膚如雪。 梳的紋絲不亂的頭發(fā)上,一...
    開封第一講書人閱讀 55,896評(píng)論 1 328
  • 那天,我揣著相機(jī)與錄音,去河邊找鬼。 笑死,一個(gè)胖子當(dāng)著我的面吹牛,可吹牛的內(nèi)容都是我干的。 我是一名探鬼主播,決...
    沈念sama閱讀 43,887評(píng)論 3 447
  • 文/蒼蘭香墨 我猛地睜開眼,長吁一口氣:“原來是場噩夢(mèng)啊……” “哼!你這毒婦竟也來了?” 一聲冷哼從身側(cè)響起,我...
    開封第一講書人閱讀 43,062評(píng)論 0 290
  • 序言:老撾萬榮一對(duì)情侶失蹤,失蹤者是張志新(化名)和其女友劉穎,沒想到半個(gè)月后,有當(dāng)?shù)厝嗽跇淞掷锇l(fā)現(xiàn)了一具尸體,經(jīng)...
    沈念sama閱讀 49,608評(píng)論 1 336
  • 正文 獨(dú)居荒郊野嶺守林人離奇死亡,尸身上長有42處帶血的膿包…… 初始之章·張勛 以下內(nèi)容為張勛視角 年9月15日...
    茶點(diǎn)故事閱讀 41,356評(píng)論 3 358
  • 正文 我和宋清朗相戀三年,在試婚紗的時(shí)候發(fā)現(xiàn)自己被綠了。 大學(xué)時(shí)的朋友給我發(fā)了我未婚夫和他白月光在一起吃飯的照片。...
    茶點(diǎn)故事閱讀 43,555評(píng)論 1 374
  • 序言:一個(gè)原本活蹦亂跳的男人離奇死亡,死狀恐怖,靈堂內(nèi)的尸體忽然破棺而出,到底是詐尸還是另有隱情,我是刑警寧澤,帶...
    沈念sama閱讀 39,077評(píng)論 5 364
  • 正文 年R本政府宣布,位于F島的核電站,受9級(jí)特大地震影響,放射性物質(zhì)發(fā)生泄漏。R本人自食惡果不足惜,卻給世界環(huán)境...
    茶點(diǎn)故事閱讀 44,769評(píng)論 3 349
  • 文/蒙蒙 一、第九天 我趴在偏房一處隱蔽的房頂上張望。 院中可真熱鬧,春花似錦、人聲如沸。這莊子的主人今日做“春日...
    開封第一講書人閱讀 35,175評(píng)論 0 28
  • 文/蒼蘭香墨 我抬頭看了看天上的太陽。三九已至,卻和暖如春,著一層夾襖步出監(jiān)牢的瞬間,已是汗流浹背。 一陣腳步聲響...
    開封第一講書人閱讀 36,489評(píng)論 1 295
  • 我被黑心中介騙來泰國打工, 沒想到剛下飛機(jī)就差點(diǎn)兒被人妖公主榨干…… 1. 我叫王不留,地道東北人。 一個(gè)月前我還...
    沈念sama閱讀 52,289評(píng)論 3 400
  • 正文 我出身青樓,卻偏偏與公主長得像,于是被迫代替她去往敵國和親。 傳聞我的和親對(duì)象是個(gè)殘疾皇子,可洞房花燭夜當(dāng)晚...
    茶點(diǎn)故事閱讀 48,516評(píng)論 2 379

推薦閱讀更多精彩內(nèi)容