Privasea Explained: AI Network with FHE
A dive into the technology around FHE with an in-depth look at how the protocol works with AI and FHE under the bonnet
Disclaimer: The content presented in this article, along with others, is based on opinions developed by the analysts at Dewhales and does not constitute sponsored content. At Dewhales, we firmly adhere to a transparency-first philosophy, making our wallets openly available to the public through our website or DeBank, and our articles serve as vehicles for self-expression, education, and contribution to the ecosystem.
Dewhales Capital does not provide investment advisory services to the public. Any information should not be taken as investment, accounting, tax or legal advice or as a recommendation to purchase, sell or hold or to pursue any investment style or strategy. The accuracy and appropriateness of the information is not guaranteed by Dewhales Capital.
1. Introduction: What is FHE, its history, and applications
2. Privasea Overview
2.1 Network participants
2.2 Privasea workflow
2.3 Privasea workflow
2.4 Secure KYC usecase workflow
3. GitHub
4. Tokenomics
5. Team
6. Partnerships, integrations and application
7. Backers
8. Conclusion
1. Introduction: What is FHE, Its History, and Applications
FHE stands for Fully Homomorphic Encryption, a type of encryption scheme that allows computations to be performed directly on ciphertexts without the need for decryption. This means that encrypted data remains encrypted throughout the entire computation process, and the result of the computation is also encrypted, without any party ever gaining access to the plaintext data at any point. FHE enables computations, including machine learning and AI analysis, on encrypted data, allowing scientists, researchers, and data-driven enterprises to extract valuable insights without decrypting or revealing the underlying data or models. Fully Homomorphic Encryption is the most powerful form of encryption, but it comes with a significant trade-off, requiring substantial computational power.
In fact, there are many types of FHE: BGV, BFV, CKKS, FHEW, NuFHE, and TFHE. Homomorphic Encryption (HE) initially developed slowly over the first three decades following its proposal in 1978. FHE, in its current form, became available in 2009 when Craig Gentry proposed a method to achieve Fully Homomorphic Encryption, although this system introduced noise with each operation. Later, a modified version with smaller ciphertexts was proposed, but the most popular scheme was introduced by Zvika Brakerski and his team, offering improved security and performance compared to earlier FHE methods. Through several successive iterations—BGV, BFV, and CKKS—GSW emerged, from which FHEW (introduced in 2014) and TFHE (introduced in 2016) were developed. Chillotti, Gama, Georgieva, and Izabachene reduced latency to less than 0.1 seconds per gate bootstrapping operation. Since then, the scheme has incorporated programmable bootstrapping into its process, accelerating FHE to make it practical for most use cases, both for web2 and web3 apps.
The end result so far has been Zama's TFHE-rs, which extends the original TFHE capabilities to support programmable bootstrapping over integers. And it is TFHE-rs that is used in the current Privasea implementation.
Usecases for FHE:
Cloud Computing: FHE can bring significant benefits to cloud computing by allowing users to store and process their data in encrypted form on remote servers. This means that users can leverage the immense computational power of the cloud while keeping their data secure and private. It’s a win-win for all parties involved.
Financial Services: Financial services can also leverage the capabilities of FHE. By securely processing financial data, FHE allows financial institutions to perform complex analyses on encrypted data. The best part is that customer privacy remains intact throughout the entire process.
Machine Learning: FHE can greatly benefit machine learning by training models on encrypted data. This enables organizations to harness the power of machine learning while keeping their data secure.
2. Privasea overview
The Privasea AI Network offers a solution to the challenges associated with data privacy in artificial intelligence. The network represents a cutting-edge architecture that combines Fully Homomorphic Encryption (FHE) with blockchain-based incentives, addressing growing concerns related to data privacy and meeting the increasing demand for collaborative AI computation. Privasea AI Network divides FHE from theory to application into the following four levels: application level, optimisation level, arithmetic level and primitive level. The network offers both generalised and tailored solutions to bridge the gap between user privacy and distributed computational resources in Al processing, covering all four levels of FHE.
A key goal of the Privasea AI Network is to ensure compliance with data protection regulations, including the stringent General Data Protection Regulation (GDPR) in the European Union. These regulations impose strict requirements on the collection, processing, and storage of personal data.
2.1 Network participants:
Network Users (Data owner, Result receivers): Initiate machine learning tasks, securely provide data, and interact with the network to obtain decrypted results.
Privanetix Nodes (Computation Nodes): Function as service providers in the Privasea AI Network, offering computational resources for privacy-preserving machine learning tasks.
Decryptors: Specialized participants who ensure the secure decryption of encrypted results generated by Privanetix Nodes. They collaborate with network users to decrypt and interpret the final results of machine learning tasks, ensuring the privacy and integrity of the decrypted results.
2.2 Privasea architecture:
The Privasea AI Network consists of four main components: the HESea Library, Privasea API, Privanetix, and the Privasea Smart Contract Kit.
HESea Library: This component forms the foundation of the Privasea AI Network, hosting a collection of highly efficient implementations of popular Fully Homomorphic Encryption schemes, such as TFHE, CKKS, BGV, BFV, and others. The HESea Library provides developers with access to a wide range of functions.
Privasea API: A comprehensive set of protocols and tools built on top of the HESea Library. The Privasea API allows developers to easily integrate advanced privacy-preserving features into their AI applications.
Privanetix: A network of computational nodes that enables secure computation on encrypted data. By distributing computations across multiple nodes, Privanetix ensures the scalability and efficiency of the Privasea AI Network.
Privasea Smart Contract Kit: This kit includes a series of smart contracts carefully designed to handle various aspects of network management.
Additionally, according to GitHub, Privasea includes several other components from Zama AI that provide state-of-the-art FHE solutions for blockchain and AI:
Concrete: An open-source FHE compiler (updated to TFHE-rs) that simplifies the use of Fully Homomorphic Encryption (FHE). It allows the transformation of Python programs into their FHE equivalents. Concrete is useful for developers who want to create high-level applications that accept encrypted inputs and produce encrypted outputs.
TFHE-rs: A pure Rust implementation of TFHE for boolean and integer arithmetic over encrypted data. TFHE-rs is designed for developers and researchers who want full control over what they can do with TFHE without worrying about low-level implementation details.
Concrete ML: An open-source toolset for privacy-preserving machine learning (PPML) built on Concrete by Zama. It aims to simplify the use of Fully Homomorphic Encryption (FHE) for data scientists, helping them automatically convert machine learning models into their homomorphic equivalent.
2.3 Privasea workflow:
The user creates an account and sets up a machine learning task. In the process, the vectors are encrypted locally using the API interface of the machine learning application and a switching key is generated locally.
Users can then submit encrypted tasks to the Privatenix network and pay for services via the blockchain.
Privatenix nodes receive and execute the encrypted tasks in the user's encryption domain, and then transmit the encrypted results to the decryptor's encryption domain using the switch key previously generated by the user.
Once the work is done, Privanetix send the results to the decryptors and get paid for their work
Next, the decryptors use their client keys to decrypt the results and send the decrypted results to the network users using the Proxy Re-encryption (PRE) scheme.
2.4 Secure KYC usecase workflow:
1. Register ID with photos: The user provides their ID, including a photo. The client extracts facial features from the ID photo using a feature extraction algorithm
2. Submit Selfie Verification Task: The user takes a selfie using a camera or mobile device. The client extracts facial features from the selfie image using the same feature extraction algorithm used in the ID registration.
3. Retrieve the Encrypted Embedding of ID Images: The assigned Privanetix node retrieves the encrypted ID embeddings associated with the user from the secure database.
4. Process the Facial Check in the Ciphertext Domain: Using FHE capabilities, the Privanetix node performs calculations on the encrypted ID and selfie embeddings, such as calculating the distance between them. The Privanetix node compares the distance with a predefined threshold to determine similarity. The result remains in the ciphertext domain.
5. Send Encrypted Result to Decryptor: The Privanetix node sends the encrypted boolean result to the decryptor for further processing.
6. Decryption and Result Extraction: The decryptor, possessing the private key, decrypts the received result to obtain the final check result (e.g., yes or no).
7. Result Delivery: The decryptor securely delivers the final result to the designated institutes or entities that require it using Proxy Re-encryption (PRE) or another suitable method.
3. GitHub
Above we've already looked at a few repositories from Zama AI, now let's look at some of the proprietary Privasea repositories. The first thing that catches your eye is that the last updates to Privasea repositories were in July 2023. This is probably because at the moment the competition in the sector around FHE and AI/ML task distribution is very high and so the team doesn't publish updates to keep their developments alive.
Privasea-general is the main Privasea repository and includes several open source packages. These are packages such as for example HESea_lib, an advanced fully homomorphic encryption (FHE) library that provides developers with a powerful, flexible and easy-to-use tool for secure computing. Or Privasea-Miscellaneous, which contains source data that can be used for other PrivateSea repositories.
HESEA_Lib - HESea is a cutting-edge fully homomorphic encryption (FHE) library that provides developers with a powerful, flexible, and easy-to-use tool for secure computation. Built with state-of-the-art cryptographic techniques and optimized for high performance, HESea is ideal for a wide range of use cases. HESea offers various FHE schemes, including TFHE, CKKS, BGV, BFV, and more, which enable users to perform computations on encrypted data without the need to decrypt it. This ensures that sensitive data remains secure and protected against privacy breaches and security threats.
comparison_demo - this is a demonstration of ciphertext sorting using the HESEA library . This demonstration program encrypts a sequence of plaintexts to be sorted into a corresponding ciphertext sequence and sorts the sequence by comparing the ciphertext sequence. The program then decrypts the ciphertext sequence and outputs the sorted plaintext sequence.
dinn_demo - this is a demonstration of secure handwritten digit recognition using the HESEA library. It represents a privacy-preserving deep learning approach using the TFHE encryption scheme. The main advantage of DINN is that it achieves competitive accuracy while maintaining normal operational efficiency through the innovative use of Discretised Neural Networks. These networks quantise weights and offsets, reducing the complexity of the underlying computation and simplifying the initial TFHE bootstrapping procedure. As a result, homomorphic estimation is further improved, leading to improved efficiency.
4. Tokenomics
The PRVA token serves as the utility token within the Privasea Al network, playing a crucial role in facilitating transactions, incentivizing participants and enabling on-chain governance. It also acts as a medium of exchange, enabling users to access privacy Al services and unlock various functionalities within the ecosystem. The value of the PRVA token is primarily driven by the demand for the network's services, which encompass privacy-preserving machine learning and other Al-based features.
Specifically, the PRVA token serves the following purposes within the ecosystem:
- Transaction Facilitation
- Incentives and Rewards
- Governance and Voting
- Staking and Network Security
- Access to Exclusive Features
Token distribution:
Mining/Staking (45%) - will be allocated to staking nodes that provide Fully Homomorphic Encryption (FHE) and other privacy services within the project.
Team Allocation (10%)
Backers (20%)
Marketing and Community Development Allocation (15%) - these tokens will be dedicated to marketing and community development initiatives.
Reserve (6%) - this category primarily applies to items that cannot be planned, such as future regulations that need to be met or licenses that need to be applied for.
Liquidity (4%) - this liquidity is essential because it ensures that participants can enter or exit their positions without causing significant price movements, facilitating growth for the market.
5. Team
David Jiao, CEO, LinkedIn - is an experienced entrepreneur with a strong background as a developer in complex systems, starting as a software engineer at Simplight Nanoelectronics in 2010, followed by Cybercom Group. From 2015 to 2020 David worked as cofounder and CPO at Golden Ridge Robotic AB - a startup R&D company mainly focused on Cyber-Physical home robotic system. In parallel, he was involved in software development for Volvo - as a system designer at Volvo, I was responsible for designing and prototyping system functions of vehicle configurations in Volvo's brand new SPA2 architecture. In 2021 he launched Nulink, which provides PRE+ZK technology for decentralised applications via APIs. In addition, David's experience at Volvo has enabled Privasea to participate in a joint project with RISE (a Swedish research institute) and Alkit Communications AB for the automotive industry.
