http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&feed=atom&action=history
可视化自编码器训练结果 - Revision history
2024-03-29T10:49:37Z
Revision history for this page on the wiki
MediaWiki 1.16.2
http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=2389&oldid=prev
Kandeng: /* 中英文对照 */
2013-05-07T15:50:48Z
<p><span class="autocomment">中英文对照</span></p>
<table style="background-color: white; color:black;">
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 15:50, 7 May 2013</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 31:</td>
<td colspan="2" class="diff-lineno">Line 31:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:非线性特征 non-linear feature</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:非线性特征 non-linear feature</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:激励 activate</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:激励 activate</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div>:平凡解 <del class="diffchange diffchange-inline">non-</del>trivial answer</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div>:平凡解 trivial answer</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:范数约束 norm constrained</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:范数约束 norm constrained</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:稀疏自编码器 sparse autoencoder</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:稀疏自编码器 sparse autoencoder</div></td></tr>
</table>
Kandeng
http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=2367&oldid=prev
Kandeng: /* 中英文对照 */
2013-04-08T08:52:15Z
<p><span class="autocomment">中英文对照</span></p>
<table style="background-color: white; color:black;">
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 08:52, 8 April 2013</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 26:</td>
<td colspan="2" class="diff-lineno">Line 26:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>==中英文对照==</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>==中英文对照==</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div> </div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:可视化 Visualizing</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:自编码器 Autoencoder</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:隐藏单元 hidden unit</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:非线性特征 non-linear feature</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:激励 activate</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:平凡解 non-trivial answer</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:范数约束 norm constrained</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:稀疏自编码器 sparse autoencoder</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:有界范数 norm bounded</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:输入域 input domains</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>==中文译者==</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>==中文译者==</div></td></tr>
</table>
Kandeng
http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=2355&oldid=prev
Wikiroot at 05:34, 8 April 2013
2013-04-08T05:34:03Z
<p></p>
<table style="background-color: white; color:black;">
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 05:34, 8 April 2013</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 31:</td>
<td colspan="2" class="diff-lineno">Line 31:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>王方(fangkey@gmail.com),胡伦(hulun499@gmail.com),谢宇(msforbus@sina.com),@小琳爱肉肉(新浪微博账号), 余凯(kai.yu.cool@gmail.com)</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>王方(fangkey@gmail.com),胡伦(hulun499@gmail.com),谢宇(msforbus@sina.com),@小琳爱肉肉(新浪微博账号), 余凯(kai.yu.cool@gmail.com)</div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;">{{稀疏自编码器}}</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>{{Languages|Visualizing_a_Trained_Autoencoder|English}}</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>{{Languages|Visualizing_a_Trained_Autoencoder|English}}</div></td></tr>
</table>
Wikiroot
http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=2350&oldid=prev
