project 07
Hybrid ViT-BiGRU Pipeline
A hybrid deep learning model combining a Vision Transformer for spatial feature extraction with Bi-directional GRU networks for capturing temporal patterns.
- Category
- AI
technologies used
- Python
- PyTorch
- TensorFlow
- NumPy
- Pandas
- Seaborn
- CUDA
key features
- 01Self-attention based feature extraction with ViT, sequence modelling through BiGRU layers.
- 02Data preprocessing, augmentation and regularisation techniques to prevent overfitting.
- 03Visualisation of confusion matrices, ROC curves and training metrics.