A comparative experiment between RNN and LSTM models to evaluate their ability to perform noise-robust sequence prediction. The project tests short-term vs long-term memory by reconstructing clean input sequences from noisy data, showing how LSTM outperforms RNN under long-dependency conditions.
lstm rnn gradient-clipping sequence-modeling noise-robustness long-term-dependency gradient-vanishing
-
Updated
Nov 24, 2025 - Jupyter Notebook