MADE: Multicurvature Adaptive Embedding for Temporal Knowledge Graph Completion.

Temporal knowledge graphs (TKGs) are receiving increased attention due to their time-dependent properties and the evolving nature of knowledge over time. TKGs typically contain complex geometric structures, such as hierarchical, ring, and chain structures, which can often be mixed together. However, embedding TKGs into Euclidean space, as is typically done with TKG completion (TKGC) models, presents a challenge when dealing with high-dimensional nonlinear data and complex geometric structures. To address this issue, we propose a novel TKGC model called multicurvature adaptive embedding (MADE). MADE models TKGs in multicurvature spaces, including flat Euclidean space (zero curvature), hyperbolic space (negative curvature), and hyperspherical space (positive curvature), to handle multiple geometric structures. We assign different weights to different curvature spaces in a data-driven manner to strengthen the ideal curvature spaces for modeling and weaken the inappropriate ones. Additionally, we introduce the quadruplet distributor (QD) to assist the information interaction in each geometric space. Ultimately, we develop an innovative temporal regularization to enhance the smoothness of timestamp embeddings by strengthening the correlation of neighboring timestamps. Experimental results show that MADE outperforms the existing state-of-the-art TKGC models.

Multi-modal long document classification based on Hierarchical Prompt and Multi-modal Transformer.

In the realm of long document classification (LDC), previous research has predominantly focused on modeling unimodal texts, overlooking the potential of multi-modal documents incorporating images. To address this gap, we introduce an innovative approach for multi-modal long document classification based on the Hierarchical Prompt and Multi-modal Transformer (HPMT). The proposed HPMT method facilitates multi-modal interactions at both the section and sentence levels, enabling a comprehensive capture of hierarchical structural features and complex multi-modal associations of long documents. Specifically, a Multi-scale Multi-modal Transformer (MsMMT) is tailored to capture the multi-granularity correlations between sentences and images. This is achieved through the incorporation of multi-scale convolutional kernels on sentence features, enhancing the model's ability to discern intricate patterns. Furthermore, to facilitate cross-level information interaction and promote learning of specific features at different levels, we introduce a Hierarchical Prompt (HierPrompt) block. This block incorporates section-level prompts and sentence-level prompts, both derived from a global prompt via distinct projection networks. Extensive experiments are conducted on four challenging multi-modal long document datasets. The results conclusively demonstrate the superiority of our proposed method, showcasing its performance advantages over existing techniques.

Bridging the Cross-Modality Semantic Gap in Visual Question Answering.

The objective of visual question answering (VQA) is to adequately comprehend a question and identify relevant contents in an image that can provide an answer. Existing approaches in VQA often combine visual and question features directly to create a unified cross-modality representation for answer inference. However, this kind of approach fails to bridge the semantic gap between visual and text modalities, resulting in a lack of alignment in cross-modality semantics and the inability to match key visual content accurately. In this article, we propose a model called the caption bridge-based cross-modality alignment and contrastive learning model (CBAC) to address the issue. The CBAC model aims to reduce the semantic gap between different modalities. It consists of a caption-based cross-modality alignment module and a visual-caption (V-C) contrastive learning module. By utilizing an auxiliary caption that shares the same modality as the question and has closer semantic associations with the visual, we are able to effectively reduce the semantic gap by separately matching the caption with both the question and the visual to generate pre-alignment features for each, which are then used in the subsequent fusion process. We also leverage the fact that V-C pairs exhibit stronger semantic connections compared to question-visual (Q-V) pairs to employ a contrastive learning mechanism on visual and caption pairs to further enhance the semantic alignment capabilities of single-modality encoders. Extensive experiments conducted on three benchmark datasets demonstrate that the proposed model outperforms previous state-of-the-art VQA models. Additionally, ablation experiments confirm the effectiveness of each module in our model. Furthermore, we conduct a qualitative analysis by visualizing the attention matrices to assess the reasoning reliability of the proposed model.

Light-Driven Biomimetic Nanomotors for Enhanced Photothermal Therapy.

