19
Identification from Wearable Device Brain Signals
Through comprehensive testing with supervised, unsupervised, and semi-supervised
techniques, this chapter shows the viability of the proposed data representation to effectively capture the essence of the large amount of recorded raw data. It is shown that the
various classification techniques are effective in predicting persons and activities, and that
various clustering techniques also provide reasonable results.
2.2 Review of Wearables
Wearable technologies are evolving quickly and companies are discovering innovative
ways to utilize the enormous amount of data they now have access to. There are many
different categories of wearables. Each category collects its own unique type of data that
can be used for meaningful data mining activities. This chapter will specifically investigate
the collection of EEG brain signals [9] from a commercially available wearable, the Muse
headband.
2.2.1 Wristbands and Watches
Smart watches such as the Pebble or the Apple Watch provide an interface to notifications,
such as messages, calendar events, or breaking news, that are typically viewed on a smart
phone. Additionally, as a wearable, they have their own sensors such as GPS, accelerometers, magnetometers, and heart rate sensors. These sensors provide additional capabilities
for users, who can now track activity levels or health information.
Wristbands, such as Fitbit devices or the Nymi band, provide more specialized functionality and typically do not rely on an ongoing connection to a smart phone. The Nymi band,
for example, measures an individual's ECG to create a unique biometric identifier that can
be used in place of a traditional password. Fitness trackers such as the Jawbone UP or Fitbit
devices are wearable wristbands that can measure health and activity information. They
can record information such as how long and how far the individual has performed a specific exercise, the duration and quality of their sleep, and heart rate levels. Fitness tracker
wearables such as the Fitbit devices aim to motivate users and provide feedback through
the measurements they gather. Fitness trackers often tie in to smart phone or online applications to better analyze the data they collect, and to provide feedback to users.
2.2.2 Armbands
Armband wearables, such as the Myo, use sensors to monitor the user's gestures and
movements. The Myo, for example, uses electromyography sensors to detect the electrical
activity changes as the user activates different muscles and combines the EMG input with
gyroscope, accelerometer, and magnetometer sensors.
2.2.3 Headbands, Headsets, and Smartglasses
Head mounted wearables in general either augment the wearer’s capabilities by displaying additional information through a heads up display, such as in the case of Google Glass
or the Microsoft HoloLens, or use sensors to measure the wearer’s brain activity, such as
the Emotiv and Muse headbands.
Identification from Wearable Device Brain Signals
Through comprehensive testing with supervised, unsupervised, and semi-supervised
techniques, this chapter shows the viability of the proposed data representation to effectively capture the essence of the large amount of recorded raw data. It is shown that the
various classification techniques are effective in predicting persons and activities, and that
various clustering techniques also provide reasonable results.
2.2 Review of Wearables
Wearable technologies are evolving quickly and companies are discovering innovative
ways to utilize the enormous amount of data they now have access to. There are many
different categories of wearables. Each category collects its own unique type of data that
can be used for meaningful data mining activities. This chapter will specifically investigate
the collection of EEG brain signals [9] from a commercially available wearable, the Muse
headband.
2.2.1 Wristbands and Watches
Smart watches such as the Pebble or the Apple Watch provide an interface to notifications,
such as messages, calendar events, or breaking news, that are typically viewed on a smart
phone. Additionally, as a wearable, they have their own sensors such as GPS, accelerometers, magnetometers, and heart rate sensors. These sensors provide additional capabilities
for users, who can now track activity levels or health information.
Wristbands, such as Fitbit devices or the Nymi band, provide more specialized functionality and typically do not rely on an ongoing connection to a smart phone. The Nymi band,
for example, measures an individual's ECG to create a unique biometric identifier that can
be used in place of a traditional password. Fitness trackers such as the Jawbone UP or Fitbit
devices are wearable wristbands that can measure health and activity information. They
can record information such as how long and how far the individual has performed a specific exercise, the duration and quality of their sleep, and heart rate levels. Fitness tracker
wearables such as the Fitbit devices aim to motivate users and provide feedback through
the measurements they gather. Fitness trackers often tie in to smart phone or online applications to better analyze the data they collect, and to provide feedback to users.
2.2.2 Armbands
Armband wearables, such as the Myo, use sensors to monitor the user's gestures and
movements. The Myo, for example, uses electromyography sensors to detect the electrical
activity changes as the user activates different muscles and combines the EMG input with
gyroscope, accelerometer, and magnetometer sensors.
2.2.3 Headbands, Headsets, and Smartglasses
Head mounted wearables in general either augment the wearer’s capabilities by displaying additional information through a heads up display, such as in the case of Google Glass
or the Microsoft HoloLens, or use sensors to measure the wearer’s brain activity, such as
the Emotiv and Muse headbands.