<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[Abner98414的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/Abner98414</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; Abner98414]]></copyright><item><title><![CDATA[Lecture 22 Ethics]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131133466</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131133466</guid><author>Abner98414</author><pubDate>Fri, 09 Jun 2023 19:10:11 +0800</pubDate><description><![CDATA[How we ought to live — Socrates]]></description><category></category></item><item><title><![CDATA[Lecture 21 Summarisation]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131133381</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131133381</guid><author>Abner98414</author><pubDate>Fri, 09 Jun 2023 19:07:06 +0800</pubDate><description><![CDATA[XSum。]]></description><category></category></item><item><title><![CDATA[Lecture 20 Topic Modelling]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131133248</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131133248</guid><author>Abner98414</author><pubDate>Fri, 09 Jun 2023 18:58:02 +0800</pubDate><description><![CDATA[issuesUVlearningti​∣wd∝Pti​∣wPti​∣d(Pti​∣w→ti​∣d→t1​∣wdPt1​∣mouse×Pt1​∣d1​0.010.012.010.01​×1.11.12.11.1​ogPw1​w2​...wm​log∑j0T​Pw1​∣tj​Ptj​∣dw1​​。]]></description><category></category></item><item><title><![CDATA[Lecture 19 Question Answering]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131133141</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131133141</guid><author>Abner98414</author><pubDate>Fri, 09 Jun 2023 18:46:22 +0800</pubDate><description><![CDATA[animalsAdam Bandtfivehow?datasetCompute:Pstart​i)Pend​iq1​...pm​start​i∝exppi​Ws​q)Pend​i∝exppi​We​qstart​i∝expSTTi′​)pend​i∝expETTi′​。]]></description><category></category></item><item><title><![CDATA[Lecture 18 Information Extraction]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131133075</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131133075</guid><author>Abner98414</author><pubDate>Fri, 09 Jun 2023 18:32:38 +0800</pubDate><description><![CDATA[ORGORGPER] said.B-ORGbeginningof anORGentity.I-ORG[TIME ][TIME ]]]></description><category></category></item><item><title><![CDATA[Lecture 17 Machine Translation]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131132904</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131132904</guid><author>Abner98414</author><pubDate>Fri, 09 Jun 2023 18:27:17 +0800</pubDate><description><![CDATA[feargmaxe​Pe∣frgmaxe​Pf∣ePergmaxe​Pf∣ePe)Pe)Pf∣ef∣e)alignmentfa∣ecomplexityxPy∣xPy1​∣xPy2​∣y1​x...Pyt​∣y1​...yt−1​xtrainingy∣xy∣x)OVnvariantsstT​hi​stT。]]></description><category></category></item><item><title><![CDATA[Lecture 16 Dependency Grammar]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131129263</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131129263</guid><author>Abner98414</author><pubDate>Fri, 09 Jun 2023 15:53:36 +0800</pubDate><description><![CDATA[Lecture 16 Dependency Grammar。]]></description><category></category></item><item><title><![CDATA[Lecture 15 Probabilistic Context-Free Grammar]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131129054</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131129054</guid><author>Abner98414</author><pubDate>Fri, 09 Jun 2023 15:48:50 +0800</pubDate><description><![CDATA[Lecture 15 Probabilistic Context-Free Grammar。]]></description><category></category></item><item><title><![CDATA[Lecture 14 Context-Free Grammar]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131128494</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131128494</guid><author>Abner98414</author><pubDate>Fri, 09 Jun 2023 15:32:25 +0800</pubDate><description><![CDATA[Symbols:Productions:Start symbol: Context-Free:Given production rules: E.g.And a string: Produce a valid parse tree:If English can be represented with CFG:CFG strike a good balance:Sentences are broken into constituentsConstituents have certain key propert]]></description><category></category></item><item><title><![CDATA[Lecture 13 Formal Language Theory & Finite State Automata]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131086257</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131086257</guid><author>Abner98414</author><pubDate>Fri, 09 Jun 2023 15:23:14 +0800</pubDate><description><![CDATA[语言模型隐马尔可夫模型循环神经网络但是，这些模型都没有涉及到语言的本质，因为它们可以用于处理任何符号序列，而不仅限于单词、句子等。]]></description><category></category></item><item><title><![CDATA[Lecture 12 Discourse]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131080472</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131080472</guid><author>Abner98414</author><pubDate>Wed, 07 Jun 2023 13:45:06 +0800</pubDate><description><![CDATA[对于许多任务，考虑上下文要比考虑句子本身更重要。传统上，许多流行的 NLP 应用程序的关注点都在句子层面（例如机器翻译），但是这种情况已经开始改变……010.png。]]