Nanotechnology-based strategy has recently drawn extensive attention for the therapy of malignant tumors due to its distinct strengths in cancer diagnosis and treatment. However, the limited intratumoral permeability of nanoparticles is a major hurdle to achieving the desired effect of cancer treatment. Due to their superior cargo towing and reliable penetrating property, micro-/nanomotors (MNMs) are considered as one of the most potential candidates for the coming generation of drug delivery platforms. Here, near-infrared (NIR)-actuated biomimetic nanomotors (4T1-JPGSs-IND) are fabricated successfully and we demonstrate that 4T1-JPGSs-IND selectively accumulate in homologous tumor regions due to the effective homing ability. Upon laser irradiation, hyperthermia generated by 4T1-JPGSs-IND leads to self-thermophoretic motion and photothermal therapy (PTT) to ablate tumors with a deep depth, thereby improving the photothermal therapeutic effect for cancer management. The developed nanomotor system with multifunctionalities exhibits promising potential in biomedical applications to fight against various diseases.

Beyond low-pass filtering on large-scale graphs via Adaptive Filtering Graph Neural Networks.

Graph Neural Networks (GNNs) have emerged as a crucial deep learning framework for graph-structured data. However, existing GNNs suffer from the scalability limitation, which hinders their practical implementation in industrial settings. Many scalable GNNs have been proposed to address this limitation. However, they have been proven to act as low-pass graph filters, which discard the valuable middle- and high-frequency information. This paper proposes a novel graph neural network named Adaptive Filtering Graph Neural Networks (AFGNN), which can capture all frequency information on large-scale graphs. AFGNN consists of two stages. The first stage utilizes low-, middle-, and high-pass graph filters to extract comprehensive frequency information without introducing additional parameters. This computation is a one-time task and is pre-computed before training, ensuring its scalability. The second stage incorporates a node-level attention-based feature combination, enabling the generation of customized graph filters for each node, contrary to existing spectral GNNs that employ uniform graph filters for the entire graph. AFGNN is suitable for mini-batch training, and can enhance scalability and efficiently capture all frequency information from large-scale graphs. We evaluate AFGNN by comparing its ability to capture all frequency information with spectral GNNs, and its scalability with scalable GNNs. Experimental results illustrate that AFGNN surpasses both scalable GNNs and spectral GNNs, highlighting its superiority.

Micromotor-Enabled Active Hydrogen and Tobramycin Delivery for Synergistic Sepsis Therapy.

Sepsis is a highly heterogeneous syndrome normally characterized by bacterial infection and dysregulated systemic inflammatory response that leads to multiple organ failure and death. Single anti-inflammation or anti-infection treatment exhibits limited survival benefit for severe cases. Here a biodegradable tobramycin-loaded magnesium micromotor (Mg-Tob motor) is successfully developed as a potential hydrogen generator and active antibiotic deliverer for synergistic therapy of sepsis. The peritoneal fluid of septic mouse provides an applicable space for Mg-water reaction. Hydrogen generated sustainably and controllably from the motor interface propels the motion to achieve active drug delivery along with attenuating hyperinflammation. The developed Mg-Tob motor demonstrates efficient protection from anti-inflammatory and antibacterial activity both in vitro and in vivo. Importantly, it prevents multiple organ failure and significantly improves the survival rate up to 87.5% in a high-grade sepsis model with no survival, whereas only about half of mice survive with the individual therapies. This micromotor displays the superior therapeutic effect of synergistic hydrogen-chemical therapy against sepsis, thus holding great promise to be an innovative and translational drug delivery system to treat sepsis or other inflammation-related diseases in the near future.

Bladder microenvironment actuated proteomotors with ammonia amplification for enhanced cancer treatment.

