Genetic Algorithm-Based Data Optimization for Efficient Transfer Learning in Convolutional Neural Networks: A Brain–Machine Interface Implementation

Robotics, Journal Year: 2024, Volume and Issue: 13(1), P. 14 - 14

Published: Jan. 15, 2024

In brain–machine interface (BMI) systems, the performance of trained Convolutional Neural Networks (CNNs) is significantly influenced by quality training data. Another issue time CNNs. This paper introduces a novel approach combining transfer learning and Genetic Algorithm (GA) to optimize data Transfer implemented across different subjects, chosen GA aim improve CNN performance. addition, GA-selected shed light on similarity in brain activity between subjects. Two datasets are used: (1) publicly available BCI Competition IV, which subjects performed motor imagery (MI) tasks, (2) dataset created healthy our laboratory performing movement (MO) tasks. The experimental results indicate that selected recognition accuracy target (TCNN) using pre-trained base (BCNN). improvement 11% 4% for IV datasets, respectively. reduce time. CNN, utilizing learning, tested real-time control robot manipulator.

Language: Английский

---

Investigation of the Kinetostatics and Stress-Strain State of the Links of the im Exoskeleton of the Lower Limb

Mechanisms and machine science, Journal Year: 2024, Volume and Issue: unknown, P. 315 - 323

Published: Jan. 1, 2024

Language: Английский

---

The autistic experience revealed through digital phenotyping

Elsevier eBooks, Journal Year: 2024, Volume and Issue: unknown, P. 185 - 245

Published: Oct. 1, 2024

Language: Английский

---

Clinical Feasibility of Applying Immersive Virtual Reality during Robot-Assisted Gait Training for Individuals with Neurological Diseases: A Pilot Study

Brain Sciences, Journal Year: 2024, Volume and Issue: 14(10), P. 1002 - 1002

Published: Oct. 2, 2024

Background: Immersive virtual reality has the potential to motivate and challenge patients who need want relearn movements in process of neurorehabilitation. Objective: The aim this study was evaluate feasibility user acceptance an innovative immersive system (head-mounted display) used combination with robot-assisted gait training subjects suffering from neurological diseases. Methods: Fifteen participants cerebrovascular accident or spinal cord injury completed a single session using head-mounted display during Lokomat® session. Training parameters safety indicators were collected, investigated among therapists. Results: results suggest that is feasible terms tolerance. Furthermore, very positive overall suggests it be included Lokomat®. Conclusion: Overall, demonstrates fully based on both well received by their therapists training. This such could viable alternative screen-based games currently It may especially suitable for enhancing patient motivation adherence training, particularly if application enjoyable not mentally taxing.

Language: Английский

---

Lower limb muscle activity during neurointerface control: neurointerface based on motor imagery of walking

Журнал высшей нервной деятельности им И П Павлова, Journal Year: 2024, Volume and Issue: 74(5), P. 591 - 605

Published: Nov. 27, 2024

Language: Английский

---