The GRU uses fewer parameters to be tuned, but in my opinion, has the same performance as an LSTM for prediction, a good reference is "LSTM and GRU Neural Network Performance Comparison Study: Taking Yelp Review Dataset as an Example," but they are interchangeable?
They do give different results, but the difference is typically not large. For example, in load forecasting (https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9006868) GRUs received somewhat better results. Adding attention to those models can improve the accuracy as shown in the mentioned paper.
Dear Marco Antonio Márquez-Vera
This links might be useful, have a look:
1. https://towardsdatascience.com/aifortrading-2edd6fac689d
2. Conference Paper LSTM and GRU Neural Network Performance Comparison Study: Ta...
3. https://paperswithcode.com/paper/on-attribution-of-recurrent-neural-network/review/
4. https://www.researchgate.net/publication/347267378_LSTM_and_GRU_Neural_Network_Performance_Comparison_Study_Taking_Yelp_Review_Dataset_as_an_Example
Kind Regards
Qamar Ul Islam
In my opinion, usage purpose of both are same. However, GRU is simpler and easy than LSTM.
GRU has 2 gates and LSTM has 4 gates so GRU is easy to implement as compared to LSTM
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression