Dual-Channel Adaptive Scale Hypergraph Encoders With Cross-View Contrastive Learning for Knowledge Tracing.

Knowledge tracing (KT) refers to predicting learners' performance in the future according to their historical responses, which has become an essential task in intelligent tutoring systems. Most deep learning-based methods usually model the learners' knowledge states via recurrent neural networks (RNNs) or attention mechanisms. Recently emerging graph neural networks (GNNs) assist the KT model to capture the relationships such as question-skill and question-learner. However, non-pairwise and complex higher-order information among responses is ignored. In addition, a single-channel encoded hidden vector struggles to represent multigranularity knowledge states. To tackle the above problems, we propose a novel KT model named dual-channel adaptive scale hypergraph encoders with cross-view contrastive learning (HyperKT). Specifically, we design an adaptive scale hyperedge distillation component for generating knowledge-aware hyperedges and pattern-aware hyperedges that reflect non-pairwise higher-order features among responses. Then, we propose dual-channel hypergraph encoders to capture multigranularity knowledge states from global and local state hypergraphs. The encoders consist of a simplified hypergraph convolution network and a collaborative hypergraph convolution network. To enhance the supervisory signal in the state hypergraphs, we introduce the cross-view contrastive learning mechanism, which performs among state hypergraph views and their transformed line graph views. Extensive experiments on three real-world datasets demonstrate the superior performance of our HyperKT over the state-of-the-art (SOTA).

Liquid metal-tailored gluten network for protein-based e-skin.

Designing electronic skin (e-skin) with proteins is a critical way to endow e-skin with biocompatibility, but engineering protein structures to achieve controllable mechanical properties and self-healing ability remains a challenge. Here, we develop a hybrid gluten network through the incorporation of a eutectic gallium indium alloy (EGaIn) to design a self-healable e-skin with improved mechanical properties. The intrinsic reversible disulfide bond/sulfhydryl group reconfiguration of gluten networks is explored as a driving force to introduce EGaIn as a chemical cross-linker, thus inducing secondary structure rearrangement of gluten to form additional ß-sheets as physical cross-linkers. Remarkably, the obtained gluten-based material is self-healing, achieves synthetic material-like stretchability (>1600%) and possesses the ability to promote skin cell proliferation. The final e-skin is biocompatible and biodegradable and can sense strain changes from human motions of different scales. The protein network microregulation method paves the way for future skin-like protein-based e-skin.

Hierarchically porous polyimide/Ti3C2Tx film with stable electromagnetic interference shielding after resisting harsh conditions.

Polymer-based conductive nanocomposites are promising for electromagnetic interference (EMI) shielding to ensure stable operations of electronic devices and protect humans from electromagnetic radiation. Although MXenes have shown high EMI shielding performances, it remains a great challenge to construct highly efficient EMI shielding polymer/MXene composite films with minimal MXene content and high durability to harsh conditions. Here, hierarchically porous polyimide (PI)/Ti3C2Tx films with consecutively conductive pathways have been constructed via a unidirectional PI aerogel­assisted immersion and hot-pressing strategy. Contributed by special architectures and high conductivities, PI/Ti3C2Tx films with 2.0 volume % Ti3C2Tx have high absolute EMI shielding effectiveness up to 15,527 dB cm2 g−1 at the thickness of 90 µm. Superior EMI shielding performance can be retained even after being subjected to hygrothermal or combustion environments, cryogenic (−196°C) or high (250°C) temperatures, and rapid thermal shock (∆T = 446°C), demonstrating high potential as high-performance EMI shielding materials resisting harsh conditions.

Super-elasticity at 4 K of covalently crosslinked polyimide aerogels with negative Poisson's ratio.

The deep cryogenic temperatures encountered in aerospace present significant challenges for the performance of elastic materials in spacecraft and related apparatus. Reported elastic carbon or ceramic aerogels overcome the low-temperature brittleness in conventional elastic polymers. However, complicated fabrication process and high costs greatly limited their applications. In this work, super-elasticity at a deep cryogenic temperature of covalently crosslinked polyimide (PI) aerogels is achieved based on scalable and low-cost directional dimethyl sulfoxide crystals assisted freeze gelation and freeze-drying strategy. The covalently crosslinked chemical structure, cellular architecture, negative Poisson's ratio (-0.2), low volume shrinkage (3.1%), and ultralow density (6.1 mg/cm3) endow the PI aerogels with an elastic compressive strain up to 99% even in liquid helium (4 K), almost zero loss of resilience after dramatic thermal shocks (∆T = 569 K), and fatigue resistance over 5000 times compressive cycles. This work provides a new pathway for constructing polymer-based materials with super-elasticity at deep cryogenic temperature, demonstrating much promise for extensive applications in ongoing and near-future aerospace exploration.