boss-li12/data-privacy

Preserve data privacy with k-anonymity (samarati & mondrian), differential privacy, federated learning, paillier homomorphic encryption, etc.

Data Privacy

Experiments

K-Anonymity

• Samarati

• Mondrian

Privacy Preserving Federated Learning

• Differential Privacy

  • Gaussian Mechanism

• Secure Multi-Party Computation

  • Paillier Homomorphic Encryption

Detailed reports are inside the two directories.

Assignments

HW1: Basics of Privacy

• (c, l)-diversity

• quasi-identifier & full domain generalization & suppression

• t-closeness & Earth Mover's distance

• k-anonymity

• prior and posterior probabilities & mutual information

• (α, β)-privacy & γ-amplifying

HW2: Differential Privacy (DP)

• (ε, δ)-DP

• local DP

• sensitivity; laplace, gussian, and exponential mechanisms

• composition theorem & advanced composition theorem

• random subsampling

HW3: Cryptography & Security

• one time pad

• interchangeble libraries

• negligible functions

• pseudo random generator (PRG)

• RSA