## The Model Implementation ### Hybrid Parallel Architecture Integrating FFN, 1D CNN, and LSTM for Predicting Wildfire Occurrences in Morocco Tensorflow Implementation
# Hybrid Parallel Architecture for Wildfire Prediction
This repository contains the code for a hybrid parallel architecture integrating Feedforward Neural Networks (FFNs), 1D Convolutional Neural Networks (1D CNNs), and Long Short-Term Memory Networks (LSTMs) for predicting wildfire occurrences in Morocco. The architecture is designed to leverage the strengths of each component: FFNs for broad feature transformation, CNNs for spatial pattern recognition, and LSTMs for capturing temporal dependencies.
## Dataset
The model is trained on the "Morocco Wildfire Predictions: 2010-2022 ML Dataset" available on Kaggle:
kaggle.com
The dataset incorporates a variety of features including meteorological conditions, soil moisture levels, and vegetation indices, which are critical for predicting wildfire occurrences. This dataset is particularly well-suited for LSTMs and 1D CNNs due to its inclusion of lagged features.
## Model Architecture
The architecture begins with a primary input layer that accepts the entire feature set as a flat vector. This input is processed through an FFN block consisting of two dense layers with 128 and 32 neurons, respectively, each followed by batch normalization and dropout layers to prevent overfitting and improve generalization. This block captures broad, non-sequential patterns within the dataset.
For each identified feature subset, the architecture employs parallel processing paths:
* **CNN Path:** The feature subset is reshaped to form a sequence and processed by a 1D CNN layer with 16 filters and a kernel size of 3. The convolutional layer captures local spatial patterns within the sequence. This is followed by a Global Average Pooling layer to reduce the spatial dimensions, retaining only the most crucial information.
* **LSTM Path:** Concurrently, the same reshaped feature subset is fed into an LSTM layer with 16 units. The LSTM captures temporal …