There have been many attempts to design brain-computer interfaces (BCIs) for wheelchair control based on steady state visual evoked potential (SSVEP), event-related desynchronization/synchronization (ERD/ERS) during motor imagery (MI) tasks, P300 potential, and some hybrid signals. However, those BCI systems cannot implement the navigation flexibly effectively. In this paper, we propose a scheme two-class MI four-class SSVEP tasks. It only provide multi-degree its user, but also allow user...

It is preferable for the retired batteries to balance their states-of-health (SOH) in battery energy storage system (BESS) since it can prolong lifetime and reduce maintenance burden. So far, corresponding balancing techniques mainly focus on either SOH among packs or of cells inside a pack. This article further proposes multilayer equalization scheme equalize all cells' SOHs large-scale BESS by comprehensively combining pack strategy commercial cell techniques. noted that schemes cannot be...

This paper presents a multisegment coal mill model that covers the whole milling process from startup to shutdown. mathematical is derived through analysis of energy transferring, heat exchange, and mass flow balances. The work presented in focuses on modeling E-type vertical spindle mills are widely used coal-fired power plants. An evolutionary computation technique adopted identify unknown parameters using on-site measurement data. identified then validated with different sets online...

Non-negative Canonical Polyadic decomposition (NCPD) and non-negative Tucker (NTD) were compared for extracting the multi-domain feature of visual mismatch negativity (vMMN), a small event-related potential (ERP), cognitive research. Since signal-to-noise ratio in vMMN is low, NTD outperformed NCPD. Moreover, we proposed an approach to select ERP among all extracted features discussed determination numbers components NCPD regarding context.

Despite recent advances, scene text recognition remains a challenging problem due to the significant variability, irregularity and distortion in appearance localization. Attention-based methods have become mainstream their superior vocabulary learning observation ability. Nonetheless, they are susceptible attention drift which can lead word errors. Most works focus on correcting decoding but completely ignore error accumulated during encoding process. In this paper, we propose novel scheme,...

Grade classification of gliomas is critical in clinical diagnosis and treatment decisions. Although histological images are commonly used for grading as an important factor prognostic prediction, their results prone to inter-observer variability. Recent advancements molecular genetics have significantly improved tumor classification, but challenges persist effective feature selection multi-modal data fusion. This letter proposes a hybrid encoding method based on information bottleneck...

The problem of incomplete data is common in signal processing and machine learning. Tensor completion algorithms aim to recover the from its partially observed entries. In this paper, taking advantages high compressibility flexibility recently proposed tensor ring (TR) decomposition, we propose a new approach named weighted optimization (TR-WOPT). It finds latent factors by gradient descent algorithm, then are employed predict missing entries tensor. We conduct various experiments on...

Event-related potentials (ERPs) especially P300 are popular effective features for brain-computer interface (BCI) systems based on electroencephalography (EEG). Traditional ERP-based BCI may perform poorly small training samples, i.e. the undersampling problem. In this study, ERP classification problem was investigated, in particular, high-dimensional setting with number of larger than samples studied. A flexible group sparse discriminative analysis algorithm Moreau-Yosida regularization...

Electroencephalogram (EEG) signals contain valuable information about the different physiological states of brain, with a variety linear and nonlinear features to investigate brain activity. Monitoring depth anesthesia (DoA) EEG is an ongoing challenge in studies. In this paper, we propose novel method based on Long Short-Term Memory (LSTM) sparse denoising auto-encoder (SDAE) combine hybrid monitor DoA. The were preprocessed using wavelet transform, filtering etc.. For later more than ten...

Alzheimer's disease (AD) is a neurodegenerative characterized by progressive deterioration of memory and cognition. Mild cognitive impairment (MCI) has been implicated as prodromal phase AD. Although abnormal functional connectivity (FC) demonstrated in AD MCI, the clinical differentiation AD, normal aging remains difficult, distinction between MCI especially problematic. We hypothesized that FC hippocampus other brain structures altered measurement could have diagnostic utility for...

Globally, the installed capacity of photovoltaic (PV) power plants is undergoing rapid growth. However, random output fluctuation PV has brought great challenge to system stable operation. One most effective approaches counteract impact carry out forecasting under ultra-short-term scales, among which second-level technique still absent since pure historical data analysis plant cannot deal with sudden change induced by moving cloud and traditional image detection methods actually fail achieve...

In clinical settings, the implementation of deep neural networks is impeded by prevalent problems label scarcity and class imbalance in medical images. To mitigate need for labeled data, semi-supervised learning (SSL) has gained traction. However, existing SSL schemes exhibit certain limitations. 1) They commonly fail to address problem. Training with imbalanced data makes model's prediction biased towards majority classes, consequently introducing bias. 2) usually suffer from training bias...

This paper explores two classes of model adaptation methods for Web search ranking: Model Interpolation and error-driven learning approaches based on a boosting algorithm. The results show that interpolation, though simple, achieves the best all open test sets where data is very different from training data. tree-based algorithm performance most closed are similar, but its drops significantly due to instability trees. Several explored improve robustness algorithm, with limited success.

Non-negative tensor factorization (NTF) has been successfully applied to analyze event-related potentials (ERPs), and shown superiority in terms of capturing multi-domain features. However, the time-frequency representation ERPs by higher-order tensors are usually large-scale, which prevents popularity most algorithms. To overcome this issue, we introduce a non-negative canonical polyadic decomposition (NCPD) based on low-rank approximation (LRA) hierarchical alternating least square (HALS)...

In this study, we propose a graph sequence neural network (GSNN) to accurately decode patterns of motor imagery from electroencephalograms (EEGs) in the presence distractions. GSNN aims build subgraphs by exploiting biological topologies among brain regions capture local and global relationships across characteristic channels. Specifically, model similarity between pairwise EEG channels adjacency matrix network. addition, node domain attention selection which connection sparsity can be...

The problem of analytically coordinating dissimilar voltage control actions to prevent collapse in a large power system is addressed. A framework for hybrid based on coordination controls with different response time and dynamic characteristics presented. proposed method security constrained steady-state approach. minimum distance from the operating point bifurcation boundary used evaluate security. optimal direction toward adequate then obtained by calculating sensitivity respect...