Unsupervised EEG Artifact Detection Flashcards
Problem/Motivation
A representation learning solution for artifact detection and correction.
what is representation learning?
Essentially just refers to automated methods for extraction and learning features
What is unsupervised outlier detection?
Unsupervised outlier detection is a type of anomaly detection that identifies unusual data points or outliers without requiring labeled training data. This study uses both statistical and representation learning methods to detect outliers
What is encoder decoder based correction and how is it used in the study?
Its a deep learning technique to correct missing or corrupt data points using the data points preceding and proceeding the corruput signal.
What are complexity features
features that measure the unpredictability and information content of EEG signals
What are connectivity features
features that measure the dependencies ad interactions between EEG signals from different electrodes
What are continuity features
features that explore the regularity, volatility, and stability of the signals over time
Shanon entropy
measures the degree of randomness or uncertainty of an EEG signal. High entropy indicates a more complex signal. low entropy may indicate reduce brain activity or even consciousness.
Information quantity
represents the entropy of wavelet decomposed EEG signals across different frequent bands.
Wavelet decomposition
a techique used to break down EEG signals into components at various frequency bands, better capturig transient events and
Median Frequency
Median frequency is the frequency in which the spectral power distribution is divided into two equal parts. Spectral power is the power is the signal at different frequencies. A shift in MF can be associated with changes inn alertness and arousual
Number of bursts
the number if increases in signal amplitude within a certain time frame, which is crucial for seizure detection, since a seizure are chrctrized by sudden and excessive bursts in electrical activty.
Coherence
measures the degree of synchronization of EEG signals across different channels. Crucial in understanding the functional connectivity of the brain
Mutual information
Quantifies the
