Pinned Loading

1. deep-neural-network-representation-analysis deep-neural-network-representation-analysis Public

   Representation analysis, generalization, domain adaptation, compression, and layer-wise training using SI/Cross-SI metrics across CNNs, VGG16, DANN, and Vision Transformers.

   Jupyter Notebook

2. robust-zero-shot-clip-adaptation robust-zero-shot-clip-adaptation Public

   Robust zero-shot CLIP classification on CIFAR-10 using PGD attacks, LoRA fine-tuning, TeCoA adversarial training, and visual prompt tuning.

   Jupyter Notebook

3. handwritten-character-localization-and-recognition handwritten-character-localization-and-recognition Public

   End-to-end handwritten character localization, clustering, and semi-supervised recognition using Faster R-CNN, HOG/LBP + PCA, and FixMatch.

   Jupyter Notebook

4. trustworthy-ai-experiments trustworthy-ai-experiments Public

   Trustworthy AI experiments covering spurious correlations, adversarial robustness, explainability, causal inference, and algorithmic fairness.

   Jupyter Notebook

5. deep-generative-models deep-generative-models Public

   Selected deep generative modeling experiments with β-VAE, CycleGAN, energy-based models, score-based generation, DDPM, and DDIM.

   Jupyter Notebook

6. privacy-preserving-federated-learning privacy-preserving-federated-learning Public

   Cross-silo federated learning on FEMNIST with membership inference attacks, Record-Level DP-SGD, Subject-Level DP, and privacy–utility analysis.

   Jupyter Notebook