Kandeng at 05:24, 8 April 2013
2013-04-08T05:24:07Z
<p></p>
<table style="background-color: white; color:black;">
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 05:24, 8 April 2013</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 22:</td>
<td colspan="2" class="diff-lineno">Line 22:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>显而易见,这些特征对物体识别等计算机视觉任务是十分有用的。若将其用于其他输入域(如音频),该算法也可学到对这些输入域有用的表示或特征。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>显而易见,这些特征对物体识别等计算机视觉任务是十分有用的。若将其用于其他输入域(如音频),该算法也可学到对这些输入域有用的表示或特征。</div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;">==中英文对照==</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td colspan="2" class="diff-lineno">Line 27:</td>
<td colspan="2" class="diff-lineno">Line 30:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>==中文译者==</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>==中文译者==</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div>王方(fangkey@gmail.com),胡伦(hulun499@gmail.<del class="diffchange diffchange-inline">com),</del>@<del class="diffchange diffchange-inline">数据鱼,</del>@<del class="diffchange diffchange-inline">小琳爱肉肉</del>, 余凯(kai.yu.cool@gmail.com)</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div>王方(fangkey@gmail.com),胡伦(hulun499@gmail.<ins class="diffchange diffchange-inline">com),谢宇(msforbus</ins>@<ins class="diffchange diffchange-inline">sina.com),</ins>@<ins class="diffchange diffchange-inline">小琳爱肉肉(新浪微博账号)</ins>, 余凯(kai.yu.cool@gmail.com)</div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div> </div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>{{Languages|Visualizing_a_Trained_Autoencoder|English}}</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>{{Languages|Visualizing_a_Trained_Autoencoder|English}}</div></td></tr>
</table>
Kandeng
http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=2270&oldid=prev
Kandeng at 12:48, 7 April 2013
2013-04-07T12:48:08Z
<p></p>
<table style="background-color: white; color:black;">
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 12:48, 7 April 2013</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 28:</td>
<td colspan="2" class="diff-lineno">Line 28:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>王方(fangkey@gmail.com),胡伦(hulun499@gmail.com),@数据鱼,@小琳爱肉肉, 余凯(kai.yu.cool@gmail.com)</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>王方(fangkey@gmail.com),胡伦(hulun499@gmail.com),@数据鱼,@小琳爱肉肉, 余凯(kai.yu.cool@gmail.com)</div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;">{{Languages|Visualizing_a_Trained_Autoencoder|English}}</ins></div></td></tr>
</table>
Kandeng
http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=2219&oldid=prev
Kandeng at 15:00, 5 April 2013
2013-04-05T15:00:37Z
<p></p>
<table style="background-color: white; color:black;">
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 15:00, 5 April 2013</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 22:</td>
<td colspan="2" class="diff-lineno">Line 22:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>显而易见,这些特征对物体识别等计算机视觉任务是十分有用的。若将其用于其他输入域(如音频),该算法也可学到对这些输入域有用的表示或特征。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>显而易见,这些特征对物体识别等计算机视觉任务是十分有用的。若将其用于其他输入域(如音频),该算法也可学到对这些输入域有用的表示或特征。</div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;">==中文译者==</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;"></ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;">王方(fangkey@gmail.com),胡伦(hulun499@gmail.com),@数据鱼,@小琳爱肉肉, 余凯(kai.yu.cool@gmail.com)</ins></div></td></tr>
</table>
Kandeng
http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=2160&oldid=prev
Kandeng at 16:08, 30 March 2013
2013-03-30T16:08:43Z
<p></p>
<a href="http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=2160&oldid=1455">Show changes</a>
Kandeng
http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=1455&oldid=prev
Kandeng at 12:02, 7 March 2013
2013-03-07T12:02:30Z
<p></p>
<table style="background-color: white; color:black;">
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 12:02, 7 March 2013</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 30:</td>
<td colspan="2" class="diff-lineno">Line 30:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>原文中的“in particular”,应为强调之意。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>原文中的“in particular”,应为强调之意。</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>原文“as some non-linear feature”中的“some”,似不译为好。非线性特征当然可以有很多,而这里计算出来的当然也只是其中一种,其意不言自明。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>原文“as some non-linear feature”中的“some”,似不译为好。非线性特征当然可以有很多,而这里计算出来的当然也只是其中一种,其意不言自明。</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">:</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div> </div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>By displaying the image formed by these pixel intensity values, we can begin to understand what feature hidden unit <math>\textstyle i</math> is looking for.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>By displaying the image formed by these pixel intensity values, we can begin to understand what feature hidden unit <math>\textstyle i</math> is looking for.</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【初译】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【初译】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">用求得的输入像素的值作为图像的亮度进行显示,我们就可以了解到隐藏单元i要寻找的特征值是什么样子了。</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">用求得的输入像素的值作为图像的亮度进行显示,我们就可以了解到隐藏单元<math>\textstyle i</math>要寻找的特征值是什么样子了。</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【一校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【一校】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">当我们用上式所得到的像素值构成我们所需要的2D图像,就可以了解到隐藏单元i要寻找的特征是什么样子了。</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">当我们用上式所得到的像素值构成我们所需要的2D图像,就可以了解到隐藏单元<math>\textstyle i</math>要寻找的特征是什么样子了。</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【二校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【二校】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">当我们用上式所得到的像素值构成我们所需要的2D图像,就可以了解到隐藏单元i要寻找的特征是什么样子了。</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">当我们用上式所得到的像素值构成我们所需要的2D图像,就可以了解到隐藏单元<math>\textstyle i</math>要寻找的特征是什么样子了。</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">当我们用上式算出各像素的值、把它们组成一幅图像、并将图像呈现在我们面前之时,隐藏单元i所追寻特征的真正含义也渐渐明朗起来。</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">当我们用上式算出各像素的值、把它们组成一幅图像、并将图像呈现在我们面前之时,隐藏单元<math>\textstyle i</math>所追寻特征的真正含义也渐渐明朗起来。</ins></div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">:</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div> </div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>If we have an autoencoder with 100 hidden units (say), then we our visualization will have 100 such images---one per hidden unit. By examining these 100 images, we can try to understand what the ensemble of hidden units is learning.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>If we have an autoencoder with 100 hidden units (say), then we our visualization will have 100 such images---one per hidden unit. By examining these 100 images, we can try to understand what the ensemble of hidden units is learning.</div></td></tr>
<tr><td colspan="2" class="diff-lineno">Line 52:</td>
<td colspan="2" class="diff-lineno">Line 52:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>假如我们训练的自编码器有100个隐藏单元,可视化结果就会包含100幅这样的图像——每个隐藏单元都对应一幅图像。审视这100幅图像,我们可以试着体会这些隐藏单元学出来的整体效果是什么样的。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>假如我们训练的自编码器有100个隐藏单元,可视化结果就会包含100幅这样的图像——每个隐藏单元都对应一幅图像。审视这100幅图像,我们可以试着体会这些隐藏单元学出来的整体效果是什么样的。</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">:</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div> </div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>When we do this for a sparse autoencoder (trained with 100 hidden units on 10x10 pixel inputs<sup>1</sup> we get the following result:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>When we do this for a sparse autoencoder (trained with 100 hidden units on 10x10 pixel inputs<sup>1</sup> we get the following result:</div></td></tr>
