<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:wfw="http://wellformedweb.org/CommentAPI/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	xmlns:slash="http://purl.org/rss/1.0/modules/slash/"
	
	xmlns:georss="http://www.georss.org/georss"
	xmlns:geo="http://www.w3.org/2003/01/geo/wgs84_pos#"
	>

<channel>
	<title>Machine Learning &#8211; BrilliantCode.net</title>
	<atom:link href="https://www.brilliantcode.net/tag/machine-learning/feed/" rel="self" type="application/rss+xml" />
	<link>https://www.brilliantcode.net</link>
	<description></description>
	<lastBuildDate>Mon, 25 May 2020 12:12:34 +0000</lastBuildDate>
	<language>zh-TW</language>
	<sy:updatePeriod>
	hourly	</sy:updatePeriod>
	<sy:updateFrequency>
	1	</sy:updateFrequency>
	<generator>https://wordpress.org/?v=5.2.3</generator>

<image>
	<url>https://i0.wp.com/www.brilliantcode.net/wp-content/uploads/2017/06/cropped-icon-2.png?fit=32%2C32&#038;ssl=1</url>
	<title>Machine Learning &#8211; BrilliantCode.net</title>
	<link>https://www.brilliantcode.net</link>
	<width>32</width>
	<height>32</height>
</image> 
<site xmlns="com-wordpress:feed-additions:1">130881408</site>	<item>
		<title>Convolutional Neural Networks(CNN) #6 Pooling in Backward pass</title>
		<link>https://www.brilliantcode.net/1781/convolutional-neural-networks-6-backpropagation-in-pooling-layers-of-cnns/</link>
				<comments>https://www.brilliantcode.net/1781/convolutional-neural-networks-6-backpropagation-in-pooling-layers-of-cnns/#respond</comments>
				<pubDate>Thu, 05 Sep 2019 10:48:31 +0000</pubDate>
		<dc:creator><![CDATA[Andy Wang]]></dc:creator>
				<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[教學]]></category>
		<category><![CDATA[CNN]]></category>
		<category><![CDATA[Computer Vision]]></category>
		<category><![CDATA[Neural Network]]></category>

		<guid isPermaLink="false">https://www.brilliantcode.net/?p=1781</guid>
				<description><![CDATA[<p>本文將介紹Pooling layer在反向傳遞(Backward propagation / Backward pass)的運作過程，雖然Pooling層的參數不需要被訓練，但是在大多數情況下，Pooling layer通常是承接著啟動函數的輸出。因此，本文將會詳細介紹反向傳遞時Kernel、Bias、Feature map的細節。</p>
<p>這篇文章 <a rel="nofollow" href="https://www.brilliantcode.net/1781/convolutional-neural-networks-6-backpropagation-in-pooling-layers-of-cnns/">Convolutional Neural Networks(CNN) #6 Pooling in Backward pass</a> 最早出現於 <a rel="nofollow" href="https://www.brilliantcode.net">BrilliantCode.net</a>。</p>
]]></description>
						<wfw:commentRss>https://www.brilliantcode.net/1781/convolutional-neural-networks-6-backpropagation-in-pooling-layers-of-cnns/feed/</wfw:commentRss>
		<slash:comments>0</slash:comments>
						<post-id xmlns="com-wordpress:feed-additions:1">1781</post-id>	</item>
		<item>
		<title>Convolutional Neural Networks(CNN) #5 特徵圖&#038;偏差值的導數</title>
		<link>https://www.brilliantcode.net/1748/convolutional-neural-networks-5-backpropagation-in-feature-maps-biases-of-cnns/</link>
				<comments>https://www.brilliantcode.net/1748/convolutional-neural-networks-5-backpropagation-in-feature-maps-biases-of-cnns/#respond</comments>
				<pubDate>Wed, 04 Sep 2019 16:05:06 +0000</pubDate>
		<dc:creator><![CDATA[Andy Wang]]></dc:creator>
				<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[教學]]></category>
		<category><![CDATA[CNN]]></category>
		<category><![CDATA[Computer Vision]]></category>
		<category><![CDATA[Neural Network]]></category>

