Search

Recently, more and more attention has been focused on the intellectual property protection of deep neural networks (DNNs), promoting DNN watermarking to become a hot research topic. Compared with embedding watermarks directly into DNN parameters, inserting trigger-set watermarks enables us to verify the ownership without knowing the internal details of the DNN, which is more suitable for application scenarios. The cost is we have to carefully craft the trigger samples. Mainstream methods construct the trigger samples by inserting a noticeable pattern to the clean samples in the spatial domain, which does not consider sample imperceptibility, sample robustness and model robustness, and therefore has limited the watermarking performance and the model generalization. It has motivated the authors in this paper to propose a novel DNN watermarking method based on Fourier perturbation analysis and frequency sensitivity clustering. First, we analyze the perturbation impact of different frequency components of the input sample on the task functionality of the DNN by applying random perturbation. Then, by K-means clustering, we determine the frequency components that result in superior watermarking performance for crafting the trigger samples. Our experiments show that the proposed work not only maintains the performance of the DNN on its original task, but also provides better watermarking performance compared with related works.

Dense depth recovery is crucial in autonomous driving, serving as a foundational element for obstacle avoidance, 3D object detection, and local path planning. Adverse weather conditions, including haze, dust, rain, snow, and darkness, introduce significant challenges to accurate dense depth estimation, thereby posing substantial safety risks in autonomous driving. These challenges are particularly pronounced for traditional depth estimation methods that rely on short electromagnetic wave sensors, such as visible spectrum cameras and near-infrared LiDAR, due to their susceptibility to diffraction noise and occlusion in such environments. To fundamentally overcome this issue, we present a novel approach for robust metric depth estimation by fusing a millimeter-wave radar and a monocular infrared thermal camera, which are capable of penetrating atmospheric particles and unaffected by lighting conditions. Our proposed Radar-Infrared fusion method achieves highly accurate and finely detailed dense depth estimation through three stages, including monocular depth prediction with global scale alignment, quasi-dense radar augmentation by learning radar-pixels correspondences, and local scale refinement of dense depth using a scale map learner. Our method achieves exceptional visual quality and accurate metric estimation by addressing the challenges of ambiguity and misalignment that arise from directly fusing multi-modal long-wave features. We evaluate the performance of our approach on the NTU4DRadLM dataset and our self-collected challenging ZJU-Multispectrum dataset. Especially noteworthy is the unprecedented robustness demonstrated by our proposed method in smoky scenarios. Our code will be released at https://github.com/MMOCKING/RIDERS.

The time series self-supervised contrastive learning framework has succeeded significantly in industrial fault detection scenarios. It typically consists of pretraining on abundant unlabeled data and fine-tuning on limited annotated data. However, the two-phase framework faces three challenges: Sampling bias, task-agnostic representation issue, and angular-centricity issue. These challenges hinder further development in industrial applications. This article introduces a debiased contrastive learning with supervision guidance (DCLSG) framework and applies it to industrial fault detection tasks. First, DCLSG employs channel augmentation to integrate temporal and frequency domain information. Pseudolabels based on momentum clustering operation are assigned to extracted representations, thereby mitigating the sampling bias raised by the selection of positive pairs. Second, the generated supervisory signal guides the pretraining phase, tackling the task-agnostic representation issue. Third, the angular-centricity issue is addressed using the proposed Gaussian distance metric measuring the radial distribution of representations. The experiments conducted on three industrial datasets (ISDB, CWRU, and practical datasets) validate the superior performance of the DCLSG compared to other fault detection methods.

We propose and demonstrate an approach to generating wide-range tunable coherent dual-frequency microwave signals with high side-mode suppression ratios (SMSRs), ultra-low phase noise and high stability, based on a dual-passband optoelectronic oscillator (OEO). The narrowband passband with a relatively high gain is used to generate the initial oscillation at a fixed frequency. Through injecting a single-tone signal into the OEO cavity via a microwave mixer, frequency-tunable oscillation in the other broadband passband with a relatively low gain is initiated through frequency mixing injection locking. Then, the two oscillation signals are phase-locked through frequency mixing mutual injection locking. In addition, a dual-loop configuration with the assistance of balanced photodetection is utilized to suppress the side modes and improve the phase noise performance, which also enhances the stability of the generated signal. In the experiment, coherent dual-frequency microwave signals with a fixed frequency at 20.0056 GHz and a tunable frequency in the range of 13.9705 GHz to 18.0075 GHz are generated, where the frequency tuning step is 625.6 kHz. The power fluctuation during frequency tuning is smaller than 1.5 dB, and the SMSRs are larger than 77 dB. In addition, the Allan deviation and the phase noise are in the order of 10–11@1 s and better than –138 dBc/Hz@10 kHz, respectively.

A scalable electrical method is proposed for measuring high-frequency responses of photodetectors (PDs) through harmonic up-conversion based on pilot-aided electro-optical stimulus. The proposed method provides wideband optical stimulus by the electro-optical harmonic up-conversion and subtracts the frequency response of the assisted electro-optical Mach-Zehnder modulator (MZM) based on the pilot insertion and extraction fixed at low frequency, which is free of any golden intensity modulator (IM) of known performance. Moreover, it achieves wideband and scalable high-frequency measurement of PDs with n-fold measuring frequency range, where the sweeping frequency up to fs of the assisted MZM enables the measuring frequency range of n × fs. In the experimental demonstration, a commercial PD is measured with an assisted MZM under a frequency-swept modulation up to 16.75 GHz and a fixed-low-frequency pilot signal at 80 MHz, and the frequency responses are evaluated up to 67 GHz. Moreover, the feasibility of the scalable scheme is proved by those cases of 1, 2, 3, and 4-fold measuring frequency range. The consistency between our method and the two-tone mixing method verifies that the scalable high-frequency measurement provides a simple but accurate method for the performance characterization of high-speed PDs.

Selective functionaliz0ation of C(sp3)–H bonds is a straightforward and practical method to construct complex molecule skeletons. In this field, direct transformation of unactivated C(sp3)–H bonds into C(sp3)–SCF3architectures is still a great challenge. We report a highly selective trifluoromethylthiolation of unactivated aliphatic C(sp3)–H bonds by combination of proton-coupled electron transfer (PCET) and hydrogen atom transfer (HAT) strategy. A wide range of structurally diverse alkyl trifluoromethyl sulfides are obtained. Furthermore, the use of two different photocatalysts can realize an unprecedented trifluoromethylthiolation and amidation cascade of different C(sp3)–H bonds. It can afford a good access to bifunctionalized molecules in synthetically useful yields.

Mobile edge computing (MEC) and digital twin (DT) technologies have been recognized as key enabling factors for the next generation of industrial Internet of Things (IoT) applications. In existing works, DT-assisted edge network resource optimization solutions mostly focus on short-term performance optimization, and long-term resource optimization has not been well studied. Thus, this paper introduces a digital twin-assisted heterogeneous edge network (DTHEN), aiming to minimize long-term energy consumption by jointly optimizing transmit power and computing resource. To solve the stochastic optimization problem, we propose a long-term queue-aware energy minimization (LQEM) scheme for joint communication and computing resource management. The proposed scheme uses Lyapunov optimization to transform the original problem with long-term time constraints into a deterministic upper bound problem for each time slot, decouples it into three independent sub-problems, and solves each sub-problem separately. We then theoretically prove the asymptotic optimality of the LQEM scheme and the tradeoff between system energy consumption and task queue backlog. Finally, experimental results verify the performance analysis of the LQEM scheme, demonstrating its superiority over several benchmark schemes, and reveal the impact of various parameters on the system.