></description><category></category></item><item><title><![CDATA[Lecture 11 Contextual Representation]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131073570</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131073570</guid><author>Abner98414</author><pubDate>Wed, 07 Jun 2023 09:02:31 +0800</pubDate><description><![CDATA[我们学习了基于 ELMo 和 BERT 的单词上下文表示，以及它们在下游任务中的表现，并且也学习了如何将它们应用到下游任务中。这些模型都是在非常大的语料库上训练的。因此，它们构建了一些语言相关的知识。使用无监督目标，模型训练无需专门提供带标签数据集。由于模型是在非常大的语料库上训练的，因此，当我们将它们用于下游任务时，我们不再是从零开始（“scratch”）的状态，因为模型在某种程度上已经理解了一些单词含义相关的信息，以及单词之间的关系。]]></description><category></category></item><item><title><![CDATA[Lecture 10 Distributional Semantics]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131068375</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131068375</guid><author>Abner98414</author><pubDate>Tue, 06 Jun 2023 18:06:21 +0800</pubDate><description><![CDATA[这里，每一行都表示语料库中的一个文档，每一列表示语料库的词汇表中的一个单词，单元格中的数字表示该单词在对应文档中出现的频率。Unsupervised: Unlabelled corpus 无监督 原始的、无标签的语料库 正如前面所述，我们的模型无需标签，只要从给定的语料库按照窗口大小对其中出现的单词进行计数，并计算概率，进行学习即可。可以看到，单词 “the” 对应的列的值都为 0，这是因为其 IDF 值为 0，所以无论对应单元格的 TF 值为多少，相乘后得到的结果都是 0。]]></description><category></category></item><item><title><![CDATA[Lecture 9 Lexical Semantics]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131056728</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131056728</guid><author>Abner98414</author><pubDate>Tue, 06 Jun 2023 03:47:33 +0800</pubDate><description><![CDATA[在 NLP 中，我们为什么要关注词汇语义学？我们先来看一个情感分析的例子：假设现在我们有一个情感分析任务，我们需要预测一段给定文本的情感极性。Bag-of-words, KNN classifier. Training data: 词袋模型，KNN分类器。训练数据Two problems here: 这里存在两个问题Comparing words directly will not work. How to make sure we compare word meanings? 直接比较单词并不是一种很好的]]></description><category></category></item><item><title><![CDATA[Lecture 8 Deep Learning for NLP: Recurrent Networks]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131035153</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131035153</guid><author>Abner98414</author><pubDate>Sun, 04 Jun 2023 21:15:25 +0800</pubDate><description><![CDATA[Core idea: have memory cells that preserve gradients across time. Access to the memory cells is controlled by gates. 核心思想：拥有跨时间保存梯度的记忆单元。Gated recurrent unit (GRU): Simplified variant with only 2 gates and no memory cell 门控循环单元（GRU）：简化的变体，只有2个门，没有记忆单元。]]></description><category></category></item><item><title><![CDATA[Lecture 7 Deep Learning for NLP: Feedforward Networks]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131027741</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131027741</guid><author>Abner98414</author><pubDate>Sun, 04 Jun 2023 18:02:55 +0800</pubDate><description><![CDATA[Given a document, classify it into a predefined set of topics. E.g. economy, politics, sports 给定一个文档，将其分类到预定义的主题集合中。例如：经济，政治，体育。In feed-forward neural network language model, the first layer is the sum of input word embeddings 在前向传播神经网络语言模型中，第一层是输入词嵌入的和。]]></description><category></category></item><item><title><![CDATA[Lecture 6 Sequence Tagging: Hidden Markov Models]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131027014</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131027014</guid><author>Abner98414</author><pubDate>Sun, 04 Jun 2023 00:45:26 +0800</pubDate><description><![CDATA[Output independence: An observed event(word) depends only on the hidden state(tag) 输出独立性：观察到的事件（词）只取决于隐藏状态（标签） ->Correct way: Consider all possible tag combinations, evaluate them, take the max. 正确的方法：考虑所有可能的标签组合，评估它们，取最大值。]]></description><category></category></item><item><title><![CDATA[Lecture 5 Part of Speech Tagging]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131026457</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131026457</guid><author>Abner98414</author><pubDate>Sat, 03 Jun 2023 22:54:28 +0800</pubDate><description><![CDATA[Lecture 5 Part of Speech Tagging]]></description><category></category></item><item><title><![CDATA[Lecture 4 Text Classification]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131026140</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131026140</guid><author>Abner98414</author><pubDate>Sat, 03 Jun 2023 22:15:57 +0800</pubDate><description><![CDATA[Input layer(features), output layer(class probabilities), and one or more hidden layers 输入层（特征），输出层（类别概率），以及一个或多个隐藏层。Classes: Topic categories E.g. “jobs”, “international news” 类别：主题类别 例如"工作"，“国际新闻”Classes: entailment, contradiction, neutral 类别：蕴含，矛盾，中立。]]></description><category></category></item><item><title><![CDATA[Lecture 3 N-gram Language Models]]></title><link>https://blog.youkuaiyun.com/Abner98414/article/details/131025008</link><guid>https://blog.youkuaiyun.com/Abner98414/article/details/131025008</guid><author>Abner98414</author><pubDate>Sat, 03 Jun 2023 20:05:52 +0800</pubDate><description><![CDATA[Lecture 3 N-gram Language Models]]></description><category></category></item></channel></rss>