Timeline for why an advanced LSTM model produce the same results as a simpler one?
Current License: CC BY-SA 4.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Dec 9, 2019 at 13:14 | comment | added | jalil asadi | I even tried different implementation, for example, github.com/adsieg/Multi_Text_Classification/blob/master/… . but I still get the same results as the simple model above. I really dont know what to do anymore. | |
| Dec 9, 2019 at 13:10 | comment | added | jalil asadi | by that sentence, I meant that the training process for both models are similar and I just sent a picture of the training process for one model because both models have similar behavior . I am not a native English, so sorry about poor choice of words. | |
| Dec 9, 2019 at 7:00 | comment | added | Jonathan | Looks like you model might be overfitting. Can you add a legend to the picture exactly stating what it shows? Because above your picture it currently says "the training process for both models is as below picture". That sounds very different from what you just explained in the comments to me. Moreover, can you add the same graph for the other model? | |
| Dec 8, 2019 at 23:11 | comment | added | jalil asadi | yes they are relevant. there is a picture of the performance of second model above, the gray line is for training and the orange line is for test data. it perform very well on training but not on test. I also used every technic for reducing overfitting but I always get the same result. | |
| Dec 8, 2019 at 21:41 | comment | added | Jonathan | And is your dataset also related to reviews/ratings? And how do your accuracy graphs look like for training data? @jalilasadi | |
| Dec 7, 2019 at 22:59 | comment | added | jalil asadi | I increased the dataset to 200k and still no difference in the results for both models. they both achieve 74% accuracy on validation. even when I put an attention layer I get the same accuracy! | |
| Dec 7, 2019 at 17:13 | comment | added | jalil asadi | I noticed that by increasing the size of the data set, the final accuracy of both models on the validation set reduces almost by 2%. more interestingly I noticed that using a random embedding layer produces better accuracy than using pre-trained word vectors as the embedding layer. is it normal ? | |
| Dec 7, 2019 at 17:05 | comment | added | jalil asadi | thank you for your answer. I changed the dataset to 70k but still no change in the result, and both models produce the same results. my dataset is an official dataset in Persian, which is used for classification and has been used in a lot of papers. unfortunately, authors implementation is in Java language which I am not familiar with, I tried to figure it out but I couldn't. I even tried to use the attention layer on lstm model and hope to improve the result but I got the same result. ir.hit.edu.cn/~dytang/paper/emnlp2015/codes.zip this is the link of his codes in java. | |
| Dec 7, 2019 at 11:54 | history | answered | Jonathan | CC BY-SA 4.0 |