In-depth interpretation of Privasea, can facial data cast NFT still be played like this?

Author: Shishijun

1. Introduction

Recently, a face NFT casting project initiated by Privasea is extremely popular!

It seems very simple at first glance. In the project, users can enter their own faces on the IMHUMAN(I am human) mobile application and cast their face data into an NFT. That’s it. The combination of face data on-chain + NFT has enabled the project to obtain more than 20W+ NFT minting volume since its launch at the end of April, the popularity is evident.

I am also very confused, why? Can facial data be uploaded to the blockchain even if it is large? Will my facial information be stolen? What does Privasea do?

Wait, let us continue to research the project itself and the project party Privasea to find out.

Keywords: NFT, AI, FHE (Fully Homomorphic Encryption), DePIN

2. From Web2 to Web3-human-machine confrontation never stops

First, let’s explain the face NFT casting The purpose of the project itself, if you think this project is simply to cast face data into NFT, then you are totally wrong.

The App name of the project we mentioned above IMHUMAN (I am human) has well explained this problem: in fact, the project aims to determine whether you are in front of the screen through face recognition It's a real person.

First of all, why do we need human-machine recognition?

According to the 2024Q1report provided by Akamai (see appendix)shows that Bot (an automated program that can simulate humans sending HTTP requests and other operations) accounts for an astonishing 42.1% of Internet traffic, among which malicious ones Traffic accounts for 27.5% of the entire Internet traffic.

Malicious Bots may bring catastrophic consequences such as delayed response or even downtime to centralized service providers, affecting the experience of real users.

Let’s take the ticket grabbing scenario as an example. By creating multiple virtual accounts to grab tickets, cheaters can greatly increase the probability of successful ticket grabbing, and some even directly deploy automated programs on Next to the service provider's computer room, ticket purchasing can be achieved with almost zero delay.

Ordinary users have almost no chance of winning against these high-tech users.

Service providers have also made some efforts in this regard. On the client side, in the Web2 scenario, real-name authentication, behavior verification codes and other methods are introduced to distinguish humans and machines. On the server side, feature filtering and interception are carried out through WAF policies and other means. .

Can this problem be solved?

Obviously not, because the benefits from cheating are huge.

At the same time, the confrontation between man and machine is continuous, and both cheaters and testers are constantly upgrading their arsenals.

Take cheaters as an example. Taking advantage of the rapid development of AI in recent years, the client's behavior verification code has almost been dimensionally reduced by various visual models. AI even has faster and more accurate recognition capabilities than humans. . This forces the verifiers to passively upgrade, gradually transitioning from early user behavioral feature detection (image verification code) to biometric feature detection (perceptual verification: such as client environment monitoring, device fingerprints, etc.). Some High-risk operations may require upgrading to biological feature detection (fingerprints, face recognition).

For Web3, human-machine detection is also a strong demand.

For some project airdrops, cheaters can create multiple fake accounts to launch witch attacks. At this time, we need to identify the real person.

Due to the financial attributes of Web3, for some high-risk operations, such as account login, currency withdrawal, transaction, transfer, etc., it is not only the real person who needs to verify the user, but also the account owner, so face recognition becomes indispensable. Choice of two.

The demand is certain, but the question is how to realize it?

As we all know, decentralization is the original intention of Web3. When we discuss how to implement face recognition on Web3, the deeper question is actually How should Web3 adapt to AI scenarios:

• How should we build a decentralized machine learning computing network?
• How to ensure that the privacy of user data is not leaked?
• How to maintain the operation of the network, etc.?

3. Privasea AI NetWork-Exploration of Privacy Computing + AI

For the problems mentioned at the end of the previous chapter, Privasea has given a groundbreaking solution: Privasea is based on FHE (Fully Homomorphic Encryption) ) built Privasea AI NetWork to solve the privacy computing problem of AI scenarios on Web3.

FHE is, in layman’s terms, an encryption technology that ensures that the results of the same operation on plain text and cipher text are consistent.

Privasea has optimized and encapsulated the traditional THE, divided into application layer, optimization layer, arithmetic layer and original layer, forming the HESea library to adapt it to machine learning scenarios. The following are the specific responsibilities of each layer. Features:

Through its layered structure, Privasea provides more specific and tailored solutions to meet the unique needs of each user.

Privasea’s optimized packaging is mainly focused on the application layer and optimization layer. Compared with basic solutions in other homomorphic libraries, these customized calculations can provide more than a thousand times acceleration.

3.1 Privasea AI NetWork’s network architecture

Looking from the architecture of Privasea AI NetWork:

There are a total of 4 roles on its network,

Data owner, Privanetix node, decryptor, result Receiver .

1. Data Owner: Used to submit tasks and data securely through Privasea API.

2. Privanetix Node: It is the core of the entire network, equipped with advanced HESea library and integrated with blockchain-based incentive mechanism, which can perform safe and efficient calculations while protecting the privacy of the underlying data and ensuring calculation integrity and confidentiality.

3. Decryptor: Get the decrypted result through Privasea API and verify the result.

4. Result Receiver: The task results will be returned to the person designated by the data owner and task issuer.

3.2 The core workflow of Privasea AI NetWork

The following is the general workflow diagram of Privasea AI NetWork:

• STEP 1: User registration: Data owner passed Provide the necessary authentication and authorization credentials to initiate the registration process on the Privacy AI Network. This step ensures that only authorized users can access the system and participate in network activities.

• STEP 2: Task submission: Submit the calculation task and input data. The data is encrypted by the HEsea library. At the same time, the data owner also specifies the authorized decryptor and result receiver who can access the final result. By.

