9. Hybrid LSTMs [Long Short-Term Memory]

preview_player
Показать описание
Lukas discusses how LSTMs perform classification, deep LSTMs, bidirectional LSTMs, and CNN/LSTM hybrids.

Follow us:

Check out Fully Connected, which features curated machine learning reports by researchers exploring deep learning techniques, Kagglers showcasing winning models, industry leaders sharing best practices, and more:
Рекомендации по теме
Комментарии
Автор

As per the previous video, I get no training data:
Traceback (most recent call last):
File "imdb-lstm.py", line 26, in <module>
(X_train, y_train), (X_test, y_test) = imdb.load_imdb()
File "C:\Users\su_matthewbennett\wandb\ml-class\videos\lstm-classifier\imdb.py", line 8, in load_imdb
X_train.extend([open(path + f).read() for f in os.listdir(path) if f.endswith('.txt')])
FileNotFoundError: [WinError 3] The system cannot find the path specified: './aclImdb/train/pos/'
Am I missing something here?

Mbittrainingltd
Автор

Really enjoyed the video but had a couple of doubts.
1.) Dont you need to preprocess the data to remove all the stop words
2.) if not does the stop words add some intrensic value to the structure of the sentence that would help building the model.
3.) Similarly for slangs, emojis etc etc.

spec
Автор

I want to say thank you for your presentation next Sir if you show how hybrid CNN-LSTM work more deeply its better

sufiankedirabdella
Автор

Is deep LSTM better than vanilla lstm?

lemoniall
Автор

Hi there,


An amazing video. I have been working for some time with text data and I must confess that it is great effort.


about code, I am getting an error when I try to build hybrid network i.e LSTM+CNN


TypeError: The added layer must be an instance of class Layer. Found: object at


Any ideas ?


Regards

hassansherwani