Home▸Machine Learning & Neural Networks▸Secure Aggregation in Federated Learning

Secure Aggregation in Federated Learning (2D)

Interactive 2D simulator of the pairwise-masking secure-aggregation protocol used in federated learning: watch pairwise random masks hide every client's raw update from the server while their sum still cancels out to the exact true aggregate.

Machine Learning & Neural Networks2DAdvanced60 FPS📱 Mobile-adapted⇄ 3D version

Details

CategoryMachine Learning & Neural Networks

RenderingCanvas 2D

DifficultyAdvanced

Target FPS60

ControlsSliders set the number of clients (3-8) and the pairwise mask magnitude, a Run button executes a masked-aggregation round over 2D bar-chart vectors with animated particles carrying each client's masked update to a central server node, a dropout button simulates a client vanishing mid-round so the masks fail to cancel, and Server-view/Debug-view buttons toggle between the masked bars the server actually sees (red) and the raw per-client bars underneath (green).

federated-learningsecure-aggregationprivacycryptographymachine-learningdistributed-systems

Related simulations

Federated Learning Communication Efficiency Simulator (2D)Watch 10 federated-learning clients compress their model updates with top-K sparsification…

Federated Learning Communication Efficiency SimulatorWatch 10 federated-learning clients compress their model updates with top-K sparsification…

Byzantine-Robust Federated AggregationInteractive 3D simulator of Byzantine-robust federated learning: malicious clients send… ⊞ All 226 in Machine Learning & Neural Networks →

Federated learning trains a shared model across many clients' private data without ever centralising it — but the model updates themselves can still leak information unless the server is prevented from seeing any single client's contribution. This 2D simulator renders each client as a node around a central server, applies the real pairwise-masking scheme used in production secure-aggregation protocols to a private per-client vector, and animates the masked values flowing to the server. Live readouts compare the true aggregate against what the server reconstructs after the masks cancel, a toggle exposes both the "server view" (masked, individually unreadable) and a "debug view" (raw values), and a dropout button demonstrates why real deployments need secret-sharing to stay correct when a client disappears mid-round.

⚙ Under the hood

Interactive 3D simulator of the pairwise-masking secure-aggregation protocol used in federated learning: client updates are individually hidden from the server, yet their sum still cancels the masks to reveal the exact true aggregate.

federated-learningsecure-aggregationprivacycryptographymachine-learningdistributed-systems

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install