<tr><td colspan="2" class="diff-lineno">Line 66:</td>
<td colspan="2" class="diff-lineno">Line 66:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>“一个稀疏自编码器”中的“一个”不必译出来。当然是一个,还能是几个?</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>“一个稀疏自编码器”中的“一个”不必译出来。当然是一个,还能是几个?</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>[[Image:ExampleSparseAutoencoderWeights.png|thumb|400px|center]]</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>[[Image:ExampleSparseAutoencoderWeights.png|thumb|400px|center]]</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">:</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div> </div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>Each square in the figure above shows the (norm bounded) input image<math>\textstyle x</math>that maximally actives one of 100 hidden units. We see that the different hidden units have learned to detect edges at different positions and orientations in the image.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>Each square in the figure above shows the (norm bounded) input image<math>\textstyle x</math>that maximally actives one of 100 hidden units. We see that the different hidden units have learned to detect edges at different positions and orientations in the image.</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【初译】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【初译】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">上图中每一个方格表示使这个方格对应的隐藏单元得到最大激励的输入图像x(经过归一化)。我们可以看到,不同的隐藏单元学会了在图像的不同位置和方向想进行边缘检测。</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">上图中每一个方格表示使这个方格对应的隐藏单元得到最大激励的输入图像<math>\textstyle x</math>(经过归一化)。我们可以看到,不同的隐藏单元学会了在图像的不同位置和方向想进行边缘检测。</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【一校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【一校】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">上图中每一个方格表示使这个方格对应的隐藏单元得到最大激励的输入图像x(图像均符合我们之前的规范假设)。我们可以看到,不同的隐藏单元学会了如何在图像的不同位置和方向进行边缘检测。</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">上图中每一个方格表示使这个方格对应的隐藏单元得到最大激励的输入图像<math>\textstyle x</math>(图像均符合我们之前的规范假设)。我们可以看到,不同的隐藏单元学会了如何在图像的不同位置和方向进行边缘检测。</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【二校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【二校】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">上图中每一个方格表示使这个方格对应的隐藏单元得到最大激励的输入图像x(规范假设)。我们可以看到,不同的隐藏单元学会了如何在图像的不同位置和方向进行边缘检测。</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">上图中每一个方格表示使这个方格对应的隐藏单元得到最大激励的输入图像<math>\textstyle x</math>(规范假设)。我们可以看到,不同的隐藏单元学会了如何在图像的不同位置和方向进行边缘检测。</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div>上图的每个小方块都给出了一个(带有有界范数 <del class="diffchange diffchange-inline">的)输入图像x,它可使这100个隐藏单元中的某一个获得最大激励。我们可以看到,不同的隐藏单元学会了在图像的不同位置和方向进行边缘检测。</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div>上图的每个小方块都给出了一个(带有有界范数 <ins class="diffchange diffchange-inline">的)输入图像<math>\textstyle x</math>,它可使这100个隐藏单元中的某一个获得最大激励。我们可以看到,不同的隐藏单元学会了在图像的不同位置和方向进行边缘检测。</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校说明】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校说明】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>我想这里不宜用“对应”,因为图中小方块是10x10方阵排列的,但隐藏单元不是。怎么对应?逐行对应还是逐列?这都是未知的。因此为严谨起见,还是改为“某一个”较好。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>我想这里不宜用“对应”,因为图中小方块是10x10方阵排列的,但隐藏单元不是。怎么对应?逐行对应还是逐列?这都是未知的。因此为严谨起见,还是改为“某一个”较好。</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">:</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div> </div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>These features are, not surprisingly, useful for such tasks as object recognition and other vision tasks. When applied to other input domains (such as audio), this algorithm also learns useful representations/features for those domains too.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>These features are, not surprisingly, useful for such tasks as object recognition and other vision tasks. When applied to other input domains (such as audio), this algorithm also learns useful representations/features for those domains too.</div></td></tr>
<tr><td colspan="2" class="diff-lineno">Line 90:</td>
