[ Skip to the content ]

Institute of Formal and Applied Linguistics Wiki

[ Back to the navigation ]

You are here: start » courses » rg » 2013 » convolution-kernels

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision 		Previous revision
courses:rg:2013:convolution-kernels [2013/03/11 18:49]
dusek 		courses:rg:2013:convolution-kernels [2013/03/12 11:27] (current)
popel <latex>x</latex> was not rendered
Line 19: Line 19:
       * They are able to "generate" fake inputs, but this feat is not used very often.       * They are able to "generate" fake inputs, but this feat is not used very often.
       * Examples: Naive Bayes, Mixtures of Gaussians, HMM, Bayesian Networks, Markov Random Fields       * Examples: Naive Bayes, Mixtures of Gaussians, HMM, Bayesian Networks, Markov Random Fields
-    * **Discriminative models** do everything in one-step -- they learn the posterior <latex>P(y|x)</latex> as a function of some features of <latex>x</latex>.+    * **Discriminative models** do everything in one-step -- they learn the posterior <latex>P(y|x)</latex> as a function of some features of <latex> x</latex>.
       * They are simpler and can use many more features, but are prone to missing inputs.       * They are simpler and can use many more features, but are prone to missing inputs.
-      * Examples: SVM, Logistic Regression, Neuron. sítě, k-NN, Conditional Random Fields+      * Examples: SVM, Logistic Regression, Neural network, k-NN, Conditional Random Fields
   - Each CFG rule generates just one level of the derivation tree. Therefore, using "standard" nonterminals, it is not possible to generate e.g. this sentence:   - Each CFG rule generates just one level of the derivation tree. Therefore, using "standard" nonterminals, it is not possible to generate e.g. this sentence:
     * ''(S (NP (PRP He)) (VP (VBD saw)(NP (PRP himself))))''     * ''(S (NP (PRP He)) (VP (VBD saw)(NP (PRP himself))))''
Line 35: Line 35:
   - Convolution is defined like this: <latex>(f*g)_k = \sum_i f_i g_{k-i}</latex>, so it measures the presence of structures that //complement// each other. Here, we have a measure of structures that are //similar//. So it is something different. But the main idea is the same -- we can combine smaller structures (kernels) into more complex ones.   - Convolution is defined like this: <latex>(f*g)_k = \sum_i f_i g_{k-i}</latex>, so it measures the presence of structures that //complement// each other. Here, we have a measure of structures that are //similar//. So it is something different. But the main idea is the same -- we can combine smaller structures (kernels) into more complex ones.
   - There is a (tiny) error in the last formula of Section 3. You cannot actually multiply tree parses, so it should read: <latex>\bar{w}^{*} \cdot h(\mathbf{x}) = \dots</latex>   - There is a (tiny) error in the last formula of Section 3. You cannot actually multiply tree parses, so it should read: <latex>\bar{w}^{*} \cdot h(\mathbf{x}) = \dots</latex>
 +
 +==== Report ====
 +
 +We discussed the answers to the questions most of the time. Other issues raised in the discussion were:
 +
 +  * **Usability** -- the approach is only usable for //reranking// the output of some other parser.
 +  * **Scalability** -- they only use 800 sentences and 20 candidates per sentence for training. We believe that for large data (milions of examples) this will become too complex.
 +  * **Evaluation** -- it looks as if they used a non-standard evaluation metric to get "better" results. The standard here would be F1-score.

[ Back to the navigation ] [ Back to the content ]