As the calorific value of waste-to-energy incineration continues to rise, the corrosion damage to boilers becomes increasingly severe. The impact of the grain size on the high-temperature corrosion resistance of Inconel 625 coating was investigated. The results indicated that the corrosion performance of the coating in 1:1 NaCl–KCl salts at 700 °C for 120 h was highly dependent on the grain size of the coating. The coating fabricated with a laser power of 1800 W and an overlapping distance of 1.5 mm had an average grain size of 13.7 μm and highest hardness, exhibiting the best corrosion performance, with a corrosion weight loss of 1.4 mg/cm2. Coatings processed under different conditions all formed a three-layered structure, with a NiO outermost layer, a Cr2O3intermediate layer, and an innermost Cr-depletion zone. Coatings with finer grains formed a continuous and dense protective oxide layer of NiO and Cr2O3due to the rapid diffusion of alloy elements, effectively preventing the invasion of corrosive media and protecting the substrate from corrosion.

Background: Esophageal cancer poses a significant health burden globally. Endoscopic treatment has emerged as a viable option for patient ineligible for surgery or experiencing disease recurrence post-radiotherapy. Methods: Patients visiting the Department of Endoscopy at the Cancer Hospital of China Academy of Medical Sciences between March 2009 and March 2024 were retrospectively analyzed. Inclusion criteria encompassed patients with histologically confirmed esophageal cancer who had not undergone surgery, but received radiotherapy or CRT, and subsequently opted for endoscopic treatment. Data on demographics, treatment modalities, recurrence patterns, histopathological characteristics, and outcomes were collected. Statistical analysis was conducted using SPSS 27.0, employing Kolmogorov–Smirnov tests for data normality assessment. Results: Out of 25 included patients, the mean age was 60.29 years, with a predominance of males (88%). Most patients (64%) received chemoradiotherapy (CRT), while the rest underwent radiotherapy alone. The median follow-up duration was 50.92 months, with a median recurrence time of 38.92 months. Majority (56%) presented with a solitary lesion and 76% had negative margins. Histopathological analysis revealed various stages of cancer, with the most common being high-grade squamous epithelial neoplasia (64%). Survival analysis indicated a 72% overall survival rate, with 16% surviving beyond 5-year post-treatment. Approximately, 20% succumbed during the study, primarily due to non-esophageal causes (16%). Conclusion: Endoscopic treatment shows promise as a therapeutic option for selected esophageal cancer patients, offering favorable outcomes in terms of survival and disease control. Further prospective studies are warranted to validate these findings and optimize patient selection criteria for endoscopic interventions in esophageal cancer management.

High-performance proton exchange membranes are of great importance for fuel cells. Here, we have synthesized polycarboxylate plasticizer modified MIL-101-Cr-NH2(PCP-MCN), a kind of hybrid metal−organic framework, which exhibits a superior proton conductivity. PCP-MCN nanoparticles are used as additives to fabricate PCP-MCN/Nafion composite membranes. Microstructures and characteristics of PCP-MCN and these membranes have been extensively investigated. Significant enhancement in proton conduction for PCP-MCN around 55 °C is interestingly found due to the thermal motion of the PCP molecular chains. Robust mechanical properties and higher thermal decomposition temperature of the composite membranes are directly ascribed to strong intermolecular interactions between PCP-MCN and Nafion side chains, i.e., the formation of substantial acid–base pairs (-SO3−···+H–NH-), which further improves compatibility between additive and Nafion matrix. At the same humidity and temperature condition, the water uptake of composite membranes significantly increases due to the incorporation of porous additives with abundant functional groups and thus less crystallinity degree in comparison to pristine Nafion. Proton conductivity (σ) over wide ranges of humidities (30 − 100% RH at 25 °C) and temperatures (30 − 98 °C at 100% RH) for prepared membranes is measured. The σ in PCP-MCN/Nafion composite membranes is remarkably enhanced, i.e. 0.245 S/cm for PCP-MCN-3wt.%/Nafion is twice that of Nafion membrane at 98 °C and 100% RH, because of the establishment of well-interconnected proton transport ionic water channels and perhaps faster protonation–deprotonation processes. The composite membranes possess weak humidity-dependence of proton transport and higher water uptake due to excellent water retention ability of PCP-MCN. In particular, when 3 wt.% PCP-MCN was added to Nafion, the power density of a single-cell fabricated with this composite membrane reaches impressively 0.480, 1.098 W/cm2under 40% RH, 100% RH at 60 °C, respectively, guaranteeing it to be a promising proton exchange membrane.

The development of efficient and stable electrocatalysts for direct methanol fuel cells (DMFCs) remains a challenge. Herein, nitrogen, sulfur codoped porous carbon spheres (N/S–Cs) were prepared by the extended Stöber method employing hexamethylenetetramine (HMTA) as both the nitrogen source and cross-linking agent and 4,4′-sulfonyldiphenol as the sulfur source. The results showed that N/S–Cs-800 has a large specific surface area (1098 m2g–1) and plentiful nitrogen contents (6.96 wt %). Electrochemical tests demonstrated that N/S–Cs-800 exhibited excellent catalytic activity for oxygen reduction (Eonset= 0.995 V, E1/2= 0.863 V) in alkaline media. Its long-term durability and high tolerance to methanol surpass that of the Pt/C catalyst. Furthermore, the ethylene glycol reduction method was used to load Pt nanoparticles on N/S–Cs-800 support for methanol oxidation. The mass activity was about 858.54 mA mgPt−1, and it showed good long-term stability and anti-CO poisoning ability. This work provides a feasible pathway to design high efficiency metal-free catalysts for DMFCs.

The traditional chromate passivation process for lithium copper foils is hazardous and does not meet the requirements of green development. Therefore, in this study, utilizing the characteristics of gluconate coordinated with Cu2+, gluconate is adsorbed on the surface of copper foil, providing an anchor point for the action of JH-M902. Its rich hydroxyl structure (C–OH) and the Si–OH produced by the hydrolysis of JH-M902 dehydrate and condense to form Si–O–C. Then, WO42+and Cu2+formed by CuWO4fill in the film layer. These three passivators synergistically construct a W–Si passivation film with excellent corrosion resistance. Electrochemical tests show that the corrosion inhibition rate of the passivated film is as high as 98.08%, which is much higher than that of the existing chromate passivation process. The process proposed in this study can promote the development of chrome-free passivation of lithium copper foil and provides new ideas for the protection of other metal materials.

Photodynamic therapy (PDT) has emerged as a preferred nonsurgical treatment in clinical applications due to its capacity to selectively eradicate diseased tissues while minimizing damage to normal tissue. Nevertheless, its clinical efficacy is constrained by the limited penetration of visible light. Although near-infrared (NIR) lasers offer enhanced tissue penetration, the dearth of suitable photosensitizers and a pronounced imaging-treatment disparity pose challenges. Additionally, clinical implementation via optical fiber implantation carries infection risks and necessitates minimally invasive surgery, contradicting PDT’s noninvasive advantage. In this study, we introduce a brilliant approach utilizing aggregation-induced emission luminogens (AIEgen) to develop a visible-light penetrator (VLP), coupled with wireless light emitting diodes (LEDs), enabling deep photodynamic therapy. We validate the therapeutic efficacy of this visible-light penetrator in tissues inaccessible to conventional PDT, demonstrating significant suppression of inflammatory diffusion in vivousing AIEgen TBPPM loaded within the VLP, which exhibits a transmittance of 86% in tissues with a thickness of 3 mm. This innovative visible-light penetrator effectively overcomes the substantial limitations of PDT in clinical settings and holds promise for advancing phototherapy.