• STEP 3: Task Allocation : Blockchain-based smart contracts deployed on the network allocate computing tasks to appropriate Privanetix nodes based on availability and capabilities. This dynamic allocation process ensures efficient resource allocation and distribution of computing tasks.

• STEP 4: Encrypted calculation: The designated Privanetix node receives the encrypted data and uses the HESea library to perform calculations. These calculations can be performed without decrypting sensitive data, thus maintaining its confidentiality. To further verify the integrity of the calculations, Privanetix nodes generate zero-knowledge proofs for these steps.

• STEP 5：密鑰切換：完成計算後，指定的 Privanetix 節點採用密鑰切換技術來確保最終結果是經過授權的，並且只有指定的解密器才能存取。

• STEP 6：結果驗證：完成計算後，Privanetix 節點將加密結果和相應的零知識

  ：完成計算後，Privanetix 節點將加密結果和相應的零知識
• ：完成計算後，Privanetix 節點將加密結果和相應的零知識證明。

  STEP 7：激勵機制

• ：追蹤 Privanetix 節點的貢獻，並分配獎勵

  提取器。他們的首要任務是驗證計算的完整性，確保 Privanetix 節點按照資料所有者的意圖執行了計算。

• STEP 9：結果交付：解密結果與資料擁有者預先確定的指定結果接收者共用。

在Privasea AI NetWork 的核心工作流程中，暴露給使用者的是開放的API，這使得使用者只需關注入參以及相應的結果，而無需了解網路內部複雜的運算本身，不會有太多的心智負擔。同時，端到端的加密在不影響資料處理的前提下，使資料本身不被外洩。

PoW && PoS 雙重機制疊加

Privasea 於近期推出的WorkHeart NFTc進行網路節點管理與獎勵發放。 購買 WorkHeart NFT 即可擁有成為 Privanetix 節點的資格參與網絡計算，並基於PoW機制獲取代幣收益。 StarFuel NFT 是節點增益器（限量5000），可以與 WorkHeart 組合，類似PoS，向其質押的代幣數量越多，WorkHeart節點的收益倍率越大。 那麼，為何是PoW和PoS?其實這個問題比較好解答。

PoW的本質是透過運算的時間成本來降低節點作惡率，維護網路的穩定。 不同於BTC的隨機數驗證的大量無效計算，該隱私計算網路節點的實際工作產出（運算）可以直接與工作量機制的掛鉤，天然適合PoW。

而PoS又更容易平衡了經濟資源。

這樣一來，WorkHeart NFT 透過 PoW 機制獲取收益，而 StarFuel NFT 透過 PoS 機制提高收益倍率，形成了多層次、多樣化的激勵機制，使得用戶可以根據自身資源和策略選擇適合的參與方式。兩種機制的結合，可以優化收益分配結構，平衡運算資源和經濟資源在網路中的重要性。

3.3 小結

由此可見，Privatosea AI NetWork基於FHE建構了一套加密版本的機器學習系統

。得益於FHE隱私計算的特性，把計算任務分包給分散式環境下的各個運算節點（Privanetix），透過ZKP對結果進行有效性驗證，並藉助於PoW和PoS的雙重機制對提供運算結果的節點進行獎勵或懲罰，維護網路的運作。

可以說，Privasea AI NetWork 的設計在為各領域的隱私保護 AI 應用鋪平道路。

4、FHE同態加密-新的密碼學聖杯？

上個章節我們可以看到，Privatosea AI NetWork 的安全性依賴於其底層的FHE，隨著FHE賽道領頭羊ZAMA在技術上的不斷突破，FHE 甚至被投資者冠以新的密碼學聖杯的稱號，讓我們把它與ZKP以及相關的解決方案做比較。

對比下來，可以看到，ZKP與FHE兩者的適用場景區別較大，FHE側重於隱私計算，ZKP側重於隱私驗證。

而SMC似乎與FHE有著更大的重合度，SMC的概念是安全的聯合計算，解決的是共同計算的計算機個體的資料隱私問題。

5、FHE的限制

FHE實現了資料處理權與資料所有權的分離，從而在不影響計算的情況下防止了資料外洩。但同時，犧牲的是運算速度。

加密如同一把雙刃劍，在提升了安全性的同時，導致運算速度大打折扣。

近年來，各種類型的FHE的性能提升方案被提出，有的基於演算法最佳化、有的依靠硬體加速。

• 演算法最佳化方面，新的FHE方案如CKKS和優化的bootstrap方法顯著減少了噪音增長和計算開銷；
• 硬體方面，定制的提升多項式運算的效能。

此外，混合加密方案的應用也在探索之中，透過結合部分同態加密（PHE）和搜尋加密（SE），在特定場景下可以提升效率。

儘管如此，FHE在性能上仍與明文計算有較大差距。

6、總結

Privasea 透過其獨特的架構和相對高效的隱私運算技術，不僅為使用者提供了高度安全的資料處理環境，也開啟了Web3與AI深度融合的新篇章。雖然其底層依賴的FHE有著天然的運算速度劣勢，但是Privasea 近期與ZAMA已經達成了合作，共同攻堅隱私計算的難題。未來，隨著技術上的不斷突破，Privasea 預計將在更多領域發揮其潛力，成為隱私運算和AI應用的探索者。

The above is the detailed content of In-depth interpretation of Privasea, can facial data cast NFT still be played like this?. For more information, please follow other related articles on the PHP Chinese website!