Pinned Loading

1. An intrusion detection module built on the BESS framework

   C++

2. Interpreting the outputs of an LSTM using the Captum library. Explores various methods such as DeepLIFT and integrated gradients to assign attribution scores to features.

   Jupyter Notebook 1

3. Using an LSTM built with PyTorch to predict time series of stock prices. Testing out different loss functions and optimizing hyperparameters using Ray Tune.

   Jupyter Notebook 1

4. Exploring various novel model architectures (such as TimesNet, N-BEATS, and N-HiTS) for time series forecasting of stock prices. Tests the effect of past and future exogenous variables on predictio…

   Jupyter Notebook 2

5. Build an autoencoder using LSTM layers to embed time series sequences into a lower-dimensional representation. The embedding size is optimized using Ray.

   Jupyter Notebook

6. Implements the classical DTW algorithm in PyTorch, enabling multi-pattern matching and GPU acceleration

   Jupyter Notebook 2