Many data-driven personalized services require that private data of users is scored against a trained machine learning model. In this paper we propose novel protocol for privacy-preserving classification decision trees, popular model in these scenarios. Our solutions composed out building blocks, namely secure comparison protocol, obliviously selecting inputs, and multiplication. By combining some the blocks our tree also improve previously proposed support vector machines logistic...

According to the security breach level index, millions of records are stolen worldwide on every single day. Personal health most targeted internet, and they considered sensitive, valuable. Security privacy important parameters cryptography encryption. They reduce availability data patients healthcare appropriate personnel ultimately lead a barrier in transfer into digital system. Using permission blockchain share can issues. literature, systems rely centralized system, which is more prone...

This work proposes a protocol for performing linear regression over dataset that is distributed multiple parties. The parties will jointly compute model without actually sharing their own private datasets. We provide security definitions, protocol, and proofs. Our solution information-theoretically secure based on the assumption Trusted Initializer pre-distributes random, correlated data to during setup phase. actual computation happens later on, an online phase, does not involve trusted...

Abstract Background In biomedical applications, valuable data is often split between owners who cannot openly share the because of privacy regulations and concerns. Training machine learning models on joint without violating a major technology challenge that can be addressed by combining techniques from cryptography. When collaboratively training with cryptographic technique named secure multi-party computation, price paid for keeping private an increase in computational cost runtime. A...

Clients of storage-as-a-service systems such as Amazon's S3 want to be sure that the files they have entrusted cloud are available now and will in future.

Machine learning (ML) is revolutionizing research and industry. Many ML applications rely on the use of large amounts personal data for training inference. Among most intimate exploited sources electroencephalogram (EEG) data, a kind that so rich with information application developers can easily gain knowledge beyond professed scope from unprotected EEG signals, including passwords, ATM PINs, other data. The challenge we address how to engage in meaningful while protecting privacy users....

We propose a privacy-preserving Naive Bayes classifier and apply it to the problem of private text classification. In this setting, party (Alice) holds message, while another (Bob) classifier. At end protocol, Alice will only learn result applied her input Bob learns nothing. Our solution is based on Secure Multiparty Computation (SMC). Rust implementation provides fast secure for classification unstructured text. Applying our case spam detection (the generic, can be used in any other...

Cryptocurrencies are widely used, yet current methods for analyzing transactions heavily rely on opaque, black-box models. These lack interpretability and adaptability, failing to effectively capture behavioral patterns. Many researchers, including us, believe that Large Language Models (LLMs) could bridge this gap due their robust reasoning abilities complex tasks. In paper, we test hypothesis by applying LLMs real-world cryptocurrency transaction graphs, specifically within the Bitcoin...

Machine Learning (ML) techniques now impact a wide variety of domains. Highly regulated industries such as healthcare and finance have stringent compliance data governance policies around sharing. Advances in secure multiparty computation (SMC) for privacy-preserving machine learning (PPML) can help transform these by allowing ML computations over encrypted with personally identifiable information (PII). Yet very little SMC-based PPML has been put into practice so far. In this paper we...

The McEliece public-key encryption scheme has become an interesting alternative to cryptosystems based on number-theoretical problems. Different from RSA and ElGamal, PKC is not known be broken by a quantum computer. Moreover, even though relatively big key size, decryption operations are rather efficient. In spite of all the recent results in coding-theory-based cryptosystems, date, there no constructions secure against chosen ciphertext attacks standard model-the de facto security notion...

Existing work on privacy-preserving machine learning with Secure Multiparty Computation (MPC) is almost exclusively focused model training and inference trained models, thereby overlooking the important data pre-processing stage. In this work, we propose first MPC based protocol for private feature selection filter method, which independent of training, can be used in combination any to rank features. We an efficient scoring Gini impurity end. To demonstrate feasibility our approach...

Abstract Most existing Secure Multi-Party Computation (MPC) protocols for privacy-preserving training of decision trees over distributed data assume that the features are categorical. In real-life applications, often numerical. The standard “in clear” algorithm to grow on with continuous values requires sorting examples each feature in quest an optimal cut-point range node. Sorting is expensive operation MPC, hence finding secure avoid such step a relevant problem machine learning. this...