<td colspan="2" class="diff-lineno">Line 90:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>显而易见,这些特征对物体识别等计算机视觉任务是十分有用的。若将其用于其他输入域(如音频),该算法也可学到对这些输入域有用的表示或特征。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>显而易见,这些特征对物体识别等计算机视觉任务是十分有用的。若将其用于其他输入域(如音频),该算法也可学到对这些输入域有用的表示或特征。</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div><del class="diffchange diffchange-inline">:</del></div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div> </div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【专业术语对照表】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【专业术语对照表】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>为了在后期校对时,使前后章节专业术语翻译统一,在此将本章中专业术语翻译的中英文对照总结到下表,以便统一修改,或用于后期专业名词附录。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>为了在后期校对时,使前后章节专业术语翻译统一,在此将本章中专业术语翻译的中英文对照总结到下表,以便统一修改,或用于后期专业名词附录。</div></td></tr>
</table>
Kandeng
http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=1454&oldid=prev
Kandeng at 11:58, 7 March 2013
2013-03-07T11:58:09Z
<p></p>
<table style="background-color: white; color:black;">
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 11:58, 7 March 2013</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 30:</td>
<td colspan="2" class="diff-lineno">Line 30:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>原文中的“in particular”,应为强调之意。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>原文中的“in particular”,应为强调之意。</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>原文“as some non-linear feature”中的“some”,似不译为好。非线性特征当然可以有很多,而这里计算出来的当然也只是其中一种,其意不言自明。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>原文“as some non-linear feature”中的“some”,似不译为好。非线性特征当然可以有很多,而这里计算出来的当然也只是其中一种,其意不言自明。</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div> </div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>By displaying the image formed by these pixel intensity values, we can begin to understand what feature hidden unit <math>\textstyle i</math> is looking for.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>By displaying the image formed by these pixel intensity values, we can begin to understand what feature hidden unit <math>\textstyle i</math> is looking for.</div></td></tr>
<tr><td colspan="2" class="diff-lineno">Line 41:</td>
<td colspan="2" class="diff-lineno">Line 41:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>当我们用上式算出各像素的值、把它们组成一幅图像、并将图像呈现在我们面前之时,隐藏单元i所追寻特征的真正含义也渐渐明朗起来。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>当我们用上式算出各像素的值、把它们组成一幅图像、并将图像呈现在我们面前之时,隐藏单元i所追寻特征的真正含义也渐渐明朗起来。</div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;">:</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>If we have an autoencoder with 100 hidden units (say), then we our visualization will have 100 such images---one per hidden unit. By examining these 100 images, we can try to understand what the ensemble of hidden units is learning.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>If we have an autoencoder with 100 hidden units (say), then we our visualization will have 100 such images---one per hidden unit. By examining these 100 images, we can try to understand what the ensemble of hidden units is learning.</div></td></tr>
<tr><td colspan="2" class="diff-lineno">Line 51:</td>
<td colspan="2" class="diff-lineno">Line 52:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>假如我们训练的自编码器有100个隐藏单元,可视化结果就会包含100幅这样的图像——每个隐藏单元都对应一幅图像。审视这100幅图像,我们可以试着体会这些隐藏单元学出来的整体效果是什么样的。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>假如我们训练的自编码器有100个隐藏单元,可视化结果就会包含100幅这样的图像——每个隐藏单元都对应一幅图像。审视这100幅图像,我们可以试着体会这些隐藏单元学出来的整体效果是什么样的。</div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;">:</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>When we do this for a sparse autoencoder (trained with 100 hidden units on 10x10 pixel inputs<sup>1</sup> we get the following result:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>When we do this for a sparse autoencoder (trained with 100 hidden units on 10x10 pixel inputs<sup>1</sup> we get the following result:</div></td></tr>
