Picture for Moran Baruch

Moran Baruch

IBM Research

Power-Softmax: Towards Secure LLM Inference over Encrypted Data

Add code
Oct 12, 2024
Viaarxiv icon

Converting Transformers to Polynomial Form for Secure Inference Over Homomorphic Encryption

Add code
Nov 15, 2023
Viaarxiv icon

Sensitive Tuning of Large Scale CNNs for E2E Secure Prediction using Homomorphic Encryption

Add code
Apr 26, 2023
Viaarxiv icon

HE-PEx: Efficient Machine Learning under Homomorphic Encryption using Pruning, Permutation and Expansion

Add code
Jul 07, 2022
Figure 1 for HE-PEx: Efficient Machine Learning under Homomorphic Encryption using Pruning, Permutation and Expansion
Figure 2 for HE-PEx: Efficient Machine Learning under Homomorphic Encryption using Pruning, Permutation and Expansion
Figure 3 for HE-PEx: Efficient Machine Learning under Homomorphic Encryption using Pruning, Permutation and Expansion
Figure 4 for HE-PEx: Efficient Machine Learning under Homomorphic Encryption using Pruning, Permutation and Expansion
Viaarxiv icon

Fighting COVID-19 in the Dark: Methodology for Improved Inference Using Homomorphically Encrypted DNN

Add code
Nov 17, 2021
Figure 1 for Fighting COVID-19 in the Dark: Methodology for Improved Inference Using Homomorphically Encrypted DNN
Figure 2 for Fighting COVID-19 in the Dark: Methodology for Improved Inference Using Homomorphically Encrypted DNN
Viaarxiv icon

Tile Tensors: A versatile data structure with descriptive shapes for homomorphic encryption

Add code
Nov 03, 2020
Figure 1 for Tile Tensors: A versatile data structure with descriptive shapes for homomorphic encryption
Figure 2 for Tile Tensors: A versatile data structure with descriptive shapes for homomorphic encryption
Figure 3 for Tile Tensors: A versatile data structure with descriptive shapes for homomorphic encryption
Figure 4 for Tile Tensors: A versatile data structure with descriptive shapes for homomorphic encryption
Viaarxiv icon

A Little Is Enough: Circumventing Defenses For Distributed Learning

Add code
Feb 16, 2019
Figure 1 for A Little Is Enough: Circumventing Defenses For Distributed Learning
Figure 2 for A Little Is Enough: Circumventing Defenses For Distributed Learning
Figure 3 for A Little Is Enough: Circumventing Defenses For Distributed Learning
Figure 4 for A Little Is Enough: Circumventing Defenses For Distributed Learning
Viaarxiv icon

Deceiving End-to-End Deep Learning Malware Detectors using Adversarial Examples

Add code
May 13, 2018
Figure 1 for Deceiving End-to-End Deep Learning Malware Detectors using Adversarial Examples
Viaarxiv icon