Please use this identifier to cite or link to this item:
|Title:||Embedded machine learners for classifying EEG patterns in a wearable device for brain-computer interface||Authors:||Lim, Kok Meng||Keywords:||DRNTU::Engineering||Issue Date:||2016||Abstract:||Tasks that seems trivial for a healthy and able-bodied individual such as text messaging or even speaking may present as a challenge for people with neuro-muscular disabilities. With the advent of brain computer interface (BCI), it raises hope in helping this group of people to perform simple tasks which were once thought overwhelmingly not possible. Invasive BCI system are based on direct implantation of device into the brain. While it produces the highest quality signals, formation of scar-tissue is high likely due to the device having direct contact with the brain tissue. Hence electroencephalograms (EEG) is considered the most popular solution as a non-invasive system. However, the process of translating EEG signals into computer commands is a sophisticated task that requires optimising various parameters that are tuned independently or simultaneously. In this project, emphasis is placed on the EEG-based BCIs that rely on steady state visual evoked potentials (SSVEP). With the datasets of 11 subjects containing the SSVEP (known labels), the tasks involve performing signal processing on these EEG-SSVEP signals before passing it into the Logistic Regression (LR) and Multi Layer Perceptron (MLP) classifiers respectively. The datasets are portioned in accordance to leave-one-subject-out cross validation (LOSO CV) strategy to train the classifiers in learning these datasets before evaluating (testing) the degree of accuracy in the classifiers learning. Comparative evaluations on the usage of various algorithms were also made in assessing the effectiveness of optimising the classification accuracy. Parameters tuning on the classifiers algorithm were also done to study the effect it has on these classification accuracy results. Keywords: Steady State Visual Evoked Potentials, Logistic Regression, Multi Layer Perceptron, Leave-One-Subject-Out Cross Validation, Comparative Evaluation, Classification Accuracy, Parameters Tuning||URI:||http://hdl.handle.net/10356/69264||Rights:||Nanyang Technological University||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Student Reports (FYP/IA/PA/PI)|
Page view(s) 5089
checked on Oct 28, 2020
checked on Oct 28, 2020
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.