Search

Recognizing 3D point cloud plays a pivotal role in many real-world applications. However, deploying 3D point cloud deep learning model is vulnerable to adversarial attacks. Despite many efforts into developing robust model by adversarial training, they may become less effective against emerging attacks. This limitation motivates the development of adversarial purification which employs generative model to mitigate the impact of adversarial attacks. In this work, we highlight the remaining challenges from two perspectives. First, the purification based method requires retraining the classifier on purified samples which introduces additional computation overhead. Moreover, in a more realistic scenario, testing samples arrives in a streaming fashion and adversarial samples are not isolated from clean samples. These challenges motivates us to explore dynamically update model upon observing testing samples. We proposed a test-time purified self-training strategy to achieve this objective. Adaptive thresholding and feature distribution alignment are introduced to improve the robustness of self-training. Extensive results on different adversarial attacks suggest the proposed method is complementary to purification based method in handling continually changing adversarial attacks on the testing data stream.

Segment Anything Model (SAM) is an advanced foundational model for image segmentation, widely applied to remote sensing images (RSIs). Due to the domain gap between RSIs and natural images, traditional methods typically use SAM as a source pre-trained model and fine-tune it with fully supervised masks. Unlike these methods, our work focuses on fine-tuning SAM using more convenient and challenging point annotations. Leveraging SAM's zero-shot capabilities, we adopt a self-training framework that iteratively generates pseudo-labels for training. However, if the pseudo-labels contain noisy labels, there is a risk of error accumulation. To address this issue, we extract target prototypes from the target dataset and use the Hungarian algorithm to match them with prediction prototypes, preventing the model from learning in the wrong direction. Additionally, due to the complex backgrounds and dense distribution of objects in RSI, using point prompts may result in multiple objects being recognized as one. To solve this problem, we propose a negative prompt calibration method based on the non-overlapping nature of instance masks. In brief, we use the prompts of overlapping masks as corresponding negative signals, resulting in refined masks. Combining the above methods, we propose a novel Pointly-supervised Segment Anything Model named PointSAM. We conduct experiments on RSI datasets, including WHU, HRSID, and NWPU VHR-10, and the results show that our method significantly outperforms direct testing with SAM, SAM2, and other comparison methods. Furthermore, we introduce PointSAM as a point-to-box converter and achieve encouraging results, suggesting that this method can be extended to other point-supervised tasks. The code is available at https://github.com/Lans1ng/PointSAM., Comment: 15 pages

Wash-in leptogenesis is an attractive mechanism to produce the baryon asymmetry of the Universe. It treats right-handed-neutrino interactions as spectator processes, on the same footing as electroweak sphalerons, that reprocess primordial charge asymmetries in the thermal plasma into a baryon-minus-lepton asymmetry. The origin of these primordial charges must be accounted for by new $CP$-violating dynamics at very high energies. In this paper, we propose such a scenario of chargegenesis that, unlike earlier proposals, primarily relies on new interactions in the gravitational sector. We point out that a coupling of a conserved current to the divergence of the Ricci scalar during reheating can lead to nonzero effective chemical potentials in the plasma that, together with a suitable charge-violating interaction, can result in the production of a primordial charge asymmetry. Gravitational chargegenesis represents a substantial generalization of the idea of gravitational baryogenesis. We provide a detailed analysis of a generic and minimal realization that is consistent with inflation and show that it can successfully explain the baryon asymmetry of the Universe., Comment: 5 pages + references and appendix, 3 figures

Primordial Black Holes (PBHs) are capable of emitting extremely energetic particles independent of their interactions with the Standard Model. In this work, we investigate a particularly interesting scenario in which PBHs evaporating in the early universe may be responsible for some of the observed high-energy neutrinos above the TeV or PeV scale in the present universe. We compute the energy spectrum of neutrinos directly emitted by PBHs with a monochromatic mass function and estimate the wash-out point, which determines the maximum energy of the spectrum. We find that the spectrum generally extends to high energies following a power law of $E_{\nu}^{-3}$ until it reaches the wash-out point, which crucially depends on the PBH mass. For PBHs of $10^{13}$ grams, the spectrum can extend up to the PeV scale, though the flux is too low for detection. We also consider an indirect production mechanism involving dark particles that are emitted by PBHs and decay into neutrinos at a much later epoch. This mechanism allows lighter (such as those in the gram to kilogram range) PBHs to produce more energetic neutrino fluxes without being washed out by the thermal plasma in the early universe. In this scenario, we find that ultra-high-energy neutrinos around or above the EeV scale can be generated, with sufficiently high fluxes detectable by current and future high-energy neutrino observatories such as IceCube and GRAND., Comment: 23 pages, 4 figures