		<guid isPermaLink="false">https://www.brilliantcode.net/?p=1748</guid>
				<description><![CDATA[<p>本篇文章要來介紹對特徵圖（Feature map）與偏差值（Bias）在Backward propagation的推導過程與可程式化的計算方法。由於卷積神經網路（Convolutional Neural Network, CNN）在使用倒傳遞法（Backpropagation）的反向傳遞（Backward pass）過程中，就是藉著計算參數之梯度（偏微分）作為更新並訓練參數的手段，因此了解這些原理對於初探卷積神經網路的學習者來說也非常重要。</p>
<p>這篇文章 <a rel="nofollow" href="https://www.brilliantcode.net/1748/convolutional-neural-networks-5-backpropagation-in-feature-maps-biases-of-cnns/">Convolutional Neural Networks(CNN) #5 特徵圖&#038;偏差值的導數</a> 最早出現於 <a rel="nofollow" href="https://www.brilliantcode.net">BrilliantCode.net</a>。</p>
]]></description>
						<wfw:commentRss>https://www.brilliantcode.net/1748/convolutional-neural-networks-5-backpropagation-in-feature-maps-biases-of-cnns/feed/</wfw:commentRss>
		<slash:comments>0</slash:comments>
						<post-id xmlns="com-wordpress:feed-additions:1">1748</post-id>	</item>
		<item>
		<title>Convolutional Neural Networks(CNN) #4 卷積核的Back propagation</title>
		<link>https://www.brilliantcode.net/1670/convolutional-neural-networks-4-backpropagation-in-kernels-of-cnns/</link>
				<comments>https://www.brilliantcode.net/1670/convolutional-neural-networks-4-backpropagation-in-kernels-of-cnns/#respond</comments>
				<pubDate>Tue, 03 Sep 2019 12:46:31 +0000</pubDate>
		<dc:creator><![CDATA[Andy Wang]]></dc:creator>
				<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[教學]]></category>
		<category><![CDATA[CNN]]></category>
		<category><![CDATA[Computer Vision]]></category>
		<category><![CDATA[Neural Network]]></category>

		<guid isPermaLink="false">https://www.brilliantcode.net/?p=1670</guid>
				<description><![CDATA[<p>卷積神經網路的參數最終仍會以Backpropagation（倒傳遞法）來優化，本文將由淺而深的從原理到可程式化的計算方法向各位介紹Kernel的偏微分計算方法。而本文採用的範例是包含padding的卷積層，這種設定也將更趨近於現實，希望透過這種方式讓各位完全理解。</p>
<p>這篇文章 <a rel="nofollow" href="https://www.brilliantcode.net/1670/convolutional-neural-networks-4-backpropagation-in-kernels-of-cnns/">Convolutional Neural Networks(CNN) #4 卷積核的Back propagation</a> 最早出現於 <a rel="nofollow" href="https://www.brilliantcode.net">BrilliantCode.net</a>。</p>
]]></description>
						<wfw:commentRss>https://www.brilliantcode.net/1670/convolutional-neural-networks-4-backpropagation-in-kernels-of-cnns/feed/</wfw:commentRss>
		<slash:comments>0</slash:comments>
						<post-id xmlns="com-wordpress:feed-additions:1">1670</post-id>	</item>
		<item>
		<title>Convolutional Neural Networks(CNN) #3 計算參數量</title>
		<link>https://www.brilliantcode.net/1646/convolutional-neural-networks-3-calculate-number-of-parameters/</link>
				<comments>https://www.brilliantcode.net/1646/convolutional-neural-networks-3-calculate-number-of-parameters/#respond</comments>
				<pubDate>Wed, 28 Aug 2019 12:25:48 +0000</pubDate>
		<dc:creator><![CDATA[Andy Wang]]></dc:creator>
				<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[教學]]></category>
		<category><![CDATA[CNN]]></category>
		<category><![CDATA[Computer Vision]]></category>
		<category><![CDATA[Neural Network]]></category>