<tr><td colspan="2" class="diff-lineno">Line 64:</td>
<td colspan="2" class="diff-lineno">Line 66:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>“一个稀疏自编码器”中的“一个”不必译出来。当然是一个,还能是几个?</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>“一个稀疏自编码器”中的“一个”不必译出来。当然是一个,还能是几个?</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>[[Image:ExampleSparseAutoencoderWeights.png|thumb|400px|center]]</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>[[Image:ExampleSparseAutoencoderWeights.png|thumb|400px|center]]</div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;">:</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>Each square in the figure above shows the (norm bounded) input image<math>\textstyle x</math>that maximally actives one of 100 hidden units. We see that the different hidden units have learned to detect edges at different positions and orientations in the image.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>Each square in the figure above shows the (norm bounded) input image<math>\textstyle x</math>that maximally actives one of 100 hidden units. We see that the different hidden units have learned to detect edges at different positions and orientations in the image.</div></td></tr>
<tr><td colspan="2" class="diff-lineno">Line 76:</td>
<td colspan="2" class="diff-lineno">Line 79:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校说明】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校说明】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>我想这里不宜用“对应”,因为图中小方块是10x10方阵排列的,但隐藏单元不是。怎么对应?逐行对应还是逐列?这都是未知的。因此为严谨起见,还是改为“某一个”较好。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>我想这里不宜用“对应”,因为图中小方块是10x10方阵排列的,但隐藏单元不是。怎么对应?逐行对应还是逐列?这都是未知的。因此为严谨起见,还是改为“某一个”较好。</div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;">:</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【原文】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>These features are, not surprisingly, useful for such tasks as object recognition and other vision tasks. When applied to other input domains (such as audio), this algorithm also learns useful representations/features for those domains too.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>These features are, not surprisingly, useful for such tasks as object recognition and other vision tasks. When applied to other input domains (such as audio), this algorithm also learns useful representations/features for those domains too.</div></td></tr>
<tr><td colspan="2" class="diff-lineno">Line 86:</td>
<td colspan="2" class="diff-lineno">Line 90:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>显而易见,这些特征对物体识别等计算机视觉任务是十分有用的。若将其用于其他输入域(如音频),该算法也可学到对这些输入域有用的表示或特征。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>显而易见,这些特征对物体识别等计算机视觉任务是十分有用的。若将其用于其他输入域(如音频),该算法也可学到对这些输入域有用的表示或特征。</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div> </div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">:</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【专业术语对照表】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【专业术语对照表】:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>为了在后期校对时,使前后章节专业术语翻译统一,在此将本章中专业术语翻译的中英文对照总结到下表,以便统一修改,或用于后期专业名词附录。</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>为了在后期校对时,使前后章节专业术语翻译统一,在此将本章中专业术语翻译的中英文对照总结到下表,以便统一修改,或用于后期专业名词附录。</div></td></tr>
</table>
Kandeng
http://ufldl.stanford.edu/wiki/index.php?title=%E5%8F%AF%E8%A7%86%E5%8C%96%E8%87%AA%E7%BC%96%E7%A0%81%E5%99%A8%E8%AE%AD%E7%BB%83%E7%BB%93%E6%9E%9C&diff=1453&oldid=prev
Kandeng at 11:56, 7 March 2013
2013-03-07T11:56:09Z
<p></p>
<table style="background-color: white; color:black;">
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 11:56, 7 March 2013</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 16:</td>
