bert-base-uncased-sst2-membership-attack Open-source Model - Accurately Detect Membership Attacks with an Accuracy of 86.81%

Bert Base Uncased Sst2 Membership Attack

Developed by doyoungkim

A fine-tuned model based on bert-base-uncased for membership inference attack detection tasks, achieving an accuracy of 86.81% on the evaluation set.

Downloads 116

Release Time : 3/2/2022

Model Overview

This model is a fine-tuned version of bert-base-uncased, primarily used for membership inference attack detection tasks.

High Accuracy

Achieved an accuracy of 86.81% on the evaluation set.

BERT-based Architecture

Leverages BERT's powerful contextual understanding capabilities for fine-tuning.

Linear Learning Rate Scheduling

Uses a linear learning rate scheduling strategy to optimize the training process.

Text Classification

Membership Inference Attack Detection

Natural Language Understanding

Security Detection

Membership Attack Identification

Detects whether specific text belongs to the training dataset's members.

Accuracy: 86.81%

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6921	1.0	3813	0.6263	0.8360
0.6916	2.0	7626	0.6296	0.8681
0.6904	3.0	11439	0.6105	0.8406
0.6886	4.0	15252	0.6192	0.8200
0.6845	5.0	19065	0.6250	0.7798

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base