TF-Regularization-Benchmarks

A comparative study of regularization techniques (Batch Normalization & Dropout) on Encoder-Decoder architectures using the MNIST dataset and TensorBoard.

📊 Results Comparison

Below is the comparison chart of the four architectures trained over 5 epochs:

1. Basic (No Regularization)
2. With Batch Normalization
3. With Dropout
4. With Both (Batch Normalization + Dropout)

📁 Repository Structure

• Encoder_Decoder_MNIST.ipynb: The primary notebook demonstrating model setup and Matplotlib visualizations.
• Encoder_Decoder_MNIST_TensorBoard.ipynb: Notebook with integrated TensorBoard callbacks for advanced metrics tracking.
• Result/: Contains visual exports of the training results.