We analyze the single-photon band structure and the transport of a single photon in a one-dimensional coupled-spinning-resonator chain. The time-reversal symmetry of the resonators chain is broken by the spinning of the resonators, instead of external or synthetic magnetic field. Two nonreciprocal single-photon band gaps can be obtained in the coupled-spinning-resonator chain, whose width depends on the angular velocity of the spinning resonator. Based on the nonreciprocal band gaps, we can implement a single photon circulator at multiple frequency windows, and the direction of photon cycling is opposite for different band gaps. In addition, reciprocal single-photon band structures can also be realized in the coupled-spinning-resonator chain when all resonators rotate in the same direction with equal angular velocity. Our work open a new route to achieve, manipulate, and switch nonreciprocal or reciprocal single-photon band structures, and provides new opportunities to realize novel single-photon devices., Comment: 8 pages, 3 figures

Nonreciprocal photon blockade is of particular interest due to its potential applications in chiral quantum technologies and topological photonics. In the regular cases, nonreciprocal transmission (classical nonreciprocity) and nonreciprocal photon blockade (quantum nonreciprocity) often appear simultaneously. Nevertheless, how to achieve purely quantum nonreciprocity (no classical nonreciprocity) remains largely unexplored. Here, we propose a spatially separated transmission scheme, that the photons transport in different directions take different paths, in an optical system consisting of two spinning cavities coupled indirectly by two common drop-filter waveguides. Based on the spatially separated transmission scheme, we demonstrate a purely quantum nonreciprocity (nonreciprocal photon blockade) by considering the Kerr nonlinear interaction in one of the paths. Interestingly, we find that the nonreciprocal photon blockade is enhanced nonreciprocally, i.e., the nonreciprocal photon blockade is enhanced when the photons transport in one direction but suppressed in the reverse direction. We identify that the nonreciprocal enhancement of nonreciprocal photon blockade is induced by the destructive or constructive interference between two paths for two photons passing through the whole system. The spatially separated transmission scheme proposed in the work provides a novel approach to observe purely quantum nonreciprocal effects., Comment: 10 pages, 5 figures

We propose how to achieve chiral photon blockade by spinning a nonlinear optical resonator. We show that by driving such a device at a fixed direction, completely different quantum effects can emerge for the counter-propagating optical modes, due to the spinning-induced breaking of time-reversal symmetry, which otherwise is unattainable for the same device in the static regime. Also, we find that in comparison with the static case, robust non-classical correlations against random backscattering losses can be achieved for such a quantum chiral system. Our work, extending previous works on the spontaneous breaking of optical chiral symmetry from the classical to purely quantum regimes, can stimulate more efforts towards making and utilizing various chiral quantum effects, including applications for chiral quantum networks or noise-tolerant quantum sensors.

Existing test-time adaptation (TTA) approaches often adapt models with the unlabeled testing data stream. A recent attempt relaxed the assumption by introducing limited human annotation, referred to as Human-In-the-Loop Test-Time Adaptation (HILTTA) in this study. The focus of existing HILTTA studies lies in selecting the most informative samples to label, a.k.a. active learning. In this work, we are motivated by a pitfall of TTA, i.e. sensitivity to hyper-parameters, and propose to approach HILTTA by synergizing active learning and model selection. Specifically, we first select samples for human annotation (active learning) and then use the labeled data to select optimal hyper-parameters (model selection). To prevent the model selection process from overfitting to local distributions, multiple regularization techniques are employed to complement the validation objective. A sample selection strategy is further tailored by considering the balance between active learning and model selection purposes. We demonstrate on 5 TTA datasets that the proposed HILTTA approach is compatible with off-the-shelf TTA methods and such combinations substantially outperform the state-of-the-art HILTTA methods. Importantly, our proposed method can always prevent choosing the worst hyper-parameters on all off-the-shelf TTA methods. The source code will be released upon publication.

The neutrino force results from the exchange of a pair of neutrinos. A neutrino background can significantly influence this force. In this work, we present a comprehensive calculation of the neutrino force in various neutrino backgrounds with spin dependence taken into account. In particular, we calculate the spin-independent and spin-dependent parity-conserving neutrino forces, in addition to the spin-dependent parity-violating neutrino forces with and without the presence of a neutrino background for both isotropic and anisotropic backgrounds. Compared with the vacuum case, the neutrino background can effectively violate Lorentz invariance and lead to additional parity-violating terms that are not suppressed by the velocity of external particles. We estimate the magnitude of the effect of atomic parity-violation experiments, and it turns out to be well below the current experimental sensitivity., Comment: 34 pages, 3 figures, 2 tables

In recent years, the rapid advancement of deepfake technology has revolutionized content creation, lowering forgery costs while elevating quality. However, this progress brings forth pressing concerns such as infringements on individual rights