Brain-computer interface (BCI) speech decoding has emerged as a promising tool for assisting individuals with speech impairments. In this context, the integration of electroencephalography (EEG) and electromyography (EMG) signals offers strong potential for enhancing decoding performance. Mandarin tone classification presents particular challenges, as tonal variations convey distinct meanings even when phonemes remain identical. In this study, we propose a novel cross-subject multimodal BCI decoding framework that fuses EEG and EMG signals to classify four Mandarin tones under both audible and silent speech conditions. Inspired by the cooperative mechanisms of neural and muscular systems in speech production, our neural decoding architecture combines spatial-temporal feature extraction branches with a cross-attention fusion mechanism, enabling informative interaction between modalities. We further incorporate domain-adversarial training to improve cross-subject generalization. We collected 4,800 EEG trials and 4,800 EMG trials from 10 participants using only twenty EEG and five EMG channels, demonstrating the feasibility of minimal-channel decoding. Despite employing lightweight modules, our model outperforms state-of-the-art baselines across all conditions, achieving average classification accuracies of 87.83\% for audible speech and 88.08\% for silent speech. In cross-subject evaluations, it still maintains strong performance with accuracies of 83.27\% and 85.10\% for audible and silent speech, respectively. We further conduct ablation studies to validate the effectiveness of each component. Our findings suggest that tone-level decoding with minimal EEG-EMG channels is feasible and potentially generalizable across subjects, contributing to the development of practical BCI applications.

Recommender systems had been proposed to help people to find the interested items, such as recommending products to a buyer; identifying movies or music that a user will find interest, etc. However, the existing recommendation approaches mainly focus on capturing user-item interaction patterns for prediction, and ignore the user's side information such as visit frequency and duration. In this paper, we study the side information aided website recommendation problem that using the browsing history of a set of users and their side information to predict the websites that will be of interest to a certain user. We propose a novel recommendation approach called SI-VAE that incorporates side information with the variational autoencoders (VAEs) model for top-k recommendation. The proposed method takes both user-website interaction information and side information as input, and adopts an encoder/decoder model to generate user's interested websites from partial observations. The model of SI-VAE is implemented as a neural network, and trained with a multinomial likelihood objective function to form the ranking of user-website interaction probabilities. We conduct extensive experiments on two real-world datasets, which show that the proposed model outperforms the baselines in a number of performance metrics in website recommendation.