Zero-knowledge
We study zero-knowledge primitives in two different scenarios:
- To provide verifiability in collaborative machine learning scenarios.
- As a way to provide scalability to permissionless blockchains.
Verifiability in collaborative machine learning
In many contemporary machine learning workflows, particularly those involving edge computing, outsourced training, or federated learning, the entity that develops or deploys a model often lacks direct control over the training environment. This creates a security gap, as the training process may occur on untrusted infrastructure or with data provided by third parties. A growing concern in such scenarios involves model poisoning and data poisoning attacks, where adversaries manipulate the training process or inject malicious data to corrupt the resulting model. These threats not only undermine model accuracy but can also introduce backdoors or hidden behaviors, posing serious risks in safety-critical and privacy-sensitive applications.
To address these concerns, we make use of cryptographic primitives to design verifiable colaborative learning schemes that protect the integrity of outsourced model training against both data and model poisoning. These scheme employ a combination of cryptographic hashes, digital signatures, and zero-knowledge proofs to enforce trust and accountability throughout the training pipeline. Hashes provide continuity and traceability for data collected from authenticated sources, ensuring it remains untampered. Digital signatures authenticate the origin and integrity of training inputs, while zero-knowledge proofs enable verifiable assurance that the training entity has correctly executed the model update process without revealing sensitive information. This design allows to securely delegate training tasks to potentially untrusted parties while maintaining end-to-end verifiability.
Scalability of permissionless blockchains
Scalability is one of the most pressing challenges facing blockchain technology today. As decentralized networks grow in usage, the limitations of current consensus mechanisms and data throughput become increasingly apparent. Most blockchains, particularly those with strong security and decentralization guarantees like Bitcoin and Ethereum, can only process a limited number of transactions per second due to the need for every node to validate and store every transaction. This bottleneck leads to network congestion, high transaction fees, and slow confirmation times, making blockchains less practical for high-volume applications such as payments, gaming, or decentralized finance (DeFi). Addressing scalability without compromising security or decentralization is a critical area of ongoing research and innovation in blockchain development.
One of the solutions deployed to improve blockchain scalability are zk-rollups. zk-Rollups are a Layer 2 scaling solution that bundle many transactions off-chain and generate zero-knowledge proof that verifies their correctness. This proof is then submitted on-chain, allowing for greater throughput and lower fees while maintaining strong security guarantees from the underlying Layer 1.
We have been studying zk-rollups from different perspectives and have compared them with other scaling solutions. We have provided compartive reviews of several state-of-the-art Layer 2 scaling techniques. Our analysis focuses on key aspects such as usability, security, and cost, offering a structured evaluation of the trade-offs involved in each approach.
We have also conducted a rigorous analysis of the security foundations of the most widely adopted Layer 2 scalability mechanisms—Payment Channel Networks, Optimistic Rollups, Zero-Knowledge Rollups, Validium, and Volition. We systematically identify and categorize both the baseline security assumptions inherited from Layer 1 (L1) blockchains and the additional assumptions introduced by each Layer 2 protocol. Our findings demonstrate that these solutions rely on supplementary trust and security models beyond those of their underlying blockchains.