Long-term real-time network traffic flow prediction using LSTM recurrent neural network Presentation.pptx
- 1. Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network Cox Communication Wei Cai September 11th , 2019
- 2. What to expect We will … Provide an overview of what Long Short-Term Memory model is and the difference in internal structure of many-to-one and many-to-many models Illustrates its flexibility and relevance to generalize time series past pattern and predict for future Propose to add a sampling scheme to many-to-many deep neural network Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network
- 3. A Real-life Use Case input.json 658.7 MB Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network
- 4. In a nutshell The Long Short-Term Memory (LSTM) network is a type of Recurrent Neural Network, but has a unique formulation that allows it to train more effectively. Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network A typical RNN (Source: http://colah.github.io/posts/2015-08-Understanding-LS TMs/ ) A LSTM unit (Source : http://colah.github.io/posts/2015-08-Understanding-LS TMs )
- 5. LSTM: many-to-one vs. many-to-many Network architecture of many-to-one LSTM Network architecture of many-to-many LSTM Source Embedding Target Embedding Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network
- 6. LSTM: many-to-one Atucal data points of market “chndmcdl” Atucal data points of market “ btnrerwn” Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network
- 7. LSTM: many-to-one Prediction of market “chndmcdl” training RMSE: 4105.4180 testing RMSE: 4465.5039 Prediction of market “btnrerwn” training RMSE: 270.5595 testing RMSE: 325.1492 Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network
- 8. … X11 X12 X13 . . . X21 X22 X23 . . . Xk1 Xk2 Xk3 . . . sampling X(ke1)1 X(ke1)2 X(k+1)3 . . . X(ke2)1 X(ke2)2 X(k+2)3 . . . X(ken)1 X(ken)2 X(k+n)3 . . . … sampling Source Embedding Target Embedding Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network LSTM: many-to-many with a sampling schema
- 9. LSTM: many-to-many with and without a sampling schema testing RMSE: 4386.9658 Prediction of market “chndmcdl” before interpolation Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network Prediction of market “chndmcdl” after interpolation testing RMSE: 3897.6734
- 10. LSTM: many-to-many with and without a sampling schema testing RMSE: 5135.6238 Prediction of market “bntrbntr” before interpolation Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network Prediction of market “bntrbntr” after interpolation testing RMSE: 5855.8790
- 11. Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network LSTM: many-to-many with a sampling schema Decrease in Model Perplexity Increase in convergence speed Less prone to overfitting Denoise the impact of anomaly
- 12. Conclusion and Future Work Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network Not all data points matter equally in the duration of training try other sampling schema (i.e., importance) to accelerate the training and compare the difference in speed and performance Sampling schema can be investigated together with other techniques to speed up training
- 13. Questions? Long-Term Real-Time Network Traffic Flow Prediction Using LSTM Recurrent Neural Network Can also email questions to wei.cai@cox.com