7-1. Summary


The architecture of EMETH enables generalization of processing and is designed to decentralize various processes other than language model learning. The core EMETH Orchestrator Server architecture comprises the following principal components:

  1. Orchestrator: This is a central function that monitors information from the Cache server and manages and coordinates processes.

  2. Splitter: This function encrypts the jobs to be executed and splits them into a form suitable for distributed processing.

  3. Aggregator: This function collects and integrates learning completed models in a dispersed encrypted state.

  4. Verifier: This function verifies whether the results of jobs obtained through distributed processing are valid.

With these main components working together, the architecture of EMETH realizes the generalization and decentralization of processing. This architecture has the flexibility to decentralize various processes, not just language model learning, and can effectively perform large-scale processing in a wide range of fields, such as AI learning and data analysis.

From the next page, we will explain the detailed functions and roles of each component. We will delve deeper into important elements that support the architecture of EMETH, such as process management by the Orchestrator, secure distributed processing by the Splitter, integration of results by the Aggregator, and verification of results by the Verifier.

Last updated