		<guid isPermaLink="false">https://www.brilliantcode.net/?p=1646</guid>
				<description><![CDATA[<p>因為卷積神經網路（CNN）的特性，使得在使用CNN進行影像辨識時可以大幅度的降低參數量。本節介紹的主題就是計算CNN的參數量，但是在開始之前，會先用簡單的例子帶各位了解如何架構CNN，希望各位能徹底明白卷積神經網路的運作流程，藉此深入了解計算參數量的方法。</p>
<p>這篇文章 <a rel="nofollow" href="https://www.brilliantcode.net/1646/convolutional-neural-networks-3-calculate-number-of-parameters/">Convolutional Neural Networks(CNN) #3 計算參數量</a> 最早出現於 <a rel="nofollow" href="https://www.brilliantcode.net">BrilliantCode.net</a>。</p>
]]></description>
						<wfw:commentRss>https://www.brilliantcode.net/1646/convolutional-neural-networks-3-calculate-number-of-parameters/feed/</wfw:commentRss>
		<slash:comments>0</slash:comments>
						<post-id xmlns="com-wordpress:feed-additions:1">1646</post-id>	</item>
		<item>
		<title>Convolutional Neural Networks(CNN) #2 池化層(Pooling layer)</title>
		<link>https://www.brilliantcode.net/1586/convolutional-neural-networks-2-pooling-layer/</link>
				<comments>https://www.brilliantcode.net/1586/convolutional-neural-networks-2-pooling-layer/#comments</comments>
				<pubDate>Mon, 26 Aug 2019 05:53:07 +0000</pubDate>
		<dc:creator><![CDATA[Andy Wang]]></dc:creator>
				<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[教學]]></category>
		<category><![CDATA[CNN]]></category>
		<category><![CDATA[Computer Vision]]></category>
		<category><![CDATA[Neural Network]]></category>

		<guid isPermaLink="false">https://www.brilliantcode.net/?p=1586</guid>
				<description><![CDATA[<p>池化層在卷積類神經網路扮演的角色也很關鍵，它可以幫助我們縮小Feature map的大小，也可以用來強CNN萃取出來的特徵。本篇文章就是要介紹池化層（Pooling layer）的運算規則。池化層的概念很簡單，但它仍有許多需要注意的屬性，像是與卷積層會用到的『移動步伐Stride』在池化層也會派上用場。</p>
<p>這篇文章 <a rel="nofollow" href="https://www.brilliantcode.net/1586/convolutional-neural-networks-2-pooling-layer/">Convolutional Neural Networks(CNN) #2 池化層(Pooling layer)</a> 最早出現於 <a rel="nofollow" href="https://www.brilliantcode.net">BrilliantCode.net</a>。</p>
]]></description>
						<wfw:commentRss>https://www.brilliantcode.net/1586/convolutional-neural-networks-2-pooling-layer/feed/</wfw:commentRss>
		<slash:comments>3</slash:comments>
						<post-id xmlns="com-wordpress:feed-additions:1">1586</post-id>	</item>
		<item>
		<title>Convolutional Neural Networks(CNN) #1 Kernel, Stride, Padding</title>
		<link>https://www.brilliantcode.net/1584/convolutional-neural-networks-1-convolution-layer-stride-padding-kernel/</link>
				<comments>https://www.brilliantcode.net/1584/convolutional-neural-networks-1-convolution-layer-stride-padding-kernel/#respond</comments>
				<pubDate>Sun, 25 Aug 2019 16:18:16 +0000</pubDate>
		<dc:creator><![CDATA[Andy Wang]]></dc:creator>
				<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[教學]]></category>
		<category><![CDATA[CNN]]></category>
		<category><![CDATA[Computer Vision]]></category>
		<category><![CDATA[Neural Network]]></category>

		<guid isPermaLink="false">https://www.brilliantcode.net/?p=1584</guid>
				<description><![CDATA[<p>卷積神經網路（Convolutional Neural Network, CNN）為目前用來進行影像辨識最有效的特徵萃取演算法，這個方法是由學者Yann LeCun於1998年發表的論文『Gradient-Based Learning Applied to Document Recognition』所使用的方法。直至今日，已有許多物體分類（Object classification）與物體偵測（Object detection）的方法就是透過CNN建構。本篇主要介紹CNN演算法中的卷積層運算方式以及相關屬性，其中包括移動步伐（Stride）、補充像素（Padding）和最重要的卷積核（Kernel or Filter）。</p>
<p>這篇文章 <a rel="nofollow" href="https://www.brilliantcode.net/1584/convolutional-neural-networks-1-convolution-layer-stride-padding-kernel/">Convolutional Neural Networks(CNN) #1 Kernel, Stride, Padding</a> 最早出現於 <a rel="nofollow" href="https://www.brilliantcode.net">BrilliantCode.net</a>。</p>
]]></description>
						<wfw:commentRss>https://www.brilliantcode.net/1584/convolutional-neural-networks-1-convolution-layer-stride-padding-kernel/feed/</wfw:commentRss>
		<slash:comments>0</slash:comments>
						<post-id xmlns="com-wordpress:feed-additions:1">1584</post-id>	</item>
		<item>
		<title>Backpropagation(BP) 倒傳遞法 #3 貓貓分類器-N層類神經網路</title>
		<link>https://www.brilliantcode.net/1527/backpropagation-3-n-layer-neural-networks/</link>
				<comments>https://www.brilliantcode.net/1527/backpropagation-3-n-layer-neural-networks/#respond</comments>
				<pubDate>Thu, 25 Apr 2019 05:42:48 +0000</pubDate>
		<dc:creator><![CDATA[Andy Wang]]></dc:creator>
				<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[教學]]></category>
		<category><![CDATA[Backpropagation]]></category>
		<category><![CDATA[Gradient Descent]]></category>
		<category><![CDATA[Logistic Regression]]></category>
		<category><![CDATA[Neural Network]]></category>
		<category><![CDATA[Optimization Algorithm]]></category>

