529 B
529 B
Deep Learning Portfolio — Documentation
Deep dives into each experiment: architecture, training configuration, and results — illustrated with real plots from the notebooks.
Projects
- MLP Regularization — Dropout comparison (with an honest note on a confounded variable)
- CNN BatchNorm Comparison — clean ablation, BatchNorm's effect isolated
- Autoencoder Denoising — noisy-to-clean image reconstruction through a compressed bottleneck