Is ML-Based Cryptanalysis Inherently Limited? Simulating Cryptographic Adversaries via Gradient-Based Methods.
Avital ShafranEran MalachThomas RistenpartGil SegevStefano TessaroPublished in: CRYPTO (6) (2024)
Keyphrases
- significant improvement
- preprocessing
- neural network
- benchmark datasets
- hash functions
- database
- qualitative and quantitative
- statistical methods
- computationally expensive
- machine learning algorithms
- empirical studies
- maximum likelihood
- computational cost
- artificial neural networks
- feature extraction
- case study
- computer vision
- artificial intelligence
- data mining
- databases