Fudan University

Given the wide adoption of multimodal sensors (e.g., camera, lidar, radar) by autonomous vehicle s (AVs), deep analytics to fuse their outputs for a robust perception become imperative. However, existing fusion methods often make two assumptions rarely holding in practice: i) similar data distributions for all inputs and ii) constant availability for all sensors. Because, for example, lidars have various resolutions and failures of radars may occur, such variability often results in significant performance degradation in fusion. To this end, we present t-READi, an adaptive inference system that accommodates the variability of multimodal sensory data and thus enables robust and efficient perception. t-READi identifies variation-sensitive yet structure-specific model parameters; it then adapts only these parameters while keeping the rest intact. t-READi also leverages a cross-modality contrastive learning method to compensate for the loss from missing modalities. Both functions are implemented to maintain compatibility with existing multimodal deep fusion methods. The extensive experiments evidently demonstrate that compared with the status quo approaches, t-READi not only improves the average inference accuracy by more than 6% but also reduces the inference latency by almost 15× with the cost of only 5% extra memory overhead in the worst case under realistic data and modal variations.

Zinc metal is a promising anode material for zinc-ion batteries (ZIBs), but severe side reactions and dendrite formation hinder its commercialization. In this study, starch is introduced into the ZnSO4...

China has adhered to policies of zero-COVID for almost three years since the outbreak of COVID-19, which has remarkably affected the circulation of respiratory pathogens. However, China has begun to end the zero-COVID policies in late 2022. Here, we reported a resurgence of common respiratory viruses and Mycoplasma pneumoniae with unique epidemiological characteristics among children after ending the zero-COVID policy in Shanghai, China, 2023. Children hospitalized with acute respiratory tract infections were enrolled from January 2022 to December 2023. Nine common respiratory viruses and 2 atypical bacteria were detected in respiratory specimens from the enrolled patients using a multiplex PCR-based assay. The data were analyzed and compared between the periods before (2022) and after (2023) ending the zero-COVID policies. A total of 8550 patients were enrolled, including 6170 patients in 2023 and 2380 patients in 2022. Rhinovirus (14.2%) was the dominant pathogen in 2022, however, Mycoplasma pneumoniae (38.8%) was the dominant pathogen in 2023. Compared with 2022, the detection rates of pathogens were significantly increased in 2023 (72.9% vs. 41.8%, p < 0.001). An out‐of-season epidemic of respiratory syncytial virus was observed during the spring and summer of 2023. The median age of children infected with respiratory viruses in 2023 was significantly greater than that in 2022. Besides, mixed infections were more frequent in 2023 (23.8% vs. 28.9%, p < 0.001). China is now facing multiple respiratory pathogen epidemics with changing seasonality, altered age distribution, and increasing mixed infection rates among children in 2023. Our finding highlights the need for public health interventions to prepare for the respiratory pathogen outbreaks in the post-COVID-19 era.

Crystal‐facet heterojunction engineering of mesoporous nanoreactors with highly redox‐active represents an efficacious strategy for the transformation of CO2 into valuable C2 products (e.g., C2H4). Herein, hollow mesoporous cube‐like CuS nanoreactors (~860 nm) with controlled anisotropic crystal‐facets are prepared through an interfacial‐confined ion dynamic migration‐rearrangement strategy. The regulation of the S2‐ ion concentration facilitates the modulation of the highly active (110) to (100) crystal‐facet ratios from 0.119 to 0.288, and induces the formation of anisotropic crystal‐facet heterojunctions. The controllable crystal‐facet heterojunctions trigger the directional charge carrier migration, and are accompanied with the formation of tandem S‐defect sites (Cu0‐S1@S3). Both of them promote the efficient electron‐hole pair dissociation and attain asymmetric C‐C coupling. The hollow mesoporous CuS nanoreactors with optimized crystal‐facet ratio of 0.224 (HMe‐CuS‐3) deliver a high selectivity of 72.7% for the photocatalytic reduction of CO2 to acetylene (C2H2). Further constructed Au‐(110) and Co3O4‐(100) spatially separated cascade nanoreactors (SS‐Au@Co3O4‐CuS) achieve CO2‐C2H4 photoreduction, in which the Co‐sites enhance H2O dissociation to provide protons and the protonation of *CO to *COH. The *COH is further captured by Au‐sites to accomplish the asymmetric *CO‐*COH coupling and subsequent protonation, ensuring a high C2H4 generation rate of 4.11 μmol/g/h with a selectivity as high as 90.6%.

Vector autoregressive (VAR) models are widely used in the analysis of time series and have been extensively studied in the literature. However, in scenarios with a large number of nodes, estimating the transition matrix in VAR models can be challenging. By incorporating the structure of the network into the VAR models, the number of parameters can be significantly reduced. In this paper, we propose a time-varying network vector autoregressive (tvNAR) model. In the tvNAR model, the response of each node at a given time point is assumed to be a linear combination of its previous values and those of its connected neighbors in the network. The coefficients are node-specific and time-varying, allowing the model to capture the unique effect of each node and describe the behavior of non-stationary time series. We propose a locally linear regression estimator of the time-varying nodal coefficients and establish its asymptotic properties. To examine the temporal stability of the coefficients, we propose a Wald-type test. We illustrate the performance of the estimator and the test procedure through simulation studies and empirical analysis of daily Nasdaq stock prices data.

The concept phase has been shown to play an important role in the analysis of multi-input multi-output system. Motivated by the recent results representing the multilinear time-invariant system in terms of standard notions of tensors, the definition of phases for third-order complex sectorial tensors via the tensor–tensor product (T-product) is proposed as a natural counterpart in this paper. The properties of tensor phases are shown along with sectorial tensor decompositions, compressions, Schur complements, tensor–tensor T-product, tensor sum, Hadamard product and T-Kronecker products. Furthermore, the phases of banded sectorial tensor completion and banded sectorial tensor decomposition are studied. Besides, we derive a tensor version of the Kalman–Yakubovich–Popov lemma.

Pulmonary metastasis represents one of the most prevalent forms of metastasis in advanced melanoma, with mortality rates reaching 70%. Current treatments including chemotherapy, targeted therapy, and immunotherapy frequently exhibit limited efficacy or present high costs. To address these clinical needs, this study presents a biomimetic drug delivery system (Ce6‐pTP‐CsA) utilizing cryoshocked adipocytes (CsA) encapsulating the prodrug triptolide palmitate (pTP) and the photosensitizer Ce6, exploiting the characteristic of tumor cells to recruit and lipolyze adipocytes for energy. CsA substantially enhances the drug‐loading capacity of adipocytes, with its particle size characteristics enabling targeted delivery of pTP to the lungs. The combination of photodynamic therapy (PDT) and pTP activates the caspase cascade, promoting apoptosis in tumor cells. Notably, the cleavage of disulfide bonds in pTP depletes glutathione (GSH), reducing its scavenging effect on reactive oxygen species (ROS) and enhancing the efficacy of PDT. Results demonstrate that Ce6‐pTP‐CsA effectively inhibits the proliferation and invasion of pulmonary metastatic melanoma cells in vitro and induces apoptosis, while significantly suppressing lung metastasis of SCID mice models in vivo. In conclusion, this novel biomimetic drug delivery system based on adipocytes provides a promising strategy for targeted therapy in pulmonary metastatic melanoma.