🤖
Alaya AI
  • 🍀Overview
  • Core Features
    • Distributed Data Ecosystem
    • Open Data Platform
    • Data Auto-Labelling Toolset
  • User Manual
    • How to start
    • Task Categories
    • $AGT Token
    • NFT System
    • Auto-Labelling Model Staking Pools
  • Technology
    • Three-Layer Optimisation Architecture
    • Dynamic Visual Data Segmentation and Discrete Tracking
Powered by GitBook
On this page
  1. Core Features

Distributed Data Ecosystem

Alaya AI utilises a distributed data sampling structure that connects global data provider communities through an integrated Web3 data infrastructure network. Alaya AI’s standard data business process follows a streamlined and highly automated sequence:

  • AI training questions are provided by data customers and added to Alaya’s question bank based on each customer’s specific data requirements.

  • Alaya AI automatically distributes training tasks from our extensive AI/ML question bank to suitable users for labelling and annotation based on user records and expertise (e.g., verification rates in each area of expertise).

  • Training tasks are completed by a distributed network of individual contributors through targeted sampling algorithms optimised through HITL-assisted AI model fine-tuning.

  • Automated data preprocessing and quality assessment is applied before final delivery based on customer data specifications. ZK-encryption and data desensitisation are applied to ensure minimal privacy risks for data contributors.

Data quality is verified by preprocessing algorithms on Alaya AI’s Optimisation Layer through Gaussian approximation and particle swarm optimisation, while sampling bias is minimised through Alaya AI’s large and diverse user communities.

PreviousCore FeaturesNextOpen Data Platform

Last updated 7 months ago