Download presentation
Presentation is loading. Please wait.
1
LSTM Practical Exercise
R. Q. Feitosa, J. D. Bermudez, J. A. Chamorro
2
Objective Predict the value of π π‘+1 from past observations ( π π‘β2 , π π‘β1 , π π‘ )
3
Network Architecture Architecture: Many-to-one LSTM
4
Dataset Daily USD to BRL exchange rate from April 30th 2018 to April 30th 2019.
5
Dataset Daily USD to BRL exchange rate from April 30th 2018 to April 30th 2019.
Train Test
6
Dataset Dataset preprocessing: Sliding window
π π π π π π y 3.92 3.99 3.9 3.86 Train Test
7
Dataset Dataset preprocessing: Sliding window
π π π π π π y 3.92 3.99 3.9 3.86 3.89 Train Test
8
Dataset Dataset preprocessing: Sliding window
π π π π π π π¦ 3.92 3.99 3.9 3.86 3.89 β¦ Train Test
9
Experimental setup Network architecture: LSTM (Many-to-one)
10
Expected Results Root Mean Squared Error Train: 0.02939 Test: 0.03424
Train prediction results Test prediction results
11
Exercises Exercise 1: Exercise 2:
Define the LSTM network architecture according to this presentation Exercise 2: Evaluate the network performance in terms of RMSE and prediction graph for the next configuration modifications: 10 and 100 epochs 10 and 100 LSTM hidden units 2 LSTM layers (Deep LSTM) Use GRU instead of LSTM
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.