<td colspan="2" class="diff-lineno">Line 16:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:We will visualize the function computed by hidden unit ---which depends on the parameters <math>\textstyle W^{(1)}_{ij}</math> (ignoring the bias term for now)---using a 2D image. In particular, we think of <math>\textstyle a^{(2)}_i</math> as some non-linear feature of the input <math>\textstyle x</math>. We ask: What input image <math>\textstyle x</math> would cause <math>\textstyle a^{(2)}_i</math> to be maximally activated? (Less formally, what is the feature that hidden unit <math>\textstyle i</math> is looking for?) For this question to have a non-trivial answer, we must impose some constraints on <math>\textstyle x</math>. If we suppose that the input is norm constrained by <math>\textstyle ||x||^2 = \sum_{i=1}^{100} x_i^2 \leq 1</math>, then one can show (try doing this yourself) that the input which maximally activates hidden unit <math>\textstyle i</math> is given by setting pixel <math>\textstyle x_j</math> (for all 100 pixels, <math>\textstyle j=1,\ldots, 100</math>) to</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:We will visualize the function computed by hidden unit ---which depends on the parameters <math>\textstyle W^{(1)}_{ij}</math> (ignoring the bias term for now)---using a 2D image. In particular, we think of <math>\textstyle a^{(2)}_i</math> as some non-linear feature of the input <math>\textstyle x</math>. We ask: What input image <math>\textstyle x</math> would cause <math>\textstyle a^{(2)}_i</math> to be maximally activated? (Less formally, what is the feature that hidden unit <math>\textstyle i</math> is looking for?) For this question to have a non-trivial answer, we must impose some constraints on <math>\textstyle x</math>. If we suppose that the input is norm constrained by <math>\textstyle ||x||^2 = \sum_{i=1}^{100} x_i^2 \leq 1</math>, then one can show (try doing this yourself) that the input which maximally activates hidden unit <math>\textstyle i</math> is given by setting pixel <math>\textstyle x_j</math> (for all 100 pixels, <math>\textstyle j=1,\ldots, 100</math>) to</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【初译】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【初译】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div>我们将用2D图像对这个由隐藏单元i计算出的函数进行可视化,这个函数依赖于参数<math>\textstyle W^{(1)}_{ij}</math>(忽略掉偏置项<math>b_i<del class="diffchange diffchange-inline">\right</del></math>)。此时,如果我们将<math>\textstyle a^{(2)}_i</math>理解为输入向量<math>\textstyle x</math>的某个非线性特征值,我们需要思考:什么样的输入图像<math>\textstyle x</math>会使得激励<math>\textstyle a^{(2)}_i</math>取得最大值?(也就是说,隐藏单元<math>\textstyle i</math>找到的是一个什么样的特征值?)。因为这个问题需要有一个有实际意义的解,所以我们必须对<math>\textstyle x</math>加以限制。我们采用输入向量长度的平方<math>\textstyle ||x||^2 = \sum_{i=1}^{100} x_i^2 \leq 1</math>进行归一化限制,于是可以得到(请读者尝试自行推导。),当输入对隐藏单元<math>\textstyle i</math>产生最大的激励时,其输入像素<math>\textstyle x_j</math>(对所有100个输入像素,j=1,…,100)所取的值应为:</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div>我们将用2D图像对这个由隐藏单元i计算出的函数进行可视化,这个函数依赖于参数<math>\textstyle W^{(1)}_{ij}</math>(忽略掉偏置项<math>b_i</math>)。此时,如果我们将<math>\textstyle a^{(2)}_i</math>理解为输入向量<math>\textstyle x</math>的某个非线性特征值,我们需要思考:什么样的输入图像<math>\textstyle x</math>会使得激励<math>\textstyle a^{(2)}_i</math>取得最大值?(也就是说,隐藏单元<math>\textstyle i</math>找到的是一个什么样的特征值?)。因为这个问题需要有一个有实际意义的解,所以我们必须对<math>\textstyle x</math>加以限制。我们采用输入向量长度的平方<math>\textstyle ||x||^2 = \sum_{i=1}^{100} x_i^2 \leq 1</math>进行归一化限制,于是可以得到(请读者尝试自行推导。),当输入对隐藏单元<math>\textstyle i</math>产生最大的激励时,其输入像素<math>\textstyle x_j</math>(对所有100个输入像素,j=1,…,100)所取的值应为:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【一校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【一校】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div>接着我们将使用一个2D图像对这个由隐藏单元i负责计算的函数进行可视化,注意该函数依赖于参数集<math>\textstyle W^{(1)}_{ij}</math>(暂时忽略偏差项<math>b_i<del class="diffchange diffchange-inline">\right</del></math>)。如果再具体一些,我们可以将<math>\textstyle a^{(2)}_i</math>理解为输入向量<math>\textstyle x</math>的某个非线性特征。然后我们便想问:什么样的输入图像<math>\textstyle x</math>会最大程度上激励<math>\textstyle a^{(2)}_i</math>?(通俗一点的说法是隐藏单元<math>\textstyle i</math>需要找到的是一个什么样的特征?)。为了使这个问题有一个有实际意义的解释,我们必须对<math>\textstyle x</math>加以限制。