		<guid isPermaLink="false">https://www.brilliantcode.net/?p=1527</guid>
				<description><![CDATA[<p>本文會以上篇內容（2層類神經網路）為基礎加深難度與實用性，因此這次將會實作可自定層數的類神經網路以及使用倒傳遞法（Back propagation, BP）優化的方法。這次的模型作法也是使用邏輯回歸（Logistic Regression）建立貓貓分類器。當然，你想換成別種圖片也是OK的。</p>
<p>這篇文章 <a rel="nofollow" href="https://www.brilliantcode.net/1527/backpropagation-3-n-layer-neural-networks/">Backpropagation(BP) 倒傳遞法 #3 貓貓分類器-N層類神經網路</a> 最早出現於 <a rel="nofollow" href="https://www.brilliantcode.net">BrilliantCode.net</a>。</p>
]]></description>
						<wfw:commentRss>https://www.brilliantcode.net/1527/backpropagation-3-n-layer-neural-networks/feed/</wfw:commentRss>
		<slash:comments>0</slash:comments>
						<post-id xmlns="com-wordpress:feed-additions:1">1527</post-id>	</item>
		<item>
		<title>Backpropagation(BP) 倒傳遞法 #2 貓貓分類器-2層類神經網路</title>
		<link>https://www.brilliantcode.net/1381/backpropagation-2-forward-pass-backward-pass/</link>
				<comments>https://www.brilliantcode.net/1381/backpropagation-2-forward-pass-backward-pass/#comments</comments>
				<pubDate>Thu, 21 Feb 2019 14:30:20 +0000</pubDate>
		<dc:creator><![CDATA[Andy Wang]]></dc:creator>
				<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[教學]]></category>
		<category><![CDATA[Backpropagation]]></category>
		<category><![CDATA[Gradient Descent]]></category>
		<category><![CDATA[Logistic Regression]]></category>
		<category><![CDATA[Neural Network]]></category>
		<category><![CDATA[Optimization Algorithm]]></category>

		<guid isPermaLink="false">https://www.brilliantcode.net/?p=1381</guid>
				<description><![CDATA[<p>本篇會介紹在機器學習（machine learning）與深度學習（deep learning）領域裡很流行的倒傳遞法（Back Propagation/ Backpropagation, BP）的演算法流程與實作方法：正向傳遞（Forward pass）、反向傳遞（Backward pass）、邏輯回歸（Logistic regression）<br />
除此之外，本篇會用簡易的2層類神經網路建立一個『貓貓分類器』。</p>
<p>這篇文章 <a rel="nofollow" href="https://www.brilliantcode.net/1381/backpropagation-2-forward-pass-backward-pass/">Backpropagation(BP) 倒傳遞法 #2 貓貓分類器-2層類神經網路</a> 最早出現於 <a rel="nofollow" href="https://www.brilliantcode.net">BrilliantCode.net</a>。</p>
]]></description>
						<wfw:commentRss>https://www.brilliantcode.net/1381/backpropagation-2-forward-pass-backward-pass/feed/</wfw:commentRss>
		<slash:comments>2</slash:comments>
						<post-id xmlns="com-wordpress:feed-additions:1">1381</post-id>	</item>
		<item>
		<title>Backpropagation(BP) 倒傳遞法 #1 工作原理與說明</title>
		<link>https://www.brilliantcode.net/1326/backpropagation-1-gradient-descent-chain-rule/</link>
				<comments>https://www.brilliantcode.net/1326/backpropagation-1-gradient-descent-chain-rule/#comments</comments>
				<pubDate>Thu, 21 Feb 2019 14:04:15 +0000</pubDate>
		<dc:creator><![CDATA[Andy Wang]]></dc:creator>
				<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[教學]]></category>
		<category><![CDATA[Backpropagation]]></category>
		<category><![CDATA[Chain Rule]]></category>
		<category><![CDATA[Gradient Descent]]></category>
		<category><![CDATA[Optimization Algorithm]]></category>