Enzyme-driven micro/nanomotors consuming in situ chemical fuels have attracted lots of attention for biomedical applications. However, motor systems composed by organism-derived organics that maximize the therapeutic efficacy of enzymatic products remain challenging. Herein, swimming proteomotors based on biocompatible urease and human serum albumin are constructed for enhanced antitumor therapy via active motion and ammonia amplification. By decomposing urea into carbon dioxide and ammonia, the designed proteomotors are endowed with self-propulsive capability, which leads to improved internalization and enhanced penetration in vitro. As a glutamine synthetase inhibitor, the loaded l-methionine sulfoximine further prevents the conversion of toxic ammonia into non-toxic glutamine in both tumor and stromal cells, resulting in local ammonia amplification. After intravesical instillation, the proteomotors achieve longer bladder retention and thus significantly inhibit the growth of orthotopic bladder tumor in vivo without adverse effects. We envision that the as-developed swimming proteomotors with amplification of the product toxicity may be a potential platform for active cancer treatment.

Cancer Cell Membrane-Coated Nanoparticle Co-loaded with Photosensitizer and Toll-like Receptor 7 Agonist for the Enhancement of Combined Tumor Immunotherapy.

Tumor immunotherapy has shown considerable therapeutic potential in the past few years, but the clinical response rate of immunotherapy is less than 20%. Encountering the high heterogeneity of tumors, it will be a general trend to apply combined therapy for cancer treatment. Photodynamic therapy (PDT) transiently kills tumor cells by producing reactive oxygen species (ROS), while residual tumor cells are prone to metastasis, leading to tumor recurrence. In combination with tumor immunotherapy, it is hoped to awaken the host immune system and eradicate residual tumor cells. Herein, cancer cell membrane-coated nanoparticles as a platform to combine PDT, TLR7 agonist, and tumor antigen for the enhancement of tumor therapeutic efficacy are designed. The final biomimetic nanoparticles (CCMV/LTNPs) can specifically kill tumor cells through PDT, while strong host antitumor immune responses are elicited to eliminate residue tumor cells under the help of immune adjuvant and tumor antigen from the cancer cell membrane. In summary, a photoimmunotherapy strategy is designed that synergistically enhances the tumor therapeutic effects by killing tumor cells through PDT and activating host antitumor immune responses through the co-delivery of adjuvant and tumor antigen, which may offer a promising strategy for clinical immunotherapy in the future.

Hyperthermia-triggered biomimetic bubble nanomachines.

Nanoparticle-based drug delivery systems have gained much attention in the treatment of various malignant tumors during the past decades. However, limited tumor penetration of nanodrugs remains a significant hurdle for effective tumor therapy due to the existing biological barriers of tumoral microenvironment. Inspired by bubble machines, here we report the successful fabrication of biomimetic nanodevices capable of in-situ secreting cell-membrane-derived nanovesicles with smaller sizes under near infrared (NIR) laser irradiation for synergistic photothermal/photodynamic therapy. Porous Au nanocages (AuNC) are loaded with phase transitable perfluorohexane (PFO) and hemoglobin (Hb), followed by oxygen pre-saturation and indocyanine green (ICG) anchored 4T1 tumor cell membrane camouflage. Upon slight laser treatment, the loaded PFO undergoes phase transition due to surface plasmon resonance effect produced by AuNC framework, thus inducing the budding of outer cell membrane coating into small-scale nanovesicles based on the pore size of AuNC. Therefore, the hyperthermia-triggered generation of nanovesicles with smaller size, sufficient oxygen supply and anchored ICG results in enhanced tumor penetration for further self-sufficient oxygen-augmented photodynamic therapy and photothermal therapy. The as-developed biomimetic bubble nanomachines with temperature responsiveness show great promise as a potential nanoplatform for cancer treatment.

Graph structure learning layer and its graph convolution clustering application.

To learn the embedding representation of graph structure data corrupted by noise and outliers, existing graph structure learning networks usually follow the two-step paradigm, i.e., constructing a "good" graph structure and achieving the message passing for signals supported on the learned graph. However, the data corrupted by noise may make the learned graph structure unreliable. In this paper, we propose an adaptive graph convolutional clustering network that alternatively adjusts the graph structure and node representation layer-by-layer with back-propagation. Specifically, we design a Graph Structure Learning layer before each Graph Convolutional layer to learn the sparse graph structure from the node representations, where the graph structure is implicitly determined by the solution to the optimal self-expression problem. This is one of the first works that uses an optimization process as a Graph Network layer, which is obviously different from the function operation in traditional deep learning layers. An efficient iterative optimization algorithm is given to solve the optimal self-expression problem in the Graph Structure Learning layer. Experimental results show that the proposed method can effectively defend the negative effects of inaccurate graph structures. The code is available at https://github.com/HeXiax/SSGNN.