Ting Gao, chief research scientist, LinkedIn - has a strong background in applied mathematics and mathematical modelling: since 2010 he has been a researcher at the Illinois Institute of Technology, since 2015 he has been a data analyst and data scientist at M3, then ML engineer at Twitter. And since 2021, he is an assistant professor at Huazhong University of Science and Technology in applied mathematics, stochastic modelling, deep learning with applications in mathematical finance
Alex (R) Gaidarski, Growth manager, LinkedIn - has extensive experience as a systems admin since 2006, which positions him as a marketer with a technical bent. Prior to Privasea, he worked at NuiLink with David Jiao as part of the marketing team.
Zean Darren, Community Growth and Management, LinkedIn - has extensive experience in community management, having participated as an ambassador for Manta, Polyhedra, Tanssi Network, and has experience as a Moderator and Helper at Arcomia and Story Chain.
6. Partnerships, integrations and application
Privasea's main practical implementation at the moment is ImHuman, an app that is already available on Google Play and AppStore. It is an implementation of PoH (Proof of Human) technology, which confirms human identity, protecting your digital presence from bots and artificial intelligence imitations. In this application, identity through facial biometrics is confirmed through NFT personalised proof of human identity. The biometric data is syberised through encrypted vectors on the user's device. These vectors are then securely encrypted using the user's client key and sent to Privasea's secure network servers. Privasea is currently planning to implement its PoH solution in Linea, for Movement, Gate and to integrate with Telegra,/Discord/Reddit.
Example of integration with TG bot:
Mind Network is a pioneer in decentralised zero-trust data lakes. Privasea is currently building a fine-tuned library optimised for core operations of fully homomorphic encryption, providing efficient and convenient solutions for both web2 and web3 customers. Mind Network, on the other hand, focuses on secure data-driven smart contracts and AI for encrypted data. One of the key aspects of this collaboration is Mind Network's integration of Mind Network's FHE datalake into Privasea's AI network.
BNB Grienfield - combines data management with decentralised finance (DeFi) potential in the BNB Smart Chain (BSC). Privasea AI Network and BNB Greenfield have joined forces to change the landscape of data storage and privacy. Privasea's FHE technology will enable persistent encryption of user data on the network. Also, developers on the BNB Greenfield platform can seamlessly perform computations with encrypted data, covering operations such as data statistics, logical analysis and machine learning model evaluation, without having to delve into the complex nuances of encryption methods including the ability to utilise Privanetix's powerful nodes.
Ton Network - Privasea has unveiled Secure LivenessCheck Bot ‘ a solution for TON Netwok that aims to redefine user authentication by harnessing the power of advanced facial recognition technology combined with fully homomorphic encryption [FHE].
Pri-Auto is a project for the sustainable automotive industry that Sweden's Vinnova Fordonsstrategisk Forskning och Innovation (FFI) has selected it for funding in 2023. Under the Pri-Auto project, Privasea will build infrastructure to create a secure data source for the automotive industry and enable data sharing between multiple parties such as OEMs, MaaS and insurance companies through intelligent access rights management. For this project, Privasea is collaborating with RISE (a Swedish research institute) and Alkit Communications AB, a data collection provider for OEMs such as Volvo Group and Volvo Cars.
7. Backers
Privasea has received commitment of $5m from backers such as Dewhales Capital, Binance Labs, Gate Labs, OKX Ventures, MH Ventures, K300 Ventures, QB Ventures, Crypto Times, Basics Capital, DuckDAO, as well as some business angels from the industry such as Zakaria (zak) Aves and Luke Sheng (of Chainlink).
8. Conclusion
By addressing efficiency issues and focusing on improving existing algorithms, Privasea is paving the way for widespread adoption of FHE, ensuring a future where data security and privacy can co-exist in this data-driven world. Their solution allows users to utilise the multiple distributed computing resources provided by the blockchain, while maintaining full control over their data and models as they process AI. Today's world is becoming more and more complex, and web3 is increasingly intertwined with AI.
Privasea links:
Website | Twitter | Discord | Documentation | GitHub | Medium
Download ImHuman App: Google Play Store | Apple App Store
To help us improve and provide you with the best content possible, we'd appreciate it if you could share your thoughts and opinions on the article you just read. Your feedback is very valuable to us and won't take more than 2-4 minutes.
Also, this post is public so feel free to share it post as well
Thank you so much! ❤️