- Research Article
1
- 10.1016/j.bspc.2025.109377
MT-DFAN: A multi-task dynamic fusion attention network for multimodal emotion recognition in naturalistic conversations
- Apr 01, 2026
- Biomedical Signal Processing and Control
- Yuyang Song + 2 more +2
Publications from 2021 to 2026
Showing 10 of 263 papers
MT-DFAN: A multi-task dynamic fusion attention network for multimodal emotion recognition in naturalistic conversations
Compact exploration for continuous action reinforcement learning
Optimal trajectory planning for collaborative robots based on improved adaptive multi-objective particle swarm algorithm
A Comparative Study of the Multi-Objective Effects of Quantitative and Price-Based Monetary Policy Tools: Dual Evidence from Same and Mixed-Frequency Models
Ionic Highways under Multivariate-Coupled Strategies: Ultrahigh Power Generation from Industrial Waste Liquors Using Robust COF Membranes.
Efficiently harvesting the intrinsic energy from low-grade heat, acidity, and high salinity of desulfurization waste liquors is crucial for sustainable management, yet remains challenging due to the instability of conventional membranes under such extreme multi-physics conditions. Herein, we report a high-performance osmotic energy conversion device engineered via a multivariate coupling strategy. The core of this device is a robust membrane based on β-ketoenamine-linked covalent organic frameworks (COFs), featuring nanochannels functionalized with tailored stimuli-responsive groups to dynamically regulate surface charge density. The β-ketoenamine linkage endows exceptional membrane stability, enabling durable operation under coupled thermal, chemical, and electrochemical stresses. Experimental and computational studies demonstrate that the remarkable power enhancement stems not only from acid-induced protonation that boosts charge density, but also from the utilization of low-grade heat to accelerate ion transport. By simulating the multi-physical field coupling in real waste liquors, the device achieves an ultrahigh power output of 258.81W m- 2, surpassing commercial benchmarks by 52-fold. This COF membrane, with its exceptional permeability, selectivity, and stability, paves the way for high-efficiency energy harvesting from hostile industrial environments.
Read moreMetallurgical origin of dual-cracking mechanism in resistance spot welds of dissimilar aluminum alloys
Coaxial 3D printing of zeolite core-shell structured catalysts for integrated NO x adsorption and selective catalytic reduction in cold start application.
NH3 selective catalytic reduction (NH3-SCR) is the most effective technology to alleviate NO x emission from diesel vehicles but faces the cold start problem. Ideally, integrating passive NO x adsorption (PNA) and NH3-SCR could achieve a cost-effective and space-friendly tandem of these two units, but designing effective catalysts to achieve both high adsorption capacity and superior catalytic activity remains a challenge. Herein, we have successfully developed an integrated PNA-SCR catalyst system based on the Pd-SSZ-13@Cu-SSZ-13 core-shell structured zeolite composite via coaxial 3D printing, which affords ultra-high NO x removal efficiency (96%) over the entire PNA and NH3-SCR process. Over the core-shell structure with spatially confined effects, NO x can be effectively adsorbed in the Pd-SSZ-13 core at low temperature (<170 °C) with less H2O competition, which is subsequently released at 200-350 °C to react completely with NH3 over the Cu-SSZ-13 shell without excessive side reactions. Based on the optimal component of composite catalysts, Pd-SSZ-13@Cu-SSZ-13 displays high adsorption capacity (NO x /Pd = 0.54), high adsorption rate, optimized desorption temperature (∼250 °C), and excellent NH3-SCR activity, providing a potential solution to the cold start challenge in NO x elimination.
Read moreDigital Infrastructure and Industrial Decarbonization: Evidence from the Dual Perspectives of Green Innovation and Digital Coordination in China
Molecular imprinting toward living systems: From cells to extracellular vesicles
The role of FOXO1/PPARγ in atrazine-induced hepatic lipid metabolism disorders.