Unsupervised domain adaptation (UDA) transfers knowledge from a label-rich source domain to a fullyunlabeled target domain. To tackle this task, recent approaches resort to discriminative domain transfer in virtue of pseudo-labels to enforce the class-level distribution alignment across the source and target domains. These methods, however, are vulnerable to the error accumulation and thus incapable of preserving cross-domain category consistency, as the pseudo-labeling accuracy is not guaranteed explicitly. In this paper, we propose the Progressive Feature Alignment Network (PFAN) to align the discriminative features across domains progressively and effectively, via exploiting the intra-class variation in the target domain. To be specific, we first develop an Easyto-Hard Transfer Strategy (EHTS) and an Adaptive Prototype Alignment (APA) step to train our model iteratively and alternatively. Moreover, upon observing that a good domain adaptation usually requires a non-saturated source classifier, we consider a simple yet efficient way to retard the convergence speed of the source classification loss by further involving a temperature variate into the soft-max function. The extensive experimental results reveal that the proposed PFAN exceeds the state-of-the-art performance on three UDA datasets. * indicates equal contributions.† Corresponding authors
We study the angular scattering properties of individual core-shell nanoparticles that support simultaneously both electric and optically-induced magnetic resonances of different orders. In contrast to the approach to suppress the backward scattering and enhance the forward scattering relying on overlapping electric and magnetic dipoles, we reveal that the directionality of the forward scattering can be further improved through the interferences of higher order electric and magnetic modes. Since the major contributing electric and magnetic responses can be tuned to close magnitudes, ultra-directional forward scattering can be achieved by single nanoparticles without compromising the feature of backward scattering suppression, which may offer new opportunities for nanoantennas, photovoltaic devices, bio-sensing and many other interdisciplinary researches.
To enable next‐generation high‐power, high‐energy‐density lithium (Li) metal batteries (LMBs), an electrolyte possessing both high Li Coulombic efficiency (CE) at a high rate and good anodic stability on cathodes is critical. Acetonitrile (AN) is a well‐known organic solvent for high anodic stability and high ionic conductivity, yet its application in LMBs is limited due to its poor compatibility with Li metal anodes even at high salt concentration conditions. Here, a highly concentrated AN‐based electrolyte is developed with a vinylene carbonate (VC) additive to suppress Li+ depletion at high current densities. Addition of VC to the AN‐based electrolyte leads to the formation of a polycarbonate‐based solid electrolyte interphase, which minimizes Li corrosion and leads to a very high Li CE of up to 99.2% at a current density of 0.2 mA cm‐2. Using such an electrolyte, fast charging of Li||NMC333 cells is realized at a high current density of 3.6 mA cm‐2, and stable cycling of Li||NMC622 cells with a high cathode loading of 4 mAh cm‐2 is also demonstrated.
Tillandsia species with degenerated roots have evolved into hygroscopic leaves that absorb moisture from air.T his interesting biological adaptability has inspired us to develop an integrated hygroscopic photothermal organogel (POG) to achieve asolar-powered atmospheric water harvesting (AWH). The well-designed hydrophilic co-polymeric skeleton is fabricated to accommodate hygroscopic glycerin medium, which enables the POG self-contained property, mechanically flexibility and synergistic enhancement of moisture sorption. The integration of interpenetrated photothermal component of poly-pyrrole-dopamine (P-Py-DA) can endow the POG an efficient solar-to-thermal property for controllable solar-driven interfacial water releasing.T he integrated POG has an equilibrium moisture sorption of 16.01 kg m À2 at the RH of 90 %, and daily water production as high as 2.43 kg m À2 day À1 is achieved in actual outdoor experiments.
Cephalopods can display variable body color/patterns upon environmental stimulation via bioelectricity‐controlled muscle contraction/expansion of skin chromatophores. However, it remains challenging to produce artificial display analogs that exhibit reversible and rapid switching between multiple expected luminescent patterns, although such systems are very appealing for many practical uses (e.g., data encryption). Inspired by the bioelectromechanical display tactic of cephalopods, in this work, a conceptually new photomechanically modulated fluorescent system that enables on‐demand display of fluorescent patterns via a cascading stimulation−mechanical movement−optical output conduction mechanism is presented. Specifically, this biomimetic system comprises a customizable hollow display panel and a bottom‐tethered photothermally responsive fluorescent actuator. Under NIR light, the photomechanically bending movements of the fluorescent actuator will immediately cover the hollow window of the display panel and synchronously manifest as the display of fluorescent patterns. Owing to its desirable time‐ and light‐power‐dependent actuating behaviors, diverse fluorescent patterns/information can be dynamically and reversibly displayed by facilely controlling this single remote NIR signal. This bioinspired strategy is universal and promising for fabricating on‐demand fluorescent display platforms that combine a wide choice of fluorophores, remote control with high spatial/temporal precision, and especially single‐input multiple‐output features.
In this paper, a tunable broadband metamaterial absorber (MA) based on graphene is investigated theoretically and numerically at mid-infrared regions. Compared with the previously reported multiband graphene-based MAs, a broad bandwidth of 11.7 THz with the absorption over 90% is obtained in the proposed MA, which is composed of a Jerusalem cross (JC) metal encrusting into the slot graphene layer in the top layer. The results show that the origin of broadband absorption is caused by coupling effect between metal and graphene, and this effect is explained by the two-mode waveguide coupling theory. The tunability of MA is achieved via changing the external gate voltage to modify the Fermi energy of graphene. Further results show that the proposed MA can be used as the permittivity sensor with a high absorption. This work indicates that the proposed MA has the potential applications with respect to sensors and infrared absorbers.
Bessel beam is the important member of the family of non-diffracting beams and has many novel properties which can be used in many areas. However, the source of Bessel beam generated by the existing methods can be used only in a short distance due to its low power. In this paper, based on the coherent combining technology, we have proposed a method which can be used to generate a high-power Bessel beam. Even more, we give an innovative idea to form vortex phase by using discontinuous piston phase. To confirm the validity of this method, the intensity evolution of the combined beam and the Bessel-Gaussian beam at different propagation distance have been studied and compared. Meanwhile, the experimental realization has been discussed from the existing experimental result related to the coherent combining technology.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.