Abstract:Robotic arms are increasingly being used in collaborative environments, requiring an accurate understanding of human intentions to ensure both effectiveness and safety. Electroencephalogram (EEG) signals, which measure brain activity, provide a direct means of communication between humans and robotic systems. However, the inherent variability and instability of EEG signals, along with their diverse distribution, pose significant challenges in data collection and ultimately affect the reliability of EEG-based applications. This study presents an extensible network designed to improve its ability to extract essential features from EEG signals. This strategy focuses on improving performance by increasing network capacity through expansion when learning performance is insufficient. Evaluations were conducted in a pseudo-online format. Results showed that the proposed method outperformed control groups over three sessions and yielded competitive performance, confirming the ability of the network to be calibrated and personalized with data from new sessions.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the reliability and adaptability issues of EEG signals in brain - computer interfaces (BCIs). Specifically, EEG signals have inherent complexity, variability, and distribution diversity, which make EEG - based applications face significant challenges in data collection and performance. In order to improve the accuracy and robustness of EEG signal classification, especially in new sessions, this paper proposes a scalable network architecture to enhance the ability to extract key features from EEG signals. ### Main problems and solutions: 1. **Complexity and variability of EEG signals**: - The inherent complexity and variability of EEG signals require frequent calibration of decoding algorithms. - This variability leads to a wide data distribution, complicating the calibration of new - session datasets and ultimately affecting performance. 2. **Influence of user - related factors**: - Factors such as user fatigue and concentration level will lead to inconsistency in EEG signal generation, making it difficult to extract basic features. 3. **Insufficient performance of shallow convolutional neural networks (CNNs)**: - Shallow CNNs perform poorly when classifying MI (motor imagery) tasks related to similar body movements. To solve these problems, the author proposes a scalable decoding algorithm, aiming to improve EEG - driven robotic arm control by dynamically expanding the network architecture. Specific methods include: - **Network expansion strategy**: Improve performance by increasing network capacity when the network capacity is insufficient. - **Selective initialization and training**: Avoid unnecessary expansion to ensure the effectiveness and efficiency of the model. - **Multi - session evaluation**: The generalization ability and personalization potential of this method are verified through experiments in three sessions. ### Mathematical expressions: The loss function introduces L1 - norm regularization to encourage sparsity and optimize model complexity: \[ L = -\frac{1}{N} \sum_{i = 1}^{N} \hat{y} \log(f(x))+\lambda \sum_{j = 1}^{J} |W_j| \] where \( W_j \) represents the weight of the \( j \) - th layer, \( \lambda \) is the regularization parameter, and \( J \) is the total number of layers in the neural network. When the calculated loss value exceeds the preset threshold, it indicates that the network capacity needs to be increased. At this time, the network weight is expanded by adding nodes to the existing layers: \[ W = [W; \hat{W}] \] where \( W \) represents the existing weight and \( \hat{W} \) represents the weight of the newly added nodes. In addition, filter - level group LASSO regularization is also used to eliminate unnecessary nodes and determine the optimal number of nodes in each layer: \[ \min_W L(W; \hat{W}, \hat{D})+\delta |W|+\delta_g \sum_e ||W_e|| \] In summary, this research significantly improves the accuracy and robustness of EEG signal classification by introducing network expansion strategies, demonstrating its potential in practical applications.

Towards a Network Expansion Approach for Reliable Brain-Computer Interface

Iteratively Calibratable Network for Reliable EEG-Based Robotic Arm Control

Bilateral adaptation and neurofeedback for brain computer interface system.

EEGNet: a compact convolutional neural network for EEG-based brain–computer interfaces

EEG-BBNet: a Hybrid Framework for Brain Biometric using Graph Connectivity

Toward reliable signals decoding for electroencephalogram: A benchmark study to EEGNeX

A novel deep transfer learning framework integrating general and domain-specific features for EEG-based brain–computer interface

EEG-DBNet: A Dual-Branch Network for Temporal-Spectral Decoding in Motor-Imagery Brain-Computer Interfaces

Re-thinking EEG-based non-invasive brain interfaces: modeling and analysis

EEGNet: A Compact Convolutional Network for EEG-based Brain-Computer Interfaces

Less is More: Brain Functional Connectivity Empowered Generalisable Intention Classification with Task-relevant Channel Selection

Network neuroscience for optimizing brain-computer interfaces

BRIEDGE: EEG-Adaptive Edge AI for Multi-Brain to Multi-Robot Interaction

Electrophysiological Brain Network Estimation with Simultaneous Scalp EEG and Intracranial EEG: Inference Algorithm and Applications

RISE-iEEG: Robust to Inter-Subject Electrodes Implantation Variability iEEG Classifier

NeuroAssist: Enhancing Cognitive-Computer Synergy with Adaptive AI and Advanced Neural Decoding for Efficient EEG Signal Classification

Pervasive brain monitoring and data sharing based on multi-tier distributed computing and linked data technology

A Wireless BCI and BMI System for Wearable Robots.

Wearable EEG electronics for a Brain–AI Closed-Loop System to enhance autonomous machine decision-making

RLEEGNet: Integrating Brain-Computer Interfaces with Adaptive AI for Intuitive Responsiveness and High-Accuracy Motor Imagery Classification

On-device Learning of EEGNet-based Network For Wearable Motor Imagery Brain-Computer Interface