laplace-mechanism

Noisegate: a differential privacy gateway that lets an untrusted LLM agent query sensitive data over MCP (Model Context Protocol), with a formal guarantee no individual's record can leak even if the agent is adversarial - enforcement lives in trusted code below the model, validated by a runnable attack gallery.

An example of differential privacy with laplace mechanism in federated learning using Pytorch

The Dirichlet Mechanism for Differentially Private KL Divergence Minimization

Interactive differential privacy teaching demo: sensitivity, Laplace noise, privacy budgets, histograms, and trusted query servers.

Data privacy

Folder full of code using Differential Privacy; as part of our group project

An interactive differential privacy course — 9 hands-on modules spanning real-world breaches, live attack demos, the randomized-response protocol, a privacy-budget treasure hunt, and published industry deployments. Single HTML file, zero setup, includes a classroom-ready presenter's guide.