Nanorobot-Mediated Synchronized Neuron Activation.

Interactions between active materials lead to collective behavior and even intelligence beyond the capability of individuals. Such behaviors are prevalent in nature and can be observed in animal colonies, providing these species with diverse capacities for communication and cooperation. In artificial systems, however, collective intelligence systems interacting with biological entities remains unexplored. Herein, we describe black (B)-TiO2@N/Au nanorobots interacting through photocatalytic pure water splitting-induced electrophoresis that exhibit periodic swarming oscillations under programmed near-infrared light. The periodic chemical-electric field generated by the oscillating B-TiO2@N/Au nanorobot swarm leads to local neuron activation in vitro. The field oscillations and neurotransmission from synchronized neurons further trigger the resonance oscillation of neuron populations without synaptic contact (about 2 mm spacing), in different ways from normal neuron oscillation requiring direct contact. We envision that the oscillating nanorobot swarm platforms will shed light on contactless communication of neurons and offer tools to explore interactions between neurons.

A Dual-Masked Deep Structural Clustering Network With Adaptive Bidirectional Information Delivery.

Structured clustering networks, which alleviate the oversmoothing issue by delivering hidden features from autoencoder (AE) to graph convolutional networks (GCNs), involve two shortcomings for the clustering task. For one thing, they used vanilla structure to learn clustering representations without considering feature and structure corruption; for another thing, they exhibit network degradation and vanishing gradient issues after stacking multilayer GCNs. In this article, we propose a clustering method called dual-masked deep structural clustering network (DMDSC) with adaptive bidirectional information delivery (ABID). Specifically, DMDSC enables generative self-supervised learning to mine deeper interstructure and interfeature correlations by simultaneously reconstructing corrupted structures and features. Furthermore, DMDSC develops an ABID module to establish an information transfer channel between each pairwise layer of AE and GCNs to alleviate the oversmoothing and vanishing gradient problems. Numerous experiments on six benchmark datasets have shown that the proposed DMDSC outperforms the most advanced deep clustering algorithms.

Self-Propelled Proteomotors with Active Cell-Free mtDNA Clearance for Enhanced Therapy of Sepsis-Associated Acute Lung Injury.

Acute lung injury (ALI) is a frequent and serious complication of sepsis with limited therapeutic options. Gaining insights into the inflammatory dysregulation that causes sepsis-associated ALI can help develop new therapeutic strategies. Herein, the crucial role of cell-free mitochondrial DNA (cf-mtDNA) in the regulation of alveolar macrophage activation during sepsis-associated ALI is identified. Most importantly, a biocompatible hybrid protein nanomotor (NM) composed of recombinant deoxyribonuclease I (DNase-I) and human serum albumin (HSA) via glutaraldehyde-mediated crosslinking is prepared to obtain an inhalable nanotherapeutic platform targeting pulmonary cf-mtDNA clearance. The synthesized DNase-I/HSA NMs are endowed with self-propulsive capability and demonstrate superior performances in stability, DNA hydrolysis, and biosafety. Pulmonary delivery of DNase-I/HSA NMs effectively eliminates cf-mtDNAs in the lungs, and also improves sepsis survival by attenuating pulmonary inflammation and lung injury. Therefore, pulmonary cf-mtDNA clearance strategy using DNase-I/HSA NMs is considered to be an attractive approach for sepsis-associated ALI.

Magnetically Actuated Biodegradable Nanorobots for Active Immunotherapy.