如果假设输入向量符合<math>\textstyle ||x||^2 = \sum_{i=1}^{100} x_i^2 \leq 1</math>的范式限制,那么我们可以知道(请读者尝试自行推导。),当输入向量 对隐藏单元<math>\textstyle i</math>产生最大程度的激励时, 在2D图像中所对应的像素<math>\textstyle x_j</math>(对应的像素总共有100个,j=1,…,100)所取的值应为:</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div>接着我们将使用一个2D图像对这个由隐藏单元i负责计算的函数进行可视化,注意该函数依赖于参数集<math>\textstyle W^{(1)}_{ij}</math>(暂时忽略偏差项<math>b_i</math>)。如果再具体一些,我们可以将<math>\textstyle a^{(2)}_i</math>理解为输入向量<math>\textstyle x</math>的某个非线性特征。然后我们便想问:什么样的输入图像<math>\textstyle x</math>会最大程度上激励<math>\textstyle a^{(2)}_i</math>?(通俗一点的说法是隐藏单元<math>\textstyle i</math>需要找到的是一个什么样的特征?)。为了使这个问题有一个有实际意义的解释,我们必须对<math>\textstyle x</math>加以限制。如果假设输入向量符合<math>\textstyle ||x||^2 = \sum_{i=1}^{100} x_i^2 \leq 1</math>的范式限制,那么我们可以知道(请读者尝试自行推导。),当输入向量 对隐藏单元<math>\textstyle i</math>产生最大程度的激励时, 在2D图像中所对应的像素<math>\textstyle x_j</math>(对应的像素总共有100个,j=1,…,100)所取的值应为:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【二校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【二校】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div>接着我们使用一个2D图像对这个由隐藏单元i负责计算的函数进行可视化,该函数依赖于参数集<math>\textstyle W^{(1)}_{ij}</math>(暂时忽略偏差项<math>b_i<del class="diffchange diffchange-inline">\right</del></math>)。更具体些,我们可以将<math>\textstyle a^{(2)}_i</math>理解为输入向量<math>\textstyle x</math>的某个非线性特征。然后我们便想问:什么样的输入图像<math>\textstyle a^{(2)}_i</math>会使<math>\textstyle a^{(2)}_i</math>得到最大程度的激励?(说通俗点,隐藏单元<math>\textstyle i</math>需要找到的是一个什么样的特征?)。为了使这个问题有一个有实际意义的解释,我们必须对<math>\textstyle x</math>加以限制。如果假设输入向量符合<math>\textstyle ||x||^2 = \sum_{i=1}^{100} x_i^2 \leq 1</math>的范式限制,那么我们可以知道(请读者尝试自行推导。),当输入向量 对隐藏单元<math>\textstyle i</math>产生最大程度的激励时, 在2D图像中所对应的像素<math>\textstyle x_j</math>(对应的像素总共有100个,j=1,…,100)所取的值应为:</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div>接着我们使用一个2D图像对这个由隐藏单元i负责计算的函数进行可视化,该函数依赖于参数集<math>\textstyle W^{(1)}_{ij}</math>(暂时忽略偏差项<math>b_i</math>)。更具体些,我们可以将<math>\textstyle a^{(2)}_i</math>理解为输入向量<math>\textstyle x</math>的某个非线性特征。然后我们便想问:什么样的输入图像<math>\textstyle a^{(2)}_i</math>会使<math>\textstyle a^{(2)}_i</math>得到最大程度的激励?(说通俗点,隐藏单元<math>\textstyle i</math>需要找到的是一个什么样的特征?)。为了使这个问题有一个有实际意义的解释,我们必须对<math>\textstyle x</math>加以限制。如果假设输入向量符合<math>\textstyle ||x||^2 = \sum_{i=1}^{100} x_i^2 \leq 1</math>的范式限制,那么我们可以知道(请读者尝试自行推导。),当输入向量 对隐藏单元<math>\textstyle i</math>产生最大程度的激励时, 在2D图像中所对应的像素<math>\textstyle x_j</math>(对应的像素总共有100个,j=1,…,100)所取的值应为:</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:【三校】:</div></td></tr>
<tr><td class='diff-marker'>-</td><td style="background: #ffa; color:black; font-size: smaller;"><div>我们将要可视化的函数,就是上面这个以2D图像为输入、并由隐藏单元i计算出来的函数。它是依赖于参数<math>\textstyle W^{(1)}_{ij}</math>的(暂时忽略偏置项<math>b_i<del class="diffchange diffchange-inline">\right</del></math>)。需要注意的是,<math>\textstyle a^{(2)}_i</math>可看作输入<math>\textstyle x</math>的非线性特征。不过还有个问题:什么样的输入图像<math>\textstyle x</math>可让<math>\textstyle a^{(2)}_i</math>得到最大程度的激励?(通俗一点说,隐藏单元<math>\textstyle i</math>要找个什么样的特征?)。这里我们必须给<math>\textstyle x</math>加约束,否则会得到平凡解。若假设输入有范数约束<math>\textstyle ||x||^2 = \sum_{i=1}^{100} x_i^2 \leq 1</math>,则可证(请读者自行推导)令隐藏单元<math>\textstyle i</math>得到最大激励的输入应由下面公式计算的像素<math>\textstyle x_j</math>给出(共需计算100个像素,j=1,…,100):</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div>我们将要可视化的函数,就是上面这个以2D图像为输入、并由隐藏单元i计算出来的函数。它是依赖于参数<math>\textstyle W^{(1)}_{ij}</math>的(暂时忽略偏置项<math>b_i</math>)。需要注意的是,<math>\textstyle a^{(2)}_i</math>可看作输入<math>\textstyle x</math>的非线性特征。不过还有个问题:什么样的输入图像<math>\textstyle x</math>可让<math>\textstyle a^{(2)}_i</math>得到最大程度的激励?(通俗一点说,隐藏单元<math>\textstyle i</math>要找个什么样的特征?)。这里我们必须给<math>\textstyle x</math>加约束,否则会得到平凡解。若假设输入有范数约束<math>\textstyle ||x||^2 = \sum_{i=1}^{100} x_i^2 \leq 1</math>,则可证(请读者自行推导)令隐藏单元<math>\textstyle i</math>得到最大激励的输入应由下面公式计算的像素<math>\textstyle x_j</math>给出(共需计算100个像素,j=1,…,100):</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:<math>\begin{align}</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>:<math>\begin{align}</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>x_j = \frac{W^{(1)}_{ij}}{\sqrt{\sum_{j=1}^{100} (W^{(1)}_{ij})^2}}.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>x_j = \frac{W^{(1)}_{ij}}{\sqrt{\sum_{j=1}^{100} (W^{(1)}_{ij})^2}}.</div></td></tr>
</table>
Kandeng