Learning Residual Kinematic Corrections for Continuous Neural Decoding via Reinforcement Learning

Source

Source name: arXiv
Source kind: paper
Authors / organization: Jiamian Li, Niall McShane, Attila Korik, Naomi du Bois, Karl McCreadie, Leen Jabban, Benjamin Metcalfe, Özgür Şimşek, Damien Coyle
Venue: arXiv: cs.AI
Published: 2026-07-13
Collected: 2026-07-14
URL: http://arxiv.org/abs/2607.11530v1
PDF: https://arxiv.org/pdf/2607.11530v1
DOI: UNKNOWN

Plain-Language Brief

A new method using reinforcement learning (RL) has been proposed to improve the accuracy of decoding continuous three-dimensional motor imagery signals from non-invasive EEG data for brain-computer interfaces (BCIs). The method involves two stages: first, using a CNN-LSTM model for kinematic decoding, and second, applying an RL agent to correct residual errors in the predicted trajectories. The proposed method significantly improved the correlation coefficient and reduced the Root Mean Square Error compared to using CNN-LSTM alone, demonstrating its potential to advance neurorehabilitation, prosthetics, and virtual interaction.

What the Document Says

The paper presents a two-stage decoding framework for continuous 3D motor imagery (MI) using non-invasive EEG-based BCIs.
The first stage involves using CNN-LSTM models for kinematic decoding.
The second stage applies reinforcement learning (RL) to correct residual errors in the predicted trajectories.
The RL agent is trained offline and operates on predicted kinematic trajectories.
Decoding performance was quantified using Pearson correlation coefficients ($r$) and Root Mean Square Errors (RMSE) along the $x, y$, and $z$ axes.

Key Concepts Mentioned

3D motor imagery
non-invasive EEG-based BCIs
CNN-LSTM models
reinforcement learning
Pearson correlation coefficients
Root Mean Square Errors

Claims Made by the Document

The proposed method enhances 3D BCI MI decoding.
The proposed method reduces residual errors in predicted trajectories.

Evidence Used

Improved mean correlation from $0.5076$ to $0.7181$ ($p = 0.0005$) in 2D and from $0.6420$ to $0.7780$ ($p = 0.0059$) in VR.
Reduced Root Mean Square Error from $0.0890$ to $0.0532$ (2D, $p < 0.0001$) and from $0.0714$ to $0.0441$ (VR, $p < 0.0001$).

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Decoding continuous three-dimensional (3D) motor imagery (MI) using non-invasive electroencephalography (EEG)-based brain--computer interfaces (BCIs) remains challenging due to signal variability and residual decoding errors. Deep learning architectures such as convolutional neural network--long short-term memory (CNN--LSTM) models can capture spatial and temporal dynamics for continuous kinematic decoding; however, systematic residual errors persist in predicted trajectories. We propose a two-stage decoding framework that applies reinforcement learning (RL) to perform residual kinematic correction on the outputs of a CNN--LSTM decoder (CNN--LSTM--RL). The RL agent is trained offline without direct EEG input and instead operates on predicted kinematic trajectories to optimize movement accuracy relative to target trajectories. Decoding performance was quantified using Pearson correlation coefficients ($r$) and Root Mean Square Errors (RMSE) along the $x, y$, and $z$ axes. Compared to CNN--LSTM applied alone, CNN--LSTM--RL improved the mean correlation from $0.5076$ to $0.7181$ ($p = 0.0005$) in 2D and from $0.6420$ to $0.7780$ ($p = 0.0059$) in VR, with relative gains of $41.5%$ and $21.2%$, respectively. Correspondingly, RMSE was reduced from $0.0890$ to $0.0532$ (2D, $p < 0.0001$) and from $0.0714$ to $0.0441$ (VR, $p < 0.0001$), representing relative reductions of $40.2%$ and $38.2%$. These findings demonstrate that this scalable framework enhances 3D BCI MI decoding by correcting kinematic errors via offline residual RL without extra neural data, advancing neurorehabilitation, prosthetics, and virtual interaction.