An efficient and cost-effective therapeutic vaccine is highly desirable for the prevention and treatment of cancer, which helps to strengthen the immune system and activate the T cell immune response. However, initiating such an adaptive immune response efficiently remains challenging, especially the deficient antigen presentation by dendritic cells (DCs) in the immunosuppressive tumor microenvironment. Herein, an efficient and dynamic antigen delivery system based on the magnetically actuated OVA-CaCO3 -SPIO robots (OCS-robots) is rationally designed for active immunotherapy. Taking advantage of the unique dynamic features, the developed OCS-robots achieve controllable motion capability under the rotating magnetic field. Specifically, with the active motion, the acid-responsiveness of OCS-robots is beneficial for the tumor acidity attenuating and lysosome escape as well as the subsequent antigen cross-presentation of DCs. Furthermore, the dynamic OCS-robots boost the crosstalk between the DCs and antigens, which displays prominent tumor immunotherapy effect on melanoma through cytotoxic T lymphocytes (CTLs). Such a strategy of dynamic vaccine delivery system enables the active activation of immune system based on the magnetically actuated OCS-robots, which presents a plausible paradigm for incredibly efficient cancer immunotherapy by designing multifunctional and novel robot platforms in the future.

Mobile mechanical signal generator for macrophage polarization.

The importance of mechanical signals in regulating the fate of macrophages is gaining increased attention recently. However, the recently used mechanical signals normally rely on the physical characteristics of matrix with non-specificity and instability or mechanical loading devices with uncontrollability and complexity. Herein, we demonstrate the successful fabrication of self-assembled microrobots (SMRs) based on magnetic nanoparticles as local mechanical signal generators for precise macrophage polarization. Under a rotating magnetic field (RMF), the propulsion of SMRs occurs due to the elastic deformation via magnetic force and hydrodynamics. SMRs perform wireless navigation toward the targeted macrophage in a controllable manner and subsequently rotate around the cell for mechanical signal generation. Macrophages are eventually polarized from M0 to anti-inflammatory related M2 phenotypes by blocking the Piezo1-activating protein-1 (AP-1）-CCL2 signaling pathway. The as-developed microrobot system provides a new platform of mechanical signal loading for macrophage polarization, which holds great potential for precise regulation of cell fate.

QDN: A Quadruplet Distributor Network for Temporal Knowledge Graph Completion.

Temporal knowledge graph completion (TKGC) is an extension of the traditional static knowledge graph completion (SKGC) by introducing the timestamp. The existing TKGC methods generally translate the original quadruplet to the form of the triplet by integrating the timestamp into the entity/relation, and then use SKGC methods to infer the missing item. However, such an integrating operation largely limits the expressive ability of temporal information and ignores the semantic loss problem due to the fact that entities, relations, and timestamps are located in different spaces. In this article, we propose a novel TKGC method called the quadruplet distributor network (QDN), which independently models the embeddings of entities, relations, and timestamps in their specific spaces to fully capture the semantics and builds the QD to facilitate the information aggregation and distribution among them. Furthermore, the interaction among entities, relations, and timestamps is integrated using a novel quadruplet-specific decoder, which stretches the third-order tensor to the fourth-order to satisfy the TKGC criterion. Equally important, we design a novel temporal regularization that imposes a smoothness constraint on temporal embeddings. Experimental results show that the proposed method outperforms the existing state-of-the-art TKGC methods. The source codes of this article are available at https://github.com/QDN for Temporal Knowledge Graph Completion.git.

Exploiting Neighbor Effect: Conv-Agnostic GNN Framework for Graphs With Heterophily.

Due to the homophily assumption in graph convolution networks (GCNs), a common consensus in the graph node classification task is that graph neural networks (GNNs) perform well on homophilic graphs but may fail on heterophilic graphs with many interclass edges. However, the previous interclass edges' perspective and related homo-ratio metrics cannot well explain the GNNs' performance under some heterophilic datasets, which implies that not all the interclass edges are harmful to GNNs. In this work, we propose a new metric based on the von Neumann entropy to reexamine the heterophily problem of GNNs and investigate the feature aggregation of interclass edges from an entire neighbor identifiable perspective. Moreover, we propose a simple yet effective Conv-Agnostic GNN framework (CAGNNs) to enhance the performance of most GNNs on the heterophily datasets by learning the neighbor effect for each node. Specifically, we first decouple the feature of each node into the discriminative feature for downstream tasks and the aggregation feature for graph convolution (GC). Then, we propose a shared mixer module to adaptively evaluate the neighbor effect of each node to incorporate the neighbor information. The proposed framework can be regarded as a plug-in component and is compatible with most GNNs. The experimental results over nine well-known benchmark datasets indicate that our framework can significantly improve performance, especially for the heterophily graphs. The average performance gain is 9.81%, 25.81%, and 20.61% compared with graph isomorphism network (GIN), graph attention network (GAT), and GCN, respectively. Extensive ablation studies and robustness analysis further verify the effectiveness, robustness, and interpretability of our framework. Code is available at https://github.com/JC-202/CAGNN.

