Ensuring safe custody of crypto-assets, while at the same time allowing a user to easily access and transfer them, is one of the biggest challenges of this nascent industry. This is even harder considering also the multiple technological implementations behind each crypto-asset. Here we present a survey of the various solutions for this custody problem, comparing advantages and disadvantages. Then we delve deeper into some interesting schemes based on secure multiparty computation, which give a blockchain-agnostic solution that balances security, safety, and transactional ease of use, and in particular, a protocol that enhances practicality by exploiting a party that may stay offline during the key generation.
The notion that collaborative machine learning can ensure privacy by just withholding the raw data is widely acknowledged to be flawed. Over the past seven years, the literature has revealed several privacy attacks that enable adversaries to extract information about a model's training dataset by exploiting access to model parameters during or after training. In this work, we study privacy attacks in the gray-box setting, where the attacker has only limited access - in terms of view and actions - to the model. The findings of our investigation provide new insights for the development of privacy-preserving collaborative learning solutions. We deploy SmartCryptNN, a framework that tailors homomorphic encryption to protect the portions of the model posing higher privacy risks. Our solution offers a trade-off between privacy and efficiency, which varies based on the extent and selection of the model components we choose to protect. We explore it on dense neural networks, where through extensive evaluation of diverse datasets and architectures, we uncover instances where a favorable sweet spot in the trade-off can be achieved by safeguarding only a single layer of the network. In one of such instances, our approach trains ~4 times faster compared to fully encrypted solutions, while reducing membership leakage by 17.8 times compared to plaintext solutions.
Motivated by applications to noncoherent network coding, we study subspace codes defined by sets of linear cellular automata (CA). As a first remark, we show that a family of linear CA where the local rules have the same diameter -- and thus the associated polynomials have the same degree -- induces a Grassmannian code. Then, we prove that the minimum distance of such a code is determined by the maximum degree occurring among the pairwise greatest common divisors (GCD) of the polynomials in the family. Finally, we consider the setting where all such polynomials have the same GCD, and determine the cardinality of the corresponding Grassmannian code. As a particular case, we show that if all polynomials in the family are pairwise coprime, the resulting Grassmannian code has the highest minimum distance possible.