Episode #125 of the Stack Overflow podcast is here. We talk Tilde Club and mechanical keyboards. Listen now

New answers tagged


So if I use a bi-directional LSTM/GRU layer over Fasttext representations of words, will it be the same? Sep. 27, 2018 the authors published a follow on work to ELMo, "Dissecting Contextual Word Embeddings: Architecture and Representation". They test whether a bidirectional language model (biLM) can be replicated using Transformers, CNNs, and deeper LSTMs....


I am not sure if it could work but I have seen this approach in: https://github.com/chickenbestlover/RNN-Time-series-Anomaly-Detection It uses a RNN autoencoder for prediction and the it finds out anomalies. Before training the autoencoder it augments data adding many different levels of noise. Please check it.


The point of the RNN is that, in your terminology, w1...w10 are actually all the same sets of weights. The weights are not different for different time steps. So really you have w1=w2=w3=...=w10=w. There are only one set of weights for an RNN regardless of how long the input sequence is. Conceptually you are feeding your outputs back into the inputs of the ...

Top 50 recent answers are included