Industry classifier yang - Revision history

Yangzhang at 20:25, 26 September 2017

2017-09-26T20:25:38Z

Yangzhang at 20:12, 26 September 2017

2017-09-26T20:12:32Z

Yangzhang at 19:27, 26 September 2017

2017-09-26T19:27:33Z

Yangzhang at 19:25, 26 September 2017

2017-09-26T19:25:36Z

Yangzhang: Created page with "Industry Classifier - by Yang Zhang"

2017-09-26T19:14:03Z

Created page with "Industry Classifier - by Yang Zhang"

New page

Industry Classifier - by Yang Zhang

@@ Line 1: / Line 1: @@
 Industry Classifier - by Yang Zhang
-Goal:
+Goal: For each company we want to classify its industry based on its description
 Approach:

@@ Line 1: / Line 1: @@
-==Industry Classifier== - by Yang Zhang
+Industry Classifier - by Yang Zhang
-Goal: for each company we want to classify its industry based on its description
+Goal:
 Approach:
-step 1: encode the text description into numerical values
+Step 1: encode the text description into numerical values.
-step 2: build a deep neural network to learn to classify
+Step 2: build a deep neural network to learn to classify.
-For step 1, a very naive way is to use the bag of words representation. The obvious drawback of this way is that you ignore the correlations between the words and also their orders. So, instead, we use word2vec this method, where each word is represented as a vector which indicate how likely other words will appear around this center word.
+For step 1, a very naive way is to use the "bag of words" representation. The obvious drawbacks are that you just ignore the correlations between the words and also their relative orders. So, instead, we use "word2vec" (https://en.wikipedia.org/wiki/Word2vec) this method, where, in short, each word is mapped to a vector which represents how likely the other words appear around this center word.
-For step 2: we have tried 1D convolutional and LSTM. Both can achieve 90+% training accuracy and around 60% testing accuracy. Notice that this task is even hard for humans and the baseline of randomly guessing is around 10%, 60% is acceptable. Turning the parameters doesn't help much meaning we might have reached the model's max capability.
+For step 2: we have tried 1D/2D convolutional NN (Neural Network) and LSTM RNN (Recurrent Neural Network). All the models can achieve 90+% training accuracy and around 60% testing accuracy. Notice that this task is even hard for humans and the baseline of randomly guessing is around 10%, 60% is acceptable. Turning the parameters doesn't help much meaning we might have reached the model's max capability.
 Next steps:
 Try with longer descriptions and see if more information can provide us better accuracy

← Older revision		Revision as of 19:27, 26 September 2017
Line 1:		Line 1:
−	Industry Classifier - by Yang Zhang	+	==Industry Classifier== - by Yang Zhang

	Goal: for each company we want to classify its industry based on its description		Goal: for each company we want to classify its industry based on its description

← Older revision		Revision as of 19:25, 26 September 2017
Line 1:		Line 1:
	Industry Classifier - by Yang Zhang		Industry Classifier - by Yang Zhang
		+
		+	Goal: for each company we want to classify its industry based on its description
		+
		+	Approach:
		+	step 1: encode the text description into numerical values
		+	step 2: build a deep neural network to learn to classify
		+
		+	For step 1, a very naive way is to use the bag of words representation. The obvious drawback of this way is that you ignore the correlations between the words and also their orders. So, instead, we use word2vec this method, where each word is represented as a vector which indicate how likely other words will appear around this center word.
		+
		+	For step 2: we have tried 1D convolutional and LSTM. Both can achieve 90+% training accuracy and around 60% testing accuracy. Notice that this task is even hard for humans and the baseline of randomly guessing is around 10%, 60% is acceptable. Turning the parameters doesn't help much meaning we might have reached the model's max capability.
		+
		+	Next steps:
		+	Try with longer descriptions and see if more information can provide us better accuracy