Looking for help?
Search our knowledge base for answers
Search...
How fast are your learning algorithms to detect mental commands?
Mental Commands and Facial Expression detections use an epoch length of 1 second, updated 8 times per second. Detection latency is 0.25-0.5 seconds. Note that there is a human latency period also. In order to activate a command, the user must think the correct thought pattern, and sometimes this can take several seconds, even longer with novice users. Expert users can typically activate a Mental Command fast enough to feel like the action is natural and is triggered exactly when they want. Latency of <0.5 seconds can be achieved by experienced users.
Performance Metrics are trained using a large volume of data collected in controlled experimental conditions from many subjects - usually about 30 to 150 hours of EEG data in total. These metrics are designed to be universal - they work the same way on all subjects, with some minor statistically-based corrections to adjust for individual differences in offset (typical level) and range (degree of emotional response for the same input stimulus)
Mental Commands are trained on demand by the user. The minimum training requirement is 8 seconds per trained command, plus 8 seconds for Neutral. So an expert user can train 4 actions (plus Neutral) in 40 seconds. Mental Commands tend to become more stable and reliable if the user updates each training action at least a few times. Overall the training set is still very small, requiring about 2-3 minutes of total EEG data in order to make a good profile.
Was this article helpful?
Not finding what you need?
Our Support Team is just a click away.
Looking for help?
Search...
Knowledge base
How fast are your learning algorithms to detect mental commands?
Mental Commands and Facial Expression detections use an epoch length of 1 second, updated 8 times per second. Detection latency is 0.25-0.5 seconds. Note that there is a human latency period also. In order to activate a command, the user must think the correct thought pattern, and sometimes this can take several seconds, even longer with novice users. Expert users can typically activate a Mental Command fast enough to feel like the action is natural and is triggered exactly when they want. Latency of <0.5 seconds can be achieved by experienced users.
Performance Metrics are trained using a large volume of data collected in controlled experimental conditions from many subjects - usually about 30 to 150 hours of EEG data in total. These metrics are designed to be universal - they work the same way on all subjects, with some minor statistically-based corrections to adjust for individual differences in offset (typical level) and range (degree of emotional response for the same input stimulus)
Mental Commands are trained on demand by the user. The minimum training requirement is 8 seconds per trained command, plus 8 seconds for Neutral. So an expert user can train 4 actions (plus Neutral) in 40 seconds. Mental Commands tend to become more stable and reliable if the user updates each training action at least a few times. Overall the training set is still very small, requiring about 2-3 minutes of total EEG data in order to make a good profile.
Was this article helpful?
Not finding what you need?
Our Support Team is just a click away.
Looking for help?
Search...
Knowledge base
How fast are your learning algorithms to detect mental commands?
Mental Commands and Facial Expression detections use an epoch length of 1 second, updated 8 times per second. Detection latency is 0.25-0.5 seconds. Note that there is a human latency period also. In order to activate a command, the user must think the correct thought pattern, and sometimes this can take several seconds, even longer with novice users. Expert users can typically activate a Mental Command fast enough to feel like the action is natural and is triggered exactly when they want. Latency of <0.5 seconds can be achieved by experienced users.
Performance Metrics are trained using a large volume of data collected in controlled experimental conditions from many subjects - usually about 30 to 150 hours of EEG data in total. These metrics are designed to be universal - they work the same way on all subjects, with some minor statistically-based corrections to adjust for individual differences in offset (typical level) and range (degree of emotional response for the same input stimulus)
Mental Commands are trained on demand by the user. The minimum training requirement is 8 seconds per trained command, plus 8 seconds for Neutral. So an expert user can train 4 actions (plus Neutral) in 40 seconds. Mental Commands tend to become more stable and reliable if the user updates each training action at least a few times. Overall the training set is still very small, requiring about 2-3 minutes of total EEG data in order to make a good profile.
Was this article helpful?
Not finding what you need?
Our Support Team is just a click away.