This repository contains the implementation and analysis for the paper "Penerapan Singular Value Decomposition untuk Optimasi Neural Networks melalui Low-Rank Approximation".
The notebook demonstrates how low-rank approximation using Singular Value Decomposition (SVD) can optimize neural networks by:
- Reducing memory usage.
- Decreasing computation time.
- Maintaining comparable accuracy to the original model.
Key components:
- Original Model: A baseline neural network implemented using TensorFlow/Keras.
- Low-Rank Approximation: Decomposing weight matrices into low-rank representations.
- Performance Comparison: Evaluating accuracy, memory usage, and training/inference time across different ranks.
This work provides insights into making the deployment of neural networks more efficient, particularly for resource-constrained environments.