Tumor cell death induced by “cuproptosis” is a novel form of tumor death that differs from apoptosis induced by chemotherapy. It is expected to emerge as a new approach for cancer treatment. In this study, our focus was on exploiting the characteristic of “cuproptosis” which necessitates increased aerobic respiration to induce tumor cell death. To achieve this, we developed a novel drug delivery system using a CaCO3@CuO2lipid coating (CaCO3@CuO2@L). This system aimed to comprehensively modulate the tumor microenvironment and trigger “cuproptosis” in hepatocellular carcinoma (HCC) through the interaction between copper ions and peroxides. Experimental results revealed that the CaCO3@CuO2@L exhibited a distinct watermelon shape, with CuO2evenly distributed within the CaCO3nanoparticles. The nanoparticles had an average size of approximately 191 nm. In vitro studies demonstrated that the nanoparticles released CuO2in a slightly acidic environment while simultaneously elevating pH levels, reducing glutathione (GSH), and increasing oxygen production. Within liver cancer cells, the CaCO3@CuO2@L effectively regulated the acidity, GSH levels, and oxygen-depleted microenvironment through the “trinity” mechanism, ultimately inducing “cuproptosis” in HCC. Furthermore, in mouse models with transplanted tumors and orthotopic liver cancer tumors, the CaCO3@CuO2@L significantly suppressed tumor growth. By triggering “cuproptosis” in HCC, this study offers valuable insights for developing a comprehensive treatment approach for HCC. Ultimately, this research may pave the way for the clinical implementation of the drug delivery system based on “cuproptosis” in liver cancer treatment.

Solar-driven carbon dioxide (CO2) reduction into C2+products such as ethylene represents an enticing route toward achieving carbon neutrality. However, due to sluggish electron transfer and intricate C–C coupling, it remains challenging to achieve highly efficient and selective ethylene production from CO2and H2O beyond capitalizing on Cu-based catalysts. Herein, we report a judicious design to attain asymmetric C–C coupling through interfacial defect-rendered tandem catalytic centers within a sulfur-vacancy-rich MoSx/Fe2O3photocatalyst sheet, enabling a robust CO2photoreduction to ethylene without the need for copper, noble metals, and sacrificial agents. Specifically, interfacial S vacancies induce adjacent under-coordinated S atoms to form Fe–S bonds as a rapid electron-transfer pathway for yielding a Z-scheme band alignment. Moreover, these S vacancies further modulate the strong coupling interaction to generate a nitrogenase-analogous Mo–Fe heteronuclear unit and induce the upward shift of the d-band center. This bioinspired interface structure effectively suppresses electrostatic repulsion between neighboring *CO and *COH intermediates via d-p hybridization, ultimately facilitating an asymmetric C–C coupling to achieve a remarkable solar-to-chemical efficiency of 0.565% with a superior selectivity of 84.9% for ethylene production. Further strengthened by MoSx/WO3, our design unveils a promising platform for optimizing interfacial electron transfer and offers a new option for C2+synthesis from CO2and H2O using copper-free and noble metal-free catalysts.

• A Co 42 Cr 20 Ni 30 Ti 4 Al 4 quinary medium entropy alloy (MEA) with excellent mechanical properties over a wide temperature ranging from 77 to 873 K. • Post heat treatment leads to refinement of grains and heterogeneous precipitation of L1 2 phase. • Stacking faults mediated deformation behavior of the SLMed Co 42 Cr 20 Ni 30 Ti 4 Al 4 MEA. Modern engineering has long been in demand for high-performance additive manufactured materials for harsh working conditions. The idea of high entropy alloy (HEA), medium entropy alloy (MEA), and multi-principal-element alloy (MPEA) provides a new way for alloy design. In this work, we develop a Co 42 Cr 20 Ni 30 Ti 4 Al 4 quinary MEA which exhibits a superiority of mechanical properties over a wide temperature ranging from 77 to 873 K via selective laser melting (SLM) and post-heat treatment. The present MEA achieves an excellent ultimate tensile strength (UTS) of 1586 MPa with a total elongation (TE) of 22.7 % at 298 K, a UTS of 1944 MPa with a TE of 22.6 % at 77 K, and a UTS of 1147 MPa with a TE of 9.1 % at 873 K. The excellent mechanical properties stem from the microstructures composed of partially refined grains and heterogeneously precipitated L1 2 phase due to the concurrence of recrystallization and precipitation. The grain boundary hardening, precipitation hardening, and dislocation hardening contribute to the high YS at 298 and 77 K. Interactions of nano-spaced stacking faults (SFs) including SFs networks, Lomer–Cottrell locks (L–C locks), and anti-phase boundaries (APBs) induced by the shearing of L1 2 phase are responsible for the high strain hardening rate and plasticity at 77 K. Our work provides a new insight for the incorporation of precipitation hardening and additive manufacturing technology, paving the avenue for the development of high-performance structural materials. [Display omitted] [ABSTRACT FROM AUTHOR]

A numerical study based on a two-dimensional two-phase SPH (Smoothed Particle Hydrodynamics) model to analyze the action of water waves on open-type sea access roads is presented. The study is a continuation of the analyses presented by Chen et al. (2022), in which the sea access roads are semi-immersed. In this new configuration, the sea access roads are placed above the still water level, therefore the presence of the air phase becomes a relevant issue in the determination of the wave forces acting on the structures. Indeed, the comparison of wave forces on the open-type sea access roads obtained from the single and two-phase SPH models with the experimental results shows that the latter are in much better agreement. So in the numerical simulations, a two-phase δ-SPH model is adopted to investigate the dynamical problems. Based on the numerical results, the maximum horizontal and uplifting wave forces acting on the sea access roads are analyzed by considering different wave conditions and geometries of the structures. In particular, the presence of the girder is analyzed and the differences in the wave forces due to the air cushion effects which are created below the structure are highlighted.

The Lower Silurian Longmaxi Formation is a major horizon for shale gas development in the Changning block in Sichuan Basin. In this study, the fracture features in the Longmaxi Formation in the Changning block were investigated through outcrop observation, core description, inclusion testing, and Fullbore Formation MicroImage logging. The results showed that tectogenetic shear fractures dominated the Longmaxi Formation, with maximum principal stress in the SSW, NW, and NE directions. According to fracture features and stress analysis, three phases of faults were generated by tectonic movements after the buried depth reached a maximum in the study area: phase 1 nearly EW-trending faults formed in the middle-late Yanshanian, phase 2 NE-trending faults formed from the end of the Yanshanian to the early Himalayan, and phase 3 NW-trending faults formed from the middle Himalayan to the present. According to the regression analysis of the tested shale gas production and faults, the faults with a throw of > 200 m could have a great effect on gas production, and high-yield wells were generally located over 1 km away from faults. Based on the dissection of tectonic styles and typical wells drilled in the Changning block, three shale gas accumulation models were established: wide gentle syncline + internal smalxl fault, subdued anticline + large fault, and slope + internal fault. The study reveals the mechanism of fault influence on shale gas accumulation in the Longmaxi Formation in the Changning block and provides a reference for efficient shale gas production in the Sichuan Basin.