		<guid isPermaLink="false">https://www.brilliantcode.net/?p=1326</guid>
				<description><![CDATA[<p>本篇會介紹在機器學習（machine learning）與深度學習（deep learning）領域裡很流行的倒傳遞法（Back Propagation/ Backpropagation, BP）的精髓：梯度下降法（Gradient Descent）、連鎖率（Chain Rule）</p>
<p>這篇文章 <a rel="nofollow" href="https://www.brilliantcode.net/1326/backpropagation-1-gradient-descent-chain-rule/">Backpropagation(BP) 倒傳遞法 #1 工作原理與說明</a> 最早出現於 <a rel="nofollow" href="https://www.brilliantcode.net">BrilliantCode.net</a>。</p>
]]></description>
						<wfw:commentRss>https://www.brilliantcode.net/1326/backpropagation-1-gradient-descent-chain-rule/feed/</wfw:commentRss>
		<slash:comments>1</slash:comments>
						<post-id xmlns="com-wordpress:feed-additions:1">1326</post-id>	</item>
		<item>
		<title>使用 Ubuntu 作為 深度學習/機器學習/人工智慧之 平台</title>
		<link>https://www.brilliantcode.net/1012/setup-ubuntu-as-a-platform-of-deep-learning-using-cuda-toolkit-9-0-cudnn-anaconda-3-python-3-5-tensorflow/</link>
				<comments>https://www.brilliantcode.net/1012/setup-ubuntu-as-a-platform-of-deep-learning-using-cuda-toolkit-9-0-cudnn-anaconda-3-python-3-5-tensorflow/#comments</comments>
				<pubDate>Sat, 24 Mar 2018 19:07:15 +0000</pubDate>
		<dc:creator><![CDATA[Andy Wang]]></dc:creator>
				<category><![CDATA[Python]]></category>
		<category><![CDATA[Ubuntu]]></category>
		<category><![CDATA[教學]]></category>
		<category><![CDATA[AI]]></category>
		<category><![CDATA[Anaconda]]></category>
		<category><![CDATA[CUDA]]></category>
		<category><![CDATA[Deep Learning]]></category>
		<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[Tensorflow]]></category>

		<guid isPermaLink="false">https://www.brilliantcode.net/?p=1012</guid>
				<description><![CDATA[<p>人工智慧相關的新興職業將會在未來五到十年內爆炸式增加，尤其一個更是未來不可或缺的職業——“AI訓練師”。(正面迎擊人機合作的時代：AI時代3大關鍵人才)<br />
小弟我是不太敢確定AI訓練師這個職業會不會在台灣火熱地成長，不過，很確定的是以全球來說，至少中國、歐美、新加坡絕對是如此！<br />
如此明顯的大趨勢之下，還不來學學這些新東西嗎？</p>
<p>本篇文章之目的為詳列系統環境建置的步驟、相關文章以及相關troubleshooting！<br />
【環境建置】：Ubuntu 17.10 + CUDA Toolkit 9.0 + cuDNN + Anaconda 3 + Python 3.5 + Tensorflow<br />
【相關文章】：Python基礎教學</p>
<p>這篇文章 <a rel="nofollow" href="https://www.brilliantcode.net/1012/setup-ubuntu-as-a-platform-of-deep-learning-using-cuda-toolkit-9-0-cudnn-anaconda-3-python-3-5-tensorflow/">使用 Ubuntu 作為 深度學習/機器學習/人工智慧之 平台</a> 最早出現於 <a rel="nofollow" href="https://www.brilliantcode.net">BrilliantCode.net</a>。</p>
]]></description>
						<wfw:commentRss>https://www.brilliantcode.net/1012/setup-ubuntu-as-a-platform-of-deep-learning-using-cuda-toolkit-9-0-cudnn-anaconda-3-python-3-5-tensorflow/feed/</wfw:commentRss>
		<slash:comments>2</slash:comments>
						<post-id xmlns="com-wordpress:feed-additions:1">1012</post-id>	</item>
	</channel>
</rss>
