2. PRIOR WORK
• Vacation Research Project from 2014
• Aim was to test classifier on Gaming Peripheral EEG Headset
• Emotive EPOC combined with python scripting
3. INVESTIGATION INTO PRIOR WORK
• 3 Separate tests (involving cubes and movement)
• 3 main subjects
• 2 Stage test (initial condition, final condition)
• Using all 14 channels
• Separate Classifier (separate script for analysis of recordings)
• Linear Classifiers – Failed
• Non-Linear Classifiers – Passed (barely)
• Poor Channel Quality
4. PROJECT DEFINITION - EXPERIMENTS
• Group Experiment
• Human motor control (such as lifting an arm) with respect to visualisation of the
same motor control (visualisation of lifting that arm)
• Individual Experiment 1
• Conducted by Lucas Nielsen
• Analysing the effect of brain activity with respect to directional movement cues
• Individual Experiment 2
• Conducted by Thomas Parker
• Analysing the effect of brain activity with respect to musical tunes
• Individual Experiment 3
• Conducted by Michael Argyris
• Analysing the effect of brain activity with respect to colour and shapes
5. PROJECT DEFINITION – GROUP EXPERIMENT
• Hypothesis
• Visualisation of motor control and actualising motor control both produce EEG
readings which are distinguishable and are able to be discretely analysed to
determine what specific motor control was performed
• Variables
• Individuals with varying socioeconomic status and varying passions and hobbies
• Time of day test taken
• Subjects dominant hand (left or right handed)
• Type of input (physical or imaginary)
• Different movement types (throw, catch, wave, handshakes)
6. PROJECT DEFINITION – INDIVIDUAL
EXPERIMENT 1
• Hypothesis
• Emotiv EPOC system is able to accurately detect and successfully identify
features associated with particular brain activity when a given a subject
demonstrates the ability to respond to directional movement cues
• A Scenario
• Observing a static black arrow directed upward
• Observing a static black arrow directed downward
• Observing a static black arrow directed towards the right
• Observing a static black arrow directed towards the left
• Repeat 1 to 4 for observing a moving black arrow
7. PROJECT DEFINITION – INDIVIDUAL
EXPERIMENT 2
• Hypothesis
• Emotiv EPOC system is able to accurately detect and successfully identify the
difference between an “earworm” or the ability of Audiation as opposed to
actually listening to music.
• A Scenario
• Listen to song prior to experiment several times
• Playback of song whilst wearing EEG Headset & Recording Data
• Attempt recall of chorus or major section of song and ask to record which part
of song was recalled
• Test for correlations between both sets of data
8. PROJECT DEFINITION – INDIVIDUAL
EXPERIMENT 3
• Hypothesis
• The Emotiv EPOC system is able to accurately detect and successfully identify the
difference between shapes and colours when visualised.
• A Scenario
• Part 1 involves looking at uniquely coloured shapes will generate highly distinguishable
readings
• Part 2 involves having three unique shapes of the same colour
• Part 3 involves having three identical shapes of different unique colours
9. METHODOLOGY - PRECAUTIONS
• Reasonably lit room with minimal sound to reduce any possible external
influence to noise generation.
• Five (5) minutes per subject to relax and stabilise energy levels to reduce
possible unwanted noise
• Ensure the subject is fully aware of the steps and time involved in completing
the experiment
• Clean sensors, lubricate thoroughly and position correctly on subjects head
• Each experiment is to be completed by multiple subjects (Approx. 6)
10. METHODOLOGY – EQUIPMENT REQUIRED
• Emotiv EPOC headset and dongle
• A computer containing the software required for performing the test
• Both of these are being sourced personally as the EPOC has been provided
by QUT to perform the experiments required.
11. METHODOLOGY - PROCEDURE
• Software Setup:
• Software provided previously (EmoKit found on GitHub) is downloaded and installed on a
windows based computer
• Emotiv EPOC control panel is opened and the headset with Bluetooth dongle is initialised.
• The application requires two inputs before EEG recording can start:
• Name for the excel file of which recorded data is exported to
• Time (in seconds) for the particular recording to be run.
12. METHODOLOGY - PROCEDURE
• Environment Setup:
• Setup the headset by applying the saline solution to each sensor
• Attaching the sensors to the headset itself
• Position the EPOC headset and sensors on the subject’s head and achieve strong signal
readings (indicated by green lights on each sensor point on the control panel diagram)
• Load the experiment stimulus material onto the laptop
• Sit the test subject with eye level in line with the screen
• Proceed to start the experiment and collect data
13. METHODOLOGY - PROCEDURE
• Data Collection (Single Experiment):
1. Perform software and environment setup
2. Determine which experiment scenario is to be carried out
3. Enter a name for the recorded data file to be exported as and enter the duration of the
sampling time (in seconds) for the particular recording to be run
4. Start the application to begin recording simultaneously to initiating the stimulus
software
5. An excel document is generated containing this test data
6. Repeat steps 3 to 5 to collect additional samples of the same test if required
7. Once all required samples are recorded proceed to process the data as detailed in the
next slide
8. Repeat steps 2 to 7 for each experiment scenario
14. METHODOLOGY - PROCEDURE
• Data Processing:
Load raw sample
data
Concatenate data
samples
Extract the 14
sensor channels
Average the
samples
Normalise the
resulting sample
Divide sample into
a training set (80%)
and a trial set (20%)
Feature extraction
and selection
Train classifier
(using training
dataset)
Test the trained
classifier (using trial
dataset)
15. PROGRESS
• Updating Python code from outdated version to be compatible with the
latest drivers and OS in use
• Python 3.3+ significant changes required
• High Performance version
• Real time Feasibility
• EPOC Firmware research
• Emotiv SDK and Trivial applications for self study
16. DIFFICULTIES THUS FAR
• Illness during late semester
• EPOC is Outdated
• EPOC suffers from corrosion
• Compatibilities with current technologies
17. RESULTS, ANALYSIS & FINDINGS
• Delayed due to code updates required to run code and to use both original
data and new classifiers
• Findings from prior research
• Findings from use cases conducted thus far