Research
Explore papers and preprints from the FLock team on federated learning, decentralized AI, privacy, and reliable language models.
FeaturedResearch
WI-IAT · 2025
Scaling Decentralized Learning with FLock
Presents a decentralized framework for collaborative fine-tuning of 70B-parameter language models and reports improved robustness against poisoning attacks.
Read paper (opens in a new tab)IEEE GBC · 2025
Multi-Continental Healthcare Modelling Using Blockchain-Enabled Federated Learning
Uses blockchain-enabled federated learning to model glucose management from datasets across Europe, North America, and Asia without sharing local data.
Read paper (opens in a new tab)AllPublications
Showing 16 of 16 publications
Bittensor Protocol: The Bitcoin in Decentralized Artificial Intelligence? A Critical and Empirical Analysis
Analyzes stake and reward concentration in Bittensor and evaluates protocol changes intended to better align incentives and reduce concentration risk.
CircuitSynth: Reliable Synthetic Data Generation
A neuro-symbolic framework that uses a probabilistic sentential decision diagram to enforce logical constraints during structured synthetic data generation.
A Control Theoretic Approach to Decentralized AI Economy Stabilization via Dynamic Buyback-and-Burn Mechanisms
Proposes a control-theoretic buyback mechanism with solvency constraints to address volatility in decentralized AI token economies.
Identity-Decoupled Anonymization for Visual Evidence in Multi-modal Retrieval-Augmented Generation
Decouples identity signals from visual evidence used in retrieval-augmented generation pipelines.
GraphSynth: Resolving the Diversity-Reliability Trade-off with Probabilistic Factor Graphs
Uses a probabilistic factor graph, decoding-time schema constraints, and verification to balance diversity and factual reliability in synthetic data.
SoK: Blockchain-Based Decentralized AI (DeAI)
Systematization of knowledge survey of the decentralized AI field. Covers protocol design, incentives, and open challenges.
Vision: How to fully unleash the productivity of Agentic AI? Decentralized Agent Swarm Network
Outlines AgentaNet as a vision for decentralized agent swarms, with verifiable profiles, task coordination, privacy-preserving communication, and incentives.
Multi-Continental Healthcare Modelling Using Blockchain-Enabled Federated Learning
Uses blockchain-enabled federated learning to model glucose management from datasets across Europe, North America, and Asia without sharing local data.
Scaling Decentralized Learning with FLock
Presents a decentralized framework for collaborative fine-tuning of 70B-parameter language models and reports improved robustness against poisoning attacks.
On Weaponization-Resistant Large Language Models with Prospect Theoretic Alignment
Prospect-theoretic alignment technique to make large language models robust against weaponization attempts.
Privacy Preserved Blood Glucose Level Cross-Prediction: An Asynchronous Decentralized Federated Learning Approach
Asynchronous decentralized federated learning for cross-prediction of blood glucose levels while preserving patient privacy.
AIArena: A Blockchain-Based Decentralized AI Training Platform
A blockchain-native decentralized AI training platform positioning on-chain coordination as the substrate for collaborative training.
ZkFL: Zero-Knowledge Proof-Based Gradient Aggregation for Federated Learning
Zero-knowledge proof-based gradient aggregation that delivers verifiable privacy in federated learning at scale.
Multi-modal Feature Fusion Networks for GeoLifeCLEF 2024
Combines remote-sensing and geospatial data with multimodal feature fusion and ensemble models to predict plant species distributions.
Defending Against Poisoning Attacks in Federated Learning With Blockchain
Blockchain-based defense against poisoning attacks in federated learning. Published in IEEE TAI.
FLock: Defending Malicious Behaviors in Federated Learning with Blockchain
Combines federated learning with blockchain to detect and defend against malicious client behaviors during training rounds.
Learn Moreabout FLock
FLock: Federated Machine Learning on Blockchain
Democratising AI through Decentralisation of Data, Computation and Models
Federated Machine Learning on Blockchain
Facilitating an open and collaborative environment where participants can contribute models, data, and computing resources, in exchange for on-chain rewards based on their traceable contributions.
UnrivaledResearch and Groundwork
