user@devops:~$ cat README.md
CNN CIFAR-10 v2
# Description
Improved CIFAR-10 classifier with deep CNN. Redesigned architecture with more convolutional layers, additional filters and better regularization. Network consists of 4 Conv2D-BatchNorm-MaxPool blocks with 32, 64, 128 and 256 filters respectively, followed by Dense layers with aggressive Dropout (0.5). Uses advanced data augmentation (20deg rotation, 15% zoom, 20% shift, horizontal flip), adaptive learning rate with ReduceLROnPlateau, EarlyStopping with patience 15, and pixel rescaling. Aims to surpass previous iteration accuracy through systematic hyperparameter tuning and cross-validation.
# Key features
$ Deep CNN architecture with 4 convolutional blocks
$ Batch Normalization after each Conv2D layer
$ Advanced data augmentation with optimized parameters
$ Dropout regularization (0.5) in dense layers
$ EarlyStopping, ReduceLROnPlateau and model checkpoint
$ Cross-validation and accuracy comparison vs previous iteration
# Gallery
# Technologies used
