Figure 3 | Scientific Reports

Figure 3

From: Brain-inspired spiking neural networks for decoding and understanding muscle activity and kinematics from electroencephalography signals during hand movements

Figure 3

Results of predicting the object and hand kinematics using BI-SNN and GLM approaches. (A) The maximum coefficient of the cross-correlation sequence between the actual and predicted kinematic signal by BI-SNN. (B) The maximum cross-correlation coefficients between the actual and predicted kinematic signal by GLM. (C) Comparison between the BI-SNN and GLM methods on predicting the object and hand kinematics. (D) The statistical distribution of the cross-correlation coefficients within the participant group. The correlation analysis shows that the BI-SNN results in a ‘moderate’ to ‘high’ cross-correlation \((0.6 \le r \le 0.7)\) in predicting x, y, z Cartesian coordinates. (E) The actual kinematic signals and the spike response of a single neuron in SPANindex-x, SPANwrist-y, SPANwrist-z and SPANthumb-azimuth populations during a Grasp-and-Lift trial.

Back to article page