Author

tarif mouad

0 works0 citations

Recent research

  • AI & ComputingOpen access

    O-MoE: Optimized Mixture of Experts for Defensive and Specialized AI Systems

    O-MoE is a theoretical framework designed to enhance Mixture of Experts (MoE) architectures by integrating: - **Sequential defensive layers** to counter jailbreak and bias. - **Explicit domain-expert mapping** for specialized tasks. - **Hierarchical Top-K allocation** for adaptiv...

    Zenodo (CERN European Organization for Nuclear Research)2026-08-080 citationsDOI
  • AI & ComputingOpen access

    O-MoE: Optimized Mixture of Experts for Defensive and Specialized AI Systems

    O-MoE is a theoretical framework designed to enhance Mixture of Experts (MoE) architectures by integrating: - **Sequential defensive layers** to counter jailbreak and bias. - **Explicit domain-expert mapping** for specialized tasks. - **Hierarchical Top-K allocation** for adaptiv...

    Zenodo (CERN European Organization for Nuclear Research)2026-08-080 citationsDOI
  • AI & ComputingOpen access

    O-MoE: Optimized Mixture of Experts for Defensive and Specialized AI Systems

    O-MoE is a theoretical framework designed to enhance Mixture of Experts (MoE) architectures by integrating: - **Sequential defensive layers** to counter jailbreak and bias. - **Explicit domain-expert mapping** for specialized tasks. - **Hierarchical Top-K allocation** for adaptiv...

    Zenodo (CERN European Organization for Nuclear Research)2026-08-080 citationsDOI