This article investigates the multi-vehicle task assignment and motion planning (MVTAMP) problem. In a dense environment, a fleet of non-holonomic vehicles is appointed to visit a series of target positions and then move to a specific ending area for real-world applications such as clearing threat targets, aid rescue, and package delivery. We presented a novel hierarchical method to simultaneously address the multiple vehicles’ task assignment and motion planning problem. Unlike most related work, our method considers the MVTAMP problem applied to non-holonomic vehicles in large-scale scenarios. At the high level, we proposed a novel distributed algorithm to address task assignment, which produces a closer to the optimal task assignment scheme by reducing the intersection paths between vehicles and tasks or between tasks and tasks. At the low level, we proposed a novel distributed motion planning algorithm that addresses the vehicle deadlocks in local planning and then quickly generates a feasible new velocity for the non-holonomic vehicle in dense environments, guaranteeing that each vehicle efficiently visits its assigned target positions. Extensive simulation experiments in large-scale scenarios for non-holonomic vehicles and two real-world experiments demonstrate the effectiveness and advantages of our method in practical applications. The source code of our method can be available at https://github.com/wuuya1/LRGO. Note to Practitioners—The motivation for this article stems from the need to solve the multi-vehicle task assignment and motion planning (MVTAMP) problem for non-holonomic vehicles in dense environments. Many real-world applications exist, such as clearing threat targets, aid rescue, and package delivery. However, when vehicles need to continuously visit a series of assigned targets, motion planning for non-holonomic vehicles becomes more difficult because it is more likely to occur sharp turns between adjacent target path nodes. In this case, a better task allocation scheme can often lead to more efficient target visits and save all vehicles’ total traveling distance. To bridge this, we proposed a hierarchical method for solving the MVTAMP problem in large-scale complex scenarios. The numerous large-scale simulations and two real-world experiments show the effectiveness of the proposed method. Our future work will focus on the integrated task assignment and motion planning problem for non-holonomic vehicles in highly dynamic scenarios.

Sulfur dioxide (SO2) fumigation was studied in laboratory to determine its potential as an alternative treatment for postharvest control of stored product insects, confused flour beetle, Tribolium confusumJacquelin du Val (Coleoptera: Tenebrionidae), and rice weevil, Sitophilus oryzae(L.) (Coleoptera: Curculionidae). Three-hour fumigations with 0.1%–2.0% SO2were conducted against eggs, immature stages, and adults of the 2 insects at 20 °C. Effective control of both insects was achieved. However, there were considerable variations between the 2 insects and among different life stages. Confused flour beetle was more susceptible to SO2fumigation than rice weevil. Complete control of adults and all life stages of confused flour beetle was achieved in 3-h fumigations with 0.5% and 2.0% SO2, respectively. For rice weevil, 3-h fumigation with 1.5% SO2resulted in 96.5% adult mortality and the fumigation with 2.0% SO2resulted in 99.27% mortality of adults and 87.5% mortality of immature stages. Three-hour fumigations with 1% SO2resulted in <5% egg survival to adults. The study demonstrated high efficacy of SO2fumigation against the insects and suggested that SO2fumigation has good potential for postharvest pest control on stored products.

Recommender systems have been widely applied in different real-life scenarios to help us find useful information. In particular, reinforcement learning (RL)-based recommender systems have become an emerging research topic in recent years, owing to the interactive nature and autonomous learning ability. Empirical results show that RL-based recommendation methods often surpass supervised learning methods. Nevertheless, there are various challenges in applying RL in recommender systems. To understand the challenges and relevant solutions, there should be a reference for researchers and practitioners working on RL-based recommender systems. To this end, we first provide a thorough overview, comparisons, and summarization of RL approaches applied in four typical recommendation scenarios, including interactive recommendation, conversational recommendation, sequential recommendation, and explainable recommendation. Furthermore, we systematically analyze the challenges and relevant solutions on the basis of existing literature. Finally, under discussion for open issues of RL and its limitations of recommender systems, we highlight some potential research directions in this field.

For cross-domain pattern classification, the supervised information (i.e., labeled patterns) in the source domain is often employed to help classify the unlabeled target domain patterns. In practice, multiple target domains are usually available. The unlabeled patterns (in different target domains) which have high-confidence predictions, can also provide some pseudo-supervised information for the downstream classification task. The performance in each target domain would be further improved if the pseudo-supervised information in different target domains can be effectively used. To this end, we propose an evidential multi-target domain adaptation (EMDA) method to take full advantage of the useful information in the single-source and multiple target domains. In EMDA, we first align distributions of the source and target domains by reducing maximum mean discrepancy (MMD) and covariance difference across domains. After that, we use the classifier learned by the labeled source domain data to classify query patterns in the target domains. The query patterns with high-confidence predictions are then selected to train a new classifier for yielding an extra piece of soft classification results of query patterns. The two pieces of soft classification results are then combined by evidence theory. In practice, their reliabilities/weights are usually diverse, and an equal treatment of them often yields the unreliable combination result. Thus, we propose to use the distribution discrepancy across domains to estimate their weighting factors, and discount them before fusing. The evidential combination of the two pieces of discounted soft classification results is employed to make the final class decision. The effectiveness of EMDA was verified by comparing with many advanced domain adaptation methods on several cross-domain pattern classification benchmark datasets.

Creep rupture of the reduced activation ferritic/martensitic (RAFM) steel and 316L steel dissimilar joint by friction stir welding was investigated. The creep rupture time of the dissimilar joint was 1941 h at 600 °C/100 MPa and 120 h at 650 °C/100 MPa. The creep fracture occurred in heat affect zone (HAZ) of RAFM steel side where coarse Laves phase was detected. The formation and coarsening of the Laves phase particles in HAZ of RAFM steel side were the main reasons that caused the creep fracture of the dissimilar joint. The Laves phase particles nucleated adjacent to the large M23C6particles at the grain boundaries where W element segregated and grew fast during creep exposure. The large Laves phase would deteriorate the pinning effect of M23C6carbides and weaken the solid solution strengthening effect. Besides, the size of the Laves phase in HAZ of RAFM steel side was bigger than that in stir zone of RAFM steel side. These reasons explain the creep fracture in HAZ of RAFM steel side of dissimilar joint.

The present work proposes an actively mode-locked modulator-free optoelectronic oscillator (OEO), which is based on a directly modulated laser (DML) with no need for any electro-optic or electrical amplitude modulators. The conception is that the DML plays a triple role, that is, the laser source, the OEO modulator of the optical signal, and the active mode-locking modulator of the cavity loss. Meanwhile, a direct-current (DC) block is employed at the RF port of the DML, which acts as a differentiator besides isolating the DC component. An external modulating square wave can be converted to a series of spike pulses via the DC block and then applied on the DML, which empowers the DML as the active mode-locking module. By relating the external frequency with the free spectral range (FSR) of the OEO cavity in an integral multiple or a rational number multiple, the fundamental, the harmonic, and the rational number harmonic mode-locking can be realized. Accordingly, microwave pulse trains with flexibly tunable pulse repetition frequency (PRF) can be obtained. Notably, PRF-staggered pulses in one cavity period can be optically obtained. In the proof-of-concept experiment, microwave pulse trains with the tunable PRF are demonstrated by driving the actively mode-locked direct modulation OEO (AML-DM-OEO) under the fundamental mode-locking, the 2nd-order, 3rd-order, and 4th-order harmonic mode-locking, and the 5th-order rational number harmonic mode-locking states, respectively. The phase noise is better than −118 dBc/Hz@10 kHz for the single-tone signal at 10 GHz abstracted from the generated microwave pulses. Staggered pulses are successfully obtained with double-pulse, triple-pulse, and quadruple-pulse in one cavity period, respectively.

ABSTRACTPurposeThe aim of this study was to analyze the characteristics of CT-measured intersection angle (FB-BNLD) between the frontal bone and bony nasolacrimal duct and to provide suggestions for treating primary acquired nasolacrimal duct obstruction (PANDO) patients in West China.MethodsThree hundred and nine participants’ CT were, respectively, evaluated with RadiAnt DICOM Viewer. We defined the FB-BNLD angle >0° as the anterior type and the FB-BNLD angle ≤0° as the posterior type.ResultsThe mean FB-BNLD was −2.52° (95% CI, −3.16° to −1.88°) across all participants, of whom 37.2% were of the anterior type and 62.8% of the posterior type. Approximately 65.0% of the female patients had a posterior FB-BNLD type, and 54.2% of the male patients had an anterior FB-BNLD type (p = .002). Posterior FB-BNLD was the dominant type in the PANDO and control groups (p = .011), and the angle of FB-BNLD was statistically different in both groups (PANDO group, −2.54° to −0.71°; control group, −4.42° to −2.67°; p < .001). Among the male participants, the type of FB-BNLD differed between the two groups (p = .036), with differences in the angle of FB-BNLD (PANDO group, 0.59° to 5.13°; control group, −4.08° to 1.89°; p = .034). There was no difference in the type of FB-BNLD in female participants between the two groups (p = .051).ConclusionThe present study revealed individual differences in the type of FB-BNLD, with anterior-type majority in males and posterior-type dominance in females. Evaluating the FB-BNLD type on CT can provide a fast method for knowing the nasolacrimal duct condition during planning for lacrimal manipulation.

Self-supervised learning (SSL) has recently achieved impressive performance on various time series tasks. The most prominent advantage of SSL is that it reduces the dependence on labeled data. Based on the pre-training and fine-tuning strategy, even a small amount of labeled data can achieve high performance. Compared with many published self-supervised surveys on computer vision and natural language processing, a comprehensive survey for time series SSL is still missing. To fill this gap, we review current state-of-the-art SSL methods for time series data in this article. To this end, we first comprehensively review existing surveys related to SSL and time series, and then provide a new taxonomy of existing time series SSL methods by summarizing them from three perspectives: generative-based, contrastive-based, and adversarial-based. These methods are further divided into ten subcategories with detailed reviews and discussions about their key intuitions, main frameworks, advantages and disadvantages. To facilitate the experiments and validation of time series SSL methods, we also summarize datasets commonly used in time series forecasting, classification, anomaly detection, and clustering tasks. Finally, we present the future directions of SSL for time series analysis.

Existing multiple kernel clustering (MKC) algorithms have two ubiquitous problems. From the theoretical perspective, most MKC algorithms lack sufficient theoretical analysis, especially the consistency of learned parameters, such as the kernel weights. From the practical perspective, the high complexity makes MKC unable to handle large-scale datasets. This paper tries to address the above two issues. We first make a consistency analysis of an influential MKC method named Simple Multiple Kernel $k$k-Means (SimpleMKKM). Specifically, suppose that $\hat{\boldsymbol{\gamma }}_{n}$γ^n are the kernel weights learned by SimpleMKKM from the training samples. We also define the expected version of SimpleMKKM and denote its solution as $\boldsymbol{\gamma }^*$γ*. We establish an upper bound of $\Vert \hat{\boldsymbol{\gamma }}_{n}-\boldsymbol{\gamma }^*\Vert _\infty$∥γ^n-γ*∥∞ in the order of $\widetilde{\mathcal {O}}(1/\sqrt{n})$O˜(1/n), where $n$n is the sample number. Based on this result, we also derive its excess clustering risk calculated by a standard clustering loss function. For the large-scale extension, we replace the eigen decomposition of SimpleMKKM with singular value decomposition (SVD). Consequently, the complexity can be decreased to $\mathcal {O}(n)$O(n) such that SimpleMKKM can be implemented on large-scale datasets. We then deduce several theoretical results to verify the approximation ability of the proposed SVD-based method. The results of comprehensive experiments demonstrate the superiority of the proposed method.

The present study investigated the flow dynamics and heat transfer behavior of molten steel during converter steel tapping. This research lays the groundwork for designing accurate post-furnace temperature models and enabling lower tapping temperatures. Considering the influence of slag and the Coriolis force on the process, a mathematical model was developed to characterize both the transient flow dynamics of molten steel within the converter and the heat dissipation of the steel stream in the air. Moreover, a physical model constructed based on similarity principles elucidated the impacts of standing time and the Coriolis force on the remaining steel volume in the converter at the moment of slag entrapment. Results showed that as standing time increased or tapping hole size decreased, the remaining steel volume in the converter decreased. When the Coriolis force was included, the percentage of remaining steel volume was lower compared to simulations without it. Numerical simulations precisely determined the trajectory of the steel stream in the air domain, facilitating improved coordination between the converter and the ladle. Furthermore, at a tapping temperature of 1700°C and a tapping hole size of 140 mm, heat loss of the steel stream in the air domain led to a temperature decrease of 1.02°C. Heat loss decreased with increasing hole size. Lowering the tapping temperature by 50°C could reduce the temperature decrease from the air domain by 0.1°C. Finally, the study proposes preliminary strategies for lowering tapping temperature.

In ophthalmology, the quality of fundus images is critical for accurate diagnosis, both in clinical practice and in artificial intelligence (AI)-assisted diagnostics. Despite the broad view provided by ultrawide-field (UWF) imaging, pseudocolor images may conceal critical lesions necessary for precise diagnosis. To address this, we introduce UWF-Net, a sophisticated image enhancement algorithm that takes disease characteristics into consideration. Using the Fudan University ultra-wide-field image (FDUWI) dataset, which includes 11 294 Optos pseudocolor and 2415 Zeiss true-color UWF images, each of which is rigorously annotated, UWF-Net combines global style modeling with feature-level lesion enhancement. Pathological consistency loss is also applied to maintain fundus feature integrity, significantly improving image quality. Quantitative and qualitative evaluations demonstrated that UWF-Net outperforms existing methods such as contrast limited adaptive histogram equalization (CLAHE) and structure and illumination constrained generative adversarial network (StillGAN), delivering superior retinal image quality, higher quality scores, and preserved feature details after enhancement. In disease classification tasks, images enhanced by UWF-Net showed notable improvements when processed with existing classification systems over those enhanced by StillGAN, demonstrating a 4.62% increase in sensitivity (SEN) and a 3.97% increase in accuracy (ACC). In a multicenter clinical setting, UWF-Net-enhanced images were preferred by ophthalmologic technicians and doctors, and yielded a significant reduction in diagnostic time ((13.17 ± 8.40) s for UWF-Net enhanced images vs (19.54 ± 12.40) s for original images) and an increase in diagnostic accuracy (87.71% for UWF-Net enhanced images vs 80.40% for original images). Our research verifies that UWF-Net markedly improves the quality of UWF imaging, facilitating better clinical outcomes and more reliable AI-assisted disease classification. The clinical integration of UWF-Net holds great promise for enhancing diagnostic processes and patient care in ophthalmology.

Bolt tightening is widely used in various types of aerospace engine assembly process. To ensure the tightness of the aerospace engine, it is extremely important to reduce the deformation of the engine and achieve uniform bolt preload in the tightening process. The initial clearance value between nozzle and chamber of different types of aerospace engine is different, which makes the tightening target of low deformation and even load difficult. Besides, the impact of elastic interaction makes the tightening of aerospace engine very challenging. And aerospace engine bolt tightening is too time-consuming. A method is proposed in this paper to solve the above puzzle. The influence of elastic interaction in the tightening process can be weakened by optimizing the tightening sequence, and the optimal tightening process of different types of aerospace engines under different clearance values can be predicted based on kriging model. The experiment shows that the deformation of the engine deduce by 82% and 72%, the average preload deduce by 3.8% and 5.1%. Then kriging model is used to predict the best tightening sequence under different types and tightening clearance. The genetic algorithm is used to obtain the best sequence in the design domain, and the deviation between the final predicted value and the simulation value is less than 3%. After comparing, the research also improves the efficiency of process planning of and medium-sized batches and multiple batches of aerospace engines.

Materials with delayed fluorescence (DF) are characterized by an available triplet state and prolonged fluorescence lifetime. In particular, those thermally activated delayed fluorescence (TADF) materials are representative of those that have been well investigated and developed. The features of DF materials ensured their impressive photophysical properties and device performances in various areas of optoelectronic applications. Recently, researchers began to apply them into biological fields and achieved great developments in bioapplications. In this Review, typical design strategies for DF molecules were presented, and some recently published TADF and aggregation-induced DF (AIDF) materials within biological fields, involving fluorescent biosensing, time-gated bioimaging, and photodynamic therapies, were systematically summarized. Furthermore, future perspectives of DF materials in biological fields were discussed, and aspects of enhancements were proposed.

As an important component of lithium-ion batteries, all-solid-state electrolytes should possess high ionic conductivity, excellent flexibility, and relatively high mechanical strength. All-solid-state polymer electrolytes (ASSPEs) based on polymers seem to be able to meet these requirements. However, pure ASSPEs have relatively low ionic conductivity, and the addition of inorganic fillers such as lithium salts will reduce their flexibility and mechanical strength. To address the above issues, in this paper, the solvent-free method was used to prepare a poly(vinylidenefluoride-co-hexafluoropropylene)/lithium bis(trifluoromethanesulfonyl) imide/poly(ethylene oxide) all-solid-state polymer electrolyte, which was then subjected to 4 × 4 magnification synchronous bidirectional stretching. Subsequently, it was multilayered with PEO-based composite polymer electrolytes to obtain multilayered composite polymer electrolytes (MCPEs). Bidirectional stretching provides superior in-plane and out-of-plane mechanical properties to MCPEs by inducing molecular chain orientation, which suppresses the growth of lithium dendrites. Concurrently, it facilitates the formation of the β-crystal form of PVDF-HFP, thereby weakening the ion solvation effect and reducing the lithium-ion migration energy barrier. Multilayered compounding improves the interfacial contact between MCPEs and electrodes, thereby reducing the interfacial impedance. Experiments have demonstrated that the MCPEs prepared in this paper exhibit high ionic conductivity at room temperature (1.83 × 10–4S cm–1), low interfacial resistance (547 Ω cm–2), excellent mechanical properties (26 MPa), and excellent cycling rate performance (a capacity retention rate of 90% after 110 cycles at 0.1 C), which can meet the performance requirements of lithium-ion batteries for ASSPEs.

Thioesters make up an important class of bioactive compounds. Due to their chemoselectivity, they have been widely used in the synthesis of a wide range of complex bioactive molecules and natural products. At present, chemists have developed a variety of methods for the preparation of thioester compounds. However, these methods usually require the use of transition metal catalysis or harsh reaction conditions. The strategy of synthesizing thioester compounds via visible light-induced electron donor–acceptor (EDA) complex reactions avoids the problems associated with conventional methods through the development of photocatalysis. Here we report a sustainable method for thiocarbonylating aryl sulfonium salts via a visible light-induced EDA complex process without transition metals.

The environment observations and learning experiences shared by the cooperative learning agents accelerate multi-agent reinforcement learning (MARL) with partial observations for mobile applications but the performance degrades due to the redundant and outdated observations under severe channel fading in wireless networks. In this paper, we propose an efficient communication scheme in MARL for mobile applications that enables each learning agent to optimize the cooperative agents and the learning parameters to integrate the shared information. The cooperative agents are chosen according to the learning environment observations, the channel states, and the task similarity with neighboring agents. The learning parameters are chosen based on the attention mechanism that exploits the correlation with the local observation to enhance the agent receptive field for efficient policy exploration. Neural networks with weights updated based on the learning factors determined by the task similarity are designed to further improve the learning efficiency. The performance bounds including the information gain from the learning agent cooperation, the communication cost and the utility are provided based on the Nash equilibrium of the cooperative MARL communication game. The proposed scheme is implemented in the anti-jamming video transmission of the unmanned aerial vehicle swarms to optimize the transmit channel and power and experimental results verify the performance gain over the benchmark.

In microdevices, local temperature serves as a crucial indicator of device performance. Traditional measurements often struggle to capture transient thermal events due to the small and short-lived nature of the heat generation region within the device. In this study, we use scanning probe microscopy with a fiber Bragg grating (FBG) near-field probe to identify hot spot locations precisely. We employ the time-stretched dispersive Fourier transform (TS-DFT) technique to enhance the demodulation rate, facilitating the mapping of the probe signal from the frequency domain to the time domain. Through numerical calculations of the optical pulse propagation process in a dispersive medium and conjunction with near-field temperature calibration experiments, we validate the capability of this technique to achieve the demodulation rate of 51.43 MHz. In practice, our measurements show that when subjected to an electrical excitation signal of 2 ms, the microheater leads a thermal impulse response of about 6 °C, and the measurement process does not affect its regular operation.

The sealed-bid auction enables bidders to secretly send their bids to the auctioneer, which compares all bids and publishes the winning one on the bid-opening day. This type of auction is friendly for protecting the bid privacy, and sufficiently fair for all bidders if the auctioneer acts faithfully. Unfortunately, the auctioneer may not always be trustworthy. The auctioneer has the ability to deliberately leak any bid information to a part of bidders for raising the final winning price based on the investigation. Meanwhile, the auctioneer can appoint any bidder as the winner, as long as the bidder accepts a higher winning price than the current highest bid. Since bidders cannot obtain any bid information from others, to the best of our knowledge, it is difficult to prevent bid leakage from the auctioneer, and support bidders to verify the bid comparison results without disclosing the winning bid, simultaneously. To alleviate these problems, we first construct a homomorphic encryption(HE)-based bid comparison circuit. All bidders can directly compute a cipher of the winning bid by using this circuit; hence, the winning bid does not need to be exposed to all bidders. Then, we propose a blockchain-based sealed-bid scheme (BSS) by integrating the circuit with commitment and zero-knowledge proof. The auctioneer only obtains the commitments of bids before the bid-opening day, and he has to prove that the winner's bid is the same as the plaintext of the bidders’ computed cipher. Thus, the auctioneer can neither leak the bid information nor publish a higher winning price during in the auction. Detailed performance analysis shows that the computational complexity of BSS is linear with the binary length of bids.

As a significant geometric feature of 3D point clouds, sharp features play an important role in shape analysis, 3D reconstruction, registration, localization, etc. Current sharp feature detection methods are still sensitive to the quality of the input point cloud, and the detection performance is affected by random noisy points and non-uniform densities. In this paper, using the prior knowledge of geometric features, we propose a Multi-scale Laplace Network (MSL-Net), a new deep-learning-based method based on an intrinsic neighbor shape descriptor, to detect sharp features from 3D point clouds. First, we establish a discrete intrinsic neighborhood of the point cloud based on the Laplacian graph, which reduces the error of local implicit surface estimation. Then, we design a new intrinsic shape descriptor based on the intrinsic neighborhood, combined with enhanced normal extraction and cosine-based field estimation function. Finally, we present the backbone of MSL-Net based on the intrinsic shape descriptor. Benefiting from the intrinsic neighborhood and shape descriptor, our MSL-Net has simple architecture and is capable of establishing accurate feature prediction that satisfies the manifold distribution while avoiding complex intrinsic metric calculations. Extensive experimental results demonstrate that with the multi-scale structure, MSL-Net has a strong analytical ability for local perturbations of point clouds. Compared with state-of-the-art methods, our MSL-Net is more robust and accurate.

Girth welding is extensively utilized in connecting pipeline structures. Excessive welding deformation may result in a decrease in ultimate bearing capacity and fatigue life. This research explores the characteristics and underlying mechanics of radial deformation in thin-walled girth welded pipes and proposes mitigation measures. Initially, several 304 stainless steel pipes were joined utilizing four distinct welding procedures. Welding deformation and residual stress distribution were measured using a 3D coordinate measuring machine (CMM) and the Cos αX-ray diffraction (XRD) method, respectively. Subsequently, numerical models were developed to simulate the welding process and verify the accuracy of the fusion zone, residual stress field, and radial deformation in the models. The mechanism of radial deformation was further elucidated based on the inherent strain method. The results show that the direction of radial deformation relies substantially on the hoop plastic strain. Hoop tensile plastic strain corresponds to radial outward deformation, while hoop shrinkage plastic strain corresponds to radial inward deformation. In addition, the influence of axial constraint on the formation process and results of hoop plastic strain, as well as the effect of heating zone width on radial deformation, was studied. Through simulation outcomes, the research suggests employing a regulated heat source method to mitigate radial deformation, which could potentially reduce deformation by 92.8%, offering considerable advantages in deformation mitigation and welding efficiency augmentation.

The time-delayed fast-slow dynamics in broadband optoelectronics oscillators (OEOs) have given birth to many important nonlinear phenomena and promising applications. In this paper, we numerically and experimentally demonstrate a late-model compound dissipative soliton dynamic phenomenon, i.e., periodic pulsating soliton packets, in a dual-time-delayed feedback broadband OEO. The dual-time-delayed feedback module is constructed by inserting a dual-branch electrical differential module into the broadband OEO loop. The pulsating dissipative solitons (DSs) are excited by injecting a perturbation signal into the OEO. With the increasing of cycles, the soliton energy is gradually coupled from the short delay loop to the long delay loop, which leads to the energy dissipation of the solitons at the front of the soliton packet. Meanwhile, the soliton energy at the back of the soliton packet is enhanced. In addition, the DSs are subjected to the amplitude attenuation induced by the spiking-DS refractory period. The total energy loss is compensated by the loop gain. Finally, the OEO system reaches nonlinear equilibrium regime, and undergoes regular pulsation in the time domain. The pulsating characteristics are analyzed in detail. This finding not only paves the way for the compound soliton dynamic research in time-delayed OEOs, but also is valuable for applications such as multi-band dense coherent microwave frequency comb generation, radar and anti-interference.

Objective. The paraspinal muscles have been extensively studied on axial lumbar magnetic resonance imaging (MRI) for better understanding of back pain; however, the acquisition of measurements mainly relies on manual segmentation, which is time consuming. The study objective was to develop and validate a deep-learning-based program for automated acquisition of quantitative measurements for major lumbar spine components on axial lumbar MRIs, the paraspinal muscles in particular. Methods. This study used a cross-sectional observational design. From the Hangzhou Lumbar Spine Study, T2-weighted axial MRIs at the L4-5 disk level of 120 participants (aged 54.8 years [SD = 15.0]) were selected to develop the deep-learning-based program Spine Explorer (Tulong). Another 30 axial lumbar MRIs were automatically measured by Spine Explorer and then manually measured using ImageJ to acquire quantitative size and compositional measurements for bilateral multifidus, erector spinae, and psoas muscles; the disk; and the spinal canal. Intersection-over-union and Dice score were used to evaluate the performance of automated segmentation. Intraclass coefficients and Bland-Altman plots were used to examine intersoftware agreements for various measurements. Results. After training, Spine Explorer (Tulong) measures an axial lumbar MRI in 1 second. The intersections-over-union were 83.3% to 88.4% for the paraspinal muscles and 92.2% and 82.1% for the disk and spinal canal, respectively. For various size and compositional measurements of paraspinal muscles, Spine Explorer (Tulong) was in good agreement with ImageJ (intraclass coefficient = 0.85 to approximately 0.99). Conclusion. Spine Explorer (Tulong) is automated, efficient, and reliable in acquiring quantitative measurements for the paraspinal muscles, the disk, and the canal, and various size and compositional measurements were simultaneously obtained for the lumbar paraspinal muscles. Impact. Such an automated program might encourage further epidemiological studies of the lumbar paraspinal muscle degeneration and enhance paraspinal muscle assessment in clinical practice. Keywords: Spine, Paraspinal Muscles, Artificial Intelligence, Magnetic Resonance Imaging, Algorithms, Introduction Most adults have suffered from back pain sometime in their lives. (1) As one of the major components in the lumbar spine, the paraspinal muscles provide motion and dynamic [...]

• The pyridine derivative could tune the hydrated Zn2+ solvation structure and be adsorbed on the Zn anode surface to form a protective layer. • The addition of pyridine derivative could enable uniform Zn plating/stripping behavior and suppress side reactions. • 2,4-dihydroxypyridine in the series of pyridine derivatives endows Zn anode with ultra-long cycling performance for 5650 h. The poor reversibility and stability of Zn anodes greatly restrict the practical application of aqueous Zn-ion batteries (AZIBs), resulting from the uncontrollable dendrite growth and H 2 O-induced side reactions during cycling. Electrolyte additive modification is considered one of the most effective and simplest methods for solving the aforementioned problems. Herein, the pyridine derivatives (PD) including 2,4-dihydroxypyridine (2,4-DHP), 2,3-dihydroxypyridine (2,3-DHP), and 2-hydroxypyrdine (2-DHP), were employed as novel electrolyte additives in ZnSO 4 electrolyte. Both density functional theory calculation and experimental findings demonstrated that the incorporation of PD additives into the electrolyte effectively modulates the solvation structure of hydrated Zn ions, thereby suppressing side reactions in AZIBs. Additionally, the adsorption of PD molecules on the zinc anode surface contributed to uniform Zn deposition and dendrite growth inhibition. Consequently, a 2,4-DHP-modified Zn/Zn symmetrical cell achieved an extremely long cyclic stability up to 5650 h at 1 mA cm–2. Furthermore, the Zn/NH 4 V 4 O 10 full cell with 2,4-DHP-containing electrolyte exhibited an outstanding initial capacity of 204 mAh g–1, with a notable capacity retention of 79 % after 1000 cycles at 5 A g–1. Hence, this study expands the selection of electrolyte additives for AZIBs, and the working mechanism of PD additives provides new insights for electrolyte modification enabling highly reversible zinc anode. [Display omitted] [ABSTRACT FROM AUTHOR]

Generative Adversarial Networks have achieved significant advancements in generating and editing high-resolution images. However, most methods suffer from either requiring extensive labeled datasets or strong prior knowledge. It is also challenging for them to disentangle correlated attributes with few-shot data. In this paper, we propose FEditNet++, a GAN-based approach to explore latent semantics. It aims to enable attribute editing with limited labeled data and disentangle the correlated attributes. We propose a layer-wise feature contrastive objective, which takes into consideration content consistency and facilitates the invariance of the unrelated attributes before and after editing. Furthermore, we harness the knowledge from the pretrained discriminative model to prevent overfitting. In particular, to solve the entanglement problem between the correlated attributes from data and semantic latent correlation, we extend our model to jointly optimize multiple attributes and propose a novel decoupling loss and cross-assessment loss to disentangle them from both latent and image space. We further propose a novel-attribute disentanglement strategy to enable editing of novel attributes with unknown entanglements. Finally, we extend our model to accurately edit the fine-grained attributes. Qualitative and quantitative assessments demonstrate that our method outperforms state-of-the-art approaches across various datasets, including CelebA-HQ, RaFD, Danbooru2018 and LSUN Church.

Applying diffusion models to image-to-image translation (I2I) has recently received increasing attention due to its practical applications. Previous attempts inject information from the source image into each denoising step for an iterative refinement, thus resulting in a time-consuming implementation. We propose an efficient method that equips a diffusion model with a lightweight translator, dubbed a Diffusion Model Translator (DMT), to accomplish I2I. Specifically, we first offer theoretical justification that in employing the pioneering DDPM work for the I2I task, it is both feasible and sufficient to transfer the distribution from one domain to another only at some intermediate step. We further observe that the translation performance highly depends on the chosen timestep for domain transfer, and therefore propose a practical strategy to automatically select an appropriate timestep for a given task. We evaluate our approach on a range of I2I applications, including image stylization, image colorization, segmentation to image, and sketch to image, to validate its efficacy and general utility. The comparisons show that our DMT surpasses existing methods in both quality and efficiency. Code is available at https://github.com/THU-LYJ-Lab/dmt.

Few-Shot Instance Segmentation (FSIS) requires detecting and segmenting novel classes with limited support examples. Existing methods based on Region Proposal Networks (RPNs) face two issues: 1) overfitting suppresses novel class objects and 2) dual-branch models require complex spatial correlation strategies to prevent spatial information loss when generating class prototypes. We introduce a unified framework, Reference Twice (RefT), to exploit the relationship between support and query features for FSIS and related tasks. Our three main contributions are: 1) a novel transformer-based baseline that avoids overfitting, offering a new direction for FSIS; 2) demonstrating that support object queries encode key factors after base training, allowing query features to be enhanced twice at both feature and query levels using simple cross-attention, thus avoiding complex spatial correlation interaction; and 3) introducing a class-enhanced base knowledge distillation loss to address the issue of DETR-like models struggling with incremental settings due to the input projection layer, enabling easy extension to incremental FSIS. Extensive experimental evaluations on the COCO dataset under three FSIS settings demonstrate that our method performs favorably against existing approaches across different shots, e.g., +8.2/+9.4 performance gain over state-of-the-art methods with 10/30-shots.

Graphical Abstract:

When vibrations generated by marine machinery propagate through a ship’s hull into the ocean, they produce low-frequency radiated noise with distinct “acoustic fingerprint” characteristics. This noise, characterized by stable and concentrated energy, long transmission distances, and difficulty in elimination, becomes the primary target for enemy sonar detection. Active vibration isolation serves as a critical method for reducing low-frequency vibrations in ships and enhancing their acoustic stealth performance. However, control challenges persist, including multi-frequency excitation, frequency fluctuation, multi-channel coupling, and slow convergence speed. To address these issues, this paper introduced an innovative multi-channel decentralized decoupling filtered-x least mean square (DMFxLMS) algorithm. Firstly, a recursive least squares identification algorithm with a forgetting factor was proposed, taking into account the characteristics of single-input, multi-output and multi-input, and multi-output control systems, effectively enhancing the algorithm’s convergence speed and control accuracy. Secondly, based on the decentralized decoupling control concept, the multi-channel control system was simplified into parallel single-channel control loops. The control weight coefficient updates were only related to adjacent error signals, significantly reducing the algorithm’s computational complexity. Thirdly, an anti-impact link was designed to improve the algorithm’s robustness, considering the interference caused by other mechanical equipment during the control process. The influence of abnormal error signals in the control weight coefficient correction term was suppressed, and a percentage function was introduced to limit the output signal. Finally, the feasibility and effectiveness of the DMFxLMS algorithm were verified through simulations and experiments. The results demonstrated that the DMFxLMS algorithm achieved significant control effects for both constant frequency line spectrum excitation and frequency fluctuating line spectrum excitation, fulfilling the objective of reducing base vibration. The DMFxLMS algorithm exhibited fast convergence and excellent robustness, making it suitable for practical engineering applications.

Although precise regulation of the crystalline structures of metal oxides is an effective method to improve their antibacterial activities, the corresponding mechanisms involved in this process are still unclear. In this study, three kinds of cuprous oxide (Cu2O) samples with different structures of cubes, octahedra, and rhombic dodecahedra (c-Cu2O, o-Cu2O, and r-Cu2O) have been successfully synthesized and their antibacterial activities are compared. The antibacterial activities follow the order of r-Cu2O > o-Cu2O > c-Cu2O, revealing the significant dependence of the antibacterial activities on the crystalline structures of Cu2O. Quenching experiments, as well as the NBT and DPD experiments indicate that ≡CuII─OO•superoxo and ≡CuII─OOH peroxo, instead of •OH, O2•–, and H2O2, are the primary oxidizing species in the oxidative damage to E. coli. Raman analysis further confirms the presence of both ≡CuII─OO•superoxo and ≡CuII─OOH peroxo on the surface of r-Cu2O. On the other hand, the NCP experiment reveals that Cu+, instead of Cu2+, also contributes to the antibacterial process. This study provides new insight into the antibacterial mechanisms of Cu2O.

An Al2O3/LaAlO3/SiO2 gate stack is designed and implemented on various trench-gate channel planes for high-performance trench power MOSFETs. The designed high-k gate stack achieves a significant enhancement in the gate blocking capability (~1.73X) and maintains a low interface state density (Dit), in comparison to the SiO2­ gate. Moreover, owing to the implementation of the high-k gate stack, the high-k trench gate MOSFETs conduct a drain current of approximately 1.3 times larger than that of the SiO2 trench gate MOSFETs on all 24 channel planes at the same overdrive voltage (Vgs-Vth) of 10 V. An analysis of the mobility limiting mechanisms on different channel planes reveals that the highest channel mobility on the (1120) channel plane is primarily due to its largest Coulomb scattering mobility and surface roughness scattering mobility.

Selective oxidation of cyclohexene (Cy) to cyclohexene oxide (Cy-ep) by O2under solvent- and initiator-free conditions is a challenge in chemical processes, which is still constrained by its low selectivity (theoretical selectivity of 50.0%). Given the low epoxide selectivity (<20.0%) of single metal catalysts, we propose a relay catalysis strategy on the oxidation of Cy to Cy-ep with high selectivity by the spherical Mg–Mo bimetallic oxide (MMO). Such a relay catalyst exhibits a desirable Cy conversion of 31.0% and a high selectivity of 49.3% toward Cy-ep. Meanwhile, the high stability and reusability of MMO also indicate its potential to be used for industrial scale-ups. A series of characterizations and DFT calculations reveal that the negatively charged lattice oxygen in [MgII–O–II] adsorbs allylic H of Cy to weaken the bond energy of the allylic C–H bond and form Cy•, and then Cy• rapidly traps O2to give CyOO•, which subsequently abstracts H from allylic C–H of another Cy to form 2-cyclohexene-1-hydroperoxide (Cy-OOH) and Cy•, achieving the propagation cycle. On the other hand, the Mo(VI) site in MMO catalyzes the produced Cy-OOH peroxo, which then oxidizes Cy to Cy-ep and 2-cyclohexene-1-ol. This work provides insight into the relay catalyst for the epoxidation of olefin with O2with high selectivity under solvent- and initiator-free conditions.