Individuals’ differences may cause changes in the baseline of brains and affect behavioral performance and the relationship between brain activities. It is hard to use generalized prediction models to fit each state. Therefore, we purpose an ensemble-based weighted fatigue prediction system.

The class imbalance problem in machine learning occurs when certain cognitive feature classes are underrepresented relative to the others, leading to a learning bias toward the majority classes. To reduce information loss during feature space projection, we proposes a novel oversampling algorithm, named minority oversampling in kernel adaptive subspaces (MOKAS), which exploits the invariant feature extraction capability of a kernel version of the adaptive subspace self-organizing maps. The synthetic instances are generated from well-trained subspaces and then their pre-images are reconstructed in the input space.

Electroencephalography (EEG) has been demonstrated to be effective for monitoring changes in the human brain state and behavior. Three domains, namely, frequency, temporal, and channel spatial information were comprehensively considered. A 4-D convolutional neural-network (4-D CNN) algorithm was then proposed to associate all information from the EEG signals and the changes in the human state and behavioral performance. A 4-D CNN achieves superior forecasting performance over 2-D CNN, 3-D CNN, and shallow networks. This work also contributes to enhancing our understanding of deep learning methods in the analysis of EEG signals.

Decoding of human sensorimotor cortex for speech articulation is the grand challenge for neural scientist. Applying brain-computer interface is a promising strategy to recognize human expressions from the continuous decoding of neural signals underlying speech imagination. For ordinary people, direct interpretation in words/semantics can provide more accurate and intuitive approach instructions for brain operation applications than imagination. However, Mandarin Chinese is one of the popular languages in the world but has few study to explorer the feature brain signals while speaking Mandarin systemically. This project will integrate real-time multi-channel high-density Electroencephalography (EEG) to explore brain dynamics while speaking Mandarin for extracting the speech-related EEG signals feature. Additionally, DS-BCI is proposed by machine learning techniques to achieve a brain-computer interface of Mandarin speech expression.

Dementia is a degenerative disease of old age that is well known and valued by the public. In recent years, the prevalence of dementia has become trending as it enters the advanced age society. Related topics such as treatment and care are major issues facing the development of human society. Mild cognitive impairment (MCI) is considered a transitional stage from a normal cognitive state to the development of dementia. The prediction of MCI and early dementia is considered to help prevent or delay dementia.
We propose the integrative biomarkers to predict MCI and early dementia, and establish a predictive model that enable an earlier, and potentially more effective, treatment of MCI and early dementia. Electroencephalography (EEG) biomarkers would provide a non-invasive and relatively cheap screening tool to predict conversion to early dementia.