SEGFAULT

• 

  SEGFAULT_logo

• 

  front_end_demo

• 

  picking_up_cup_gen_data

To combat the scarcity of EEG data, S.E.G.F.A.U.L.T. - Synthetic EEG Generator For Automatic Utility / Learning / Training generates synthetic electroencephalogram (EEG) data.

Current neuroscience studies involve real-time EEG collection, pairing recorded brain activity with specific images, actions, emotions, or spoken words. However, the scarcity of accessible EEG datasets limits broader research - gathering data can be costly and time-consuming. To tackle this problem, we reversed the typical data flow by training machine learning models to generate EEG data.

By utilizing Generative Adversarial Networks (GANs), SEGFAULT simulates EEG outputs corresponding to specific brain signals. With the current implementation, users can select specific physical tasks (e.g., right/left hand/leg movement) and then generate synthetic EEG data for download and analysis corresponding to the brainwave activity during the selected task. This can be downloaded to incorporate into any application or study.

To further expand this tool, it could generate other types of EEG data corresponding to images, emotions, spoken words, and more. Since the data is trained from random noise, less participants are needed, saving time and money.

This tool will be a significant asset for researchers, allowing the creation of large-scale, high-quality EEG datasets without extensive and costly data collection. The synthetic EEG data can be used for training, validating, and enhancing research. By leveraging this system, this project aims to solve the issue of limited EEG data amidst the growing demand for data-driven neuroscience research.

Built With

• python
• tensorflow
• tkinter