Covariance Density Neural Networks
This work addresses the challenge of transferability in Brain-Computer Interfaces for unseen individuals, though it is incremental as it builds on existing Covariance Neural Networks.
The paper tackles the problem of selecting graph structures for graph neural networks by proposing Covariance Density Neural Networks, which use a density matrix derived from the sample covariance matrix as a Graph Shift Operator to improve performance, achieving strong results in EEG motor imagery classification and outperforming EEGnet while being faster.
Graph neural networks have re-defined how we model and predict on network data but there lacks a consensus on choosing the correct underlying graph structure on which to model signals. CoVariance Neural Networks (VNN) address this issue by using the sample covariance matrix as a Graph Shift Operator (GSO). Here, we improve on the performance of VNNs by constructing a Density Matrix where we consider the sample Covariance matrix as a quasi-Hamiltonian of the system in the space of random variables. Crucially, using this density matrix as the GSO allows components of the data to be extracted at different scales, allowing enhanced discriminability and performance. We show that this approach allows explicit control of the stability-discriminability trade-off of the network, provides enhanced robustness to noise compared to VNNs, and outperforms them in useful real-life applications where the underlying covariance matrix is informative. In particular, we show that our model can achieve strong performance in subject-independent Brain Computer Interface EEG motor imagery classification, outperforming EEGnet while being faster. This shows how covariance density neural networks provide a basis for the notoriously difficult task of transferability of BCIs when evaluated on unseen individuals.