AI Paper Atlas

Research map / Federated learning and privacy / Privacy attacks and secure learning

TAPAS: Tricks to Accelerate (encrypted) Prediction As a Service

Amartya Sanyal, Matt Kusner, Adria Gascon, Varun Kanade

ICML 2018 · Federated learning and privacy › Privacy attacks and secure learning · approach: Encryption-based private learning · 59 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Apply the +1 trick to sparsify encrypted sums (ablation)TAPAS BNN inference · MNISTprediction time: 65.1h -> 37.22 hImproved
Apply the +1 trick to sparsify encrypted sums (ablation)TAPAS BNN inference · Facesprediction time: 763.5h -> 564hImproved
Apply the +1 trick to sparsify encrypted sums (ablation)TAPAS BNN inference · Diabetesprediction time: 283s -> 250sImproved
Apply the +1 trick to sparsify encrypted sums (ablation)TAPAS BNN inference · Cancerprediction time: 3.5s -> 3.5sNo clear effect
Drop 10% of weights (ablation)BNN · Cancer testaccuracy: 0.971 -> 0.976Improved
Drop 10% of weights (ablation)BNN · Diabetes testaccuracy: 0.549 -> 0.549No clear effect
Drop 10% of weights (ablation)BNN · Faces testaccuracy: 0.891 -> 0.879Hurt
Drop 10% of weights (ablation)BNN · MNIST testaccuracy: 0.986 -> 0.976Hurt
Drop 20% of weights (ablation)BNN · Cancer testaccuracy: 0.971 -> 0.912Hurt
Drop 20% of weights (ablation)BNN · Diabetes testaccuracy: 0.549 -> 0.541Hurt
Drop 20% of weights (ablation)BNN · Faces testaccuracy: 0.891 -> 0.878Hurt
Drop 20% of weights (ablation)BNN · MNIST testaccuracy: 0.986 -> 0.973Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy