Inferix Decentralized GPU
  • Getting Started
    • Overview
    • $IFX
    • Resources
    • Brand Kit
    • Frequently asked questions (FAQs)
  • Inferix Whitepaper
    • Introduction
      • Rendering network using crowdsourced GPU
      • Rendering verification problem
    • High-level description of ANGV
      • Noise generation
      • Noise verification
      • Thread model
    • Implementation of ANGV
      • Structure of noise
      • Noise insertion
        • Geometric constraints
        • Distortion region
      • Adaptive noise spreading
      • Verification key generation
      • Noise verification
      • Threat analysis
        • Attacks on verification keys
        • Attacks on noises
        • Attacks on verifiers
      • Performance evaluation
      • Integration
    • Decentralized visual computing
      • Client Apps plugin
      • Client API and SDK
      • Manager node
      • Worker node
      • Decentralized storage
        • Data categories
        • Multi-level 3D polygon data
        • Polygon digester
        • Decentralized storage
        • Decentralized cache
      • Data security with FHE and TEE
        • Verifier data security enhancement with FHE
        • Worker and Manager data security enhancement with FHE
    • Decentralized federated AI
      • Federated learning with TensorOpera
      • Meta LLaMA
      • Stable Diffusion
      • Other AI models
      • Inferix AI
    • Economic model
      • GPU compute market for visual computing and federated AI
      • Inferix vision
      • $IFX token
      • Burn-Mint-Work token issuance model
      • Inferix bench and IBME
        • IB and IBM
        • IBME
      • Price simulation
      • Token metrics and allocation
        • Token allocation
        • Token vesting
      • Governance
      • Node staking and rewards
        • Worker
        • Verifier
        • Manager
        • Penalty pool
      • Node sale and guaranteed node buyback
        • Node sales
        • Guaranteed Node Buyback
    • Future development
      • PoR and NFT minting for graphics creative assets
      • ZKP and PoR communication
      • Inferix RemotePC
      • Rendering professional network
    • References
    • Appendix A: Proofs
    • Appendix B: Price simulation details
    • Appendix C: Hardware requirements for nodes
    • Appendix D: Performance evaluation data
  • Worker Node Guide
    • What is Worker Node
      • How do the Worker Node work
      • Worker Node Rewards
      • How to run Worker Node
      • What is the Worker Node License (NFT)
    • Worker Node Sales
      • Guide to Purchase Worker Nodes
      • Worker Node Sale Timeline
      • Node Supply, Price, Tiers and Purchase Caps
      • Guaranteed Node Buyback
      • How to get Node Whitelisted?
      • Smart Contract Addresses
      • User Discounts & Referral Program
      • Worker Node Purchase FAQ
      • ABKK Collaboration FAQ
  • Verifier Node Guide
    • What is Verifier Node
      • How do the Verifier Node work
      • Verifier Node Rewards
      • How to run Verifier Node
      • What is the Verifier Node License (NFT)
    • Verifier Node Sales
      • Guide to Purchase Verifier Nodes
      • Verifier Node Sale Timeline
      • Node Supply, Price, Tiers and Purchase Caps
      • Guaranteed Node Buyback
      • How to get Node Whitelisted?
      • Smart Contract Addresses
      • User Discounts & Referral Program
      • Verifier Node Purchase FAQ
      • Aethir Node Winners FAQ
  • Inferix MVP
    • Tutorial: MVP for designers & GPU owners
    • PoR MVP
  • Inferix Testnet 2 on Solana & IoTeX [ENDED]
    • Adding GPUs to the Network
      • For GPU providers
      • For GPU providers without funds
      • For users without GPUs
      • For Inferix Node Holders
    • Renting GPU Devices
    • User Revenue Calculation
      • Worker Rewards
      • Rental Revenue
      • Viewing Revenue
      • Claiming Rewards
    • GPU Staking & Unstaking
      • Staking Requirements
      • Unstaking GPUs
    • Guide to get tIFX tokens
    • Why choose Inferix DePIN GPU Solutions?
  • Inferix Testnet 1 on IoTeX [ENDED]
    • Inferix GPU Solutions
    • Adding GPUs to the Network
    • Renting GPU Devices
    • User Revenue Calculation
    • GPU Staking
    • Multiple options to participate in the Staking & Mining Program
    • Special airdrop for Inferix Node Holders! 🎉
    • Guide to get tIFX tokens
    • FAQ
  • Inferix Explorer
  • Team & Achievements
    • Our Story
    • Team
    • Member of Cohort 1 DePINSurf
    • Achievements
  • Community & Events
    • Events
    • Inferix Campaign: "ALLIANCE" (ENDED)
  • Terms of Service
    • Privacy Policy
    • Airdrop Terms of Service
Powered by GitBook
On this page
  1. Inferix Whitepaper
  2. Implementation of ANGV

Structure of noise

As introduced in the previous section, a noise WWW is a random vector (wi)1≤i≤n\left(w_i\right)_{1 \leq i \leq n}(wi​)1≤i≤n​ where each element is independently chosen from a normal distribution. This atomic watermark is constructed as a rectangular image of periodic patterns as follows:

