ZKTeco Announces Publication of "BrainScape" Research Paper: Pioneering Data Security and Collaborative Computing Architectures in Brain-Computer Interface (BCI)

ZKTeco Co., Limited (ZKTeco) is pleased to announce that its wholly-owned subsidiary, ZKBrainWave, has published its latest research findings titled “BrainScape: Research on Sovereign Cloud Architecture of Brain-Computer Interface Based on Distributed Weight Merging and DARE Algorithm.” The paper has been featured in the international academic journal Biomedical Data Science (2026, Vol. 6, No. 1, ISSN: 2752-6305 / 2753-4839).

The study addresses critical bottlenecks in the large-scale application of Brain-Computer Interface (BCI) technology, specifically focusing on data security, model generalization, and distributed collaboration. It proposes a technical framework based on "Sovereignty Cloud" architecture, exploring the integration of federated learning, privacy protection, and edge-to-cloud synergy within BCI contexts.

This publication follows the official debut of the BrainScape BCI Sovereignty Cloud Platform and its accompanying white paper, which were unveiled earlier this year at the Change Maker Brain-Computer Interface (China) Launch hosted by ZKTeco on March 28.

Addressing Real-World Challenges in BCI Scalability

As non-invasive BCI technology expands into fields such as educational focus training, neurological rehabilitation, and emotional wellness, the industry faces heightened demands for data security, model adaptability, and compliant deployment.

The research identifies three primary industry-wide challenges:

1. Data Sensitivity: EEG (electroencephalogram) data is highly personal; traditional centralized processing creates significant pressure regarding privacy and data flow.

2. Individual Variability: Significant differences in EEG signals between users hinder the cross-individual generalization of models.

3. The Security-Performance Trade-off: Privacy mechanisms often incur high computational costs, making it difficult to balance robust security with high-performance model output.

Core Research Pillars of the BrainScape Architecture

To resolve these issues, the paper introduces the BrainScape architecture, centered on Distributed Sovereignty Cloud + Federated Learning Optimization + Low-Rank Adaptation. The goal is to enhance collaborative training and cross-user adaptation without data ever leaving the local domain.

Key research directions include:

Distributed Sovereignty Cloud Architecture: Utilizing edge-to-cloud synergy, user EEG data is processed locally. Only encrypted model parameters are synchronized and aggregated, significantly mitigating privacy risks associated with cross-domain raw data transfer.

Distributed Weight Merging Mechanism: To address discrepancies in model updates across different devices and users, the study proposes a weight aggregation method based on Low-Rank Bias adapters, improving model generalization in diverse user scenarios.

DARE Privacy Protection Algorithm: By combining Zero-Knowledge Proofs with federated learning mechanisms, the paper optimizes defenses against privacy risks such as membership inference attacks. Experimental data indicates a measurable reduction in attack success rates while maintaining stable model performance.

According to the study's experimental results, the BrainScape architecture demonstrates a marked improvement in model stability and generalization across various users and task scenarios.

Strengthening the Foundation of BCI Infrastructure

As a global provider of smart space evolution services driven by AI cognition, ZKTeco has long specialized in multimodal BioCV (Biometrics and Computer Vision) and AI-driven spatial computing. Through ZKBrainWave, the company has proactively positioned itself in the BCI sector, establishing a full-chain closed-loop system encompassing "Chips - Algorithms - Sovereignty Cloud - Applications."

On the software and services front, ZKTeco integrates NeuroSky’s biosensing technology with its own Minerva IoT platform to develop the Emotion Cloud Service Platform. This platform provides industry partners and end-users with specialized emotional recognition, management, and intelligent service solutions through continuous cloud-based algorithmic iterations.

ZKTeco emphasizes that this research is primarily directed toward education, rehabilitation, and emotional health, with a core focus on data compliance, deployment efficiency, and cross-terminal compatibility. As BCI transitions from scientific exploration to industrial application, the development of underlying capabilities in privacy computing and model synergy is becoming a strategic priority.

Public records indicate that ZKTeco has already secured a robust patent portfolio in emotion analysis and is currently validating these solutions in real-world scenarios.

Chong-liang Zhong, Head of ZKTeco's Intelligent Innovation Research Institute, commented: "The BrainScape research is centered on securing data and optimizing model collaboration within BCI environments. Moving forward, ZKTeco will continue to align our research with market demands, conducting rigorous validation in privacy protection, model generalization, and system-wide synergy."