Graph contrastive learning with implicit augmentations.

Existing graph contrastive learning methods rely on augmentation techniques based on random perturbations (e.g., randomly adding or dropping edges and nodes). Nevertheless, altering certain edges or nodes can unexpectedly change the graph characteristics, and choosing the optimal perturbing ratio for each dataset requires onerous manual tuning. In this paper, we introduce Implicit Graph Contrastive Learning (iGCL), which utilizes augmentations in the latent space learned from a Variational Graph Auto-Encoder by reconstructing graph topological structure. Importantly, instead of explicitly sampling augmentations from latent distributions, we further propose an upper bound for the expected contrastive loss to improve the efficiency of our learning algorithm. Thus, graph semantics can be preserved within the augmentations in an intelligent way without arbitrary manual design or prior human knowledge. Experimental results on both graph-level and node-level show that the proposed method achieves state-of-the-art accuracy on downstream classification tasks compared to other graph contrastive baselines, where ablation studies in the end demonstrate the effectiveness of modules in iGCL.

Immunostimulant In Situ Fibrin Gel for Post-operative Glioblastoma Treatment by Macrophage Reprogramming and Photo-Chemo-Immunotherapy.

Tumor recurrence remains the leading cause of treatment failure following surgical resection of glioblastoma (GBM). M2-like tumor-associated macrophages (TAMs) infiltrating the tumor tissue promote tumor progression and seriously impair the efficacy of chemotherapy and immunotherapy. In addition, designing drugs capable of crossing the blood-brain barrier and eliciting the applicable organic response is an ambitious challenge. Here, we propose an injectable nanoparticle-hydrogel system that uses doxorubicin (DOX)-loaded mesoporous polydopamine (MPDA) nanoparticles encapsulated in M1 macrophage-derived nanovesicles (M1NVs) as effectors and fibrin hydrogels as in situ delivery vehicles. In vivo fluorescence imaging shows that the hydrogel system triggers photo-chemo-immunotherapy to destroy remaining tumor cells when delivered to the tumor cavity of a model of subtotal GBM resection. Concomitantly, the result of flow cytometry indicated that M1NVs comprehensively improved the immune microenvironment by reprogramming M2-like TAMs to M1-like TAMs. This hydrogel system combined with a near-infrared laser effectively promoted the continuous infiltration of T cells, restored T cell effector function, inhibited the infiltration of myeloid-derived suppressor cells and regulatory T cells, and thereby exhibited a strong antitumor immune response and significantly inhibited tumor growth. Hence, MPDA-DOX-NVs@Gel (MD-NVs@Gel) presents a unique clinical strategy for the treatment of GBM recurrence.

Domain Adaptation as Optimal Transport on Grassmann Manifolds.

Domain adaptation in the Euclidean space is a challenging task on which researchers recently have made great progress. However, in practice, there are rich data representations that are not Euclidean. For example, many high-dimensional data in computer vision are in general modeled by a low-dimensional manifold. This prompts the demand of exploring domain adaptation between non-Euclidean manifold spaces. This article is concerned with domain adaption over the classic Grassmann manifolds. An optimal transport-based domain adaptation model on Grassmann manifolds has been proposed. The model implements the adaption between datasets by minimizing the Wasserstein distances between the projected source data and the target data on Grassmann manifolds. Four regularization terms are introduced to keep task-related consistency in the adaptation process. Furthermore, to reduce the computational cost, a simplified model preserving the necessary adaption property and its efficient algorithm is proposed and tested. The experiments on several publicly available datasets prove the proposed model outperforms several relevant baseline domain adaptation methods.