  • let fix some values M,NM,NM,N for the width and the height of the rectangle, and

  • let X_i,Y_i\mathcal{X}\_i, \mathcal{Y}\_iX_i,Y_i be independent and identically distributed normal random variables:

    Xi∼Yi∼N(μ,σ2)\mathcal{X}_i \sim \mathcal{Y}_i \sim \mathcal{N}\left(\mu, \sigma^2\right)Xi​∼Yi​∼N(μ,σ2)

    for some μ\muμ and σ\sigmaσ, then take Xi,YiX_i, Y_iXi​,Yi​ be respectively some samples of Xi,Yi\mathcal{X}_i, \mathcal{Y}_iXi​,Yi​.

The complex atomic signal wiw_iwi​ is defined by:

wi(x,y)=Ae2iπ(xXi+yYi) (0≤x<M,0≤y<N)w_i(x,y) = A e^{2i\pi\left(\frac{x}{X_i} + \frac{y}{Y_i}\right)} \ \left(0 \leq x < M, 0 \leq y < N \right)wi​(x,y)=Ae2iπ(Xi​x​+Yi​y​) (0≤x<M,0≤y<N)

for some amplitude AAA. We observe that wi(x,y)=wi(x+Xi,y)=wi(x,y+Yi) ∀x,yw_i\left(x,y\right) = w_i\left(x + X_i, y\right) = w_i\left(x, y + Y_i\right) \, \forall x,ywi​(x,y)=wi​(x+Xi​,y)=wi​(x,y+Yi​)∀x,y then Xi,YiX_i, Y_iXi​,Yi​ are actually the horizontal and the vertical periods.

Remark: Xi\mathcal{X}_iXi​ and Yi\mathcal{Y}_iYi​ are elements of a set {Xi,Yi∣1≤i≤n}\left\{ \mathcal{X}_i, \mathcal{Y}_i \mid 1 \leq i \leq n\right\}{Xi​,Yi​∣1≤i≤n} of independent and identically distributed normal random variables N(μ,σ2)\mathcal{N}\left(\mu, \sigma^2\right)N(μ,σ2). The parameters μ\muμ and σ\sigmaσ are chosen by analyzing the input scene that is discussed in~\cref{subsec:noise_spreading}.

Proposition 1. (Fourier transform of complex atomic signals)

Fi(u,v)=AM×N(1−e2iπMXi)(1−e2iπNYi)(1−e2iπ(1Xi+uM))(1−e2iπ(1Yi+vN))F_i\left(u,v\right) = \frac{A}{M \times N} \frac{\left(1 - e^{2i\pi \frac{M}{X_i}}\right) \left(1 - e^{2i\pi \frac{N}{Y_i}}\right)}{\left(1 - e^{2i\pi\left(\frac{1}{X_i} + \frac{u}{M}\right)}\right) \left(1 - e^{2i\pi\left(\frac{1}{Y_i} + \frac{v}{N}\right)}\right)}Fi​(u,v)=M×NA​(1−e2iπ(Xi​1​+Mu​))(1−e2iπ(Yi​1​+Nv​))(1−e2iπXi​M​)(1−e2iπYi​N​)​

The length nnn of the noise vector is one of the principal factors which decides the robustness of noise: the higher the value nnn, the lower the false positive of noise verification. But this size influences the quality of the rendered image: the lower value nnn, the higher fidelity of the rendered images. Consequently, the value nnn is a trade-off between the robustness of the embedded noise and the fidelity of the rendered image, it is empirically chosen to be about 8 to 15.

Figure 4:

PreviousImplementation of ANGVNextNoise insertion

Last updated 8 months ago

Proof: Direct calculation (for details, see ).

The given structure of noise has two folds: we empirically find that this form of signal makes the wrapping graphical objects (discussed in the ) persistent in the rendering of graphics scenes. Furthermore, the distortion raised by any atomic watermark is easily controlled thanks to the simple form of the signal amplitude given in the proposition.

The spectrums of atomic signals play a crucial role in the noise verification since they help to distinguish embedded noises from the original image signals. They are also completely determined by the periods Xi,YiX_i, Y_iXi​,Yi​ given fixed M,NM,NM,N since the discrete Fourier transform in the proposition. In turn, these periods statistically rely on the expectation μ\muμ, we will discuss how to choose this value in the .

shows a noise as a vector of 555 atomic watermarks and the Fourier transforms showing the corresponding frequency characteristics. The vital frequencies of energy are clearly shown in the spectrums. For illustration purpose, we take Xi=Yi∼N(25,5) (1≤i≤5)X_i = Y_i \sim \mathcal{N}\left(25,5\right) \ \left(1 \leq i \leq 5\right)Xi​=Yi​∼N(25,5) (1≤i≤5), and M=N=512M = N = 512M=N=512.

Appendix 1
noise insertion
noise spreading
Figure 4
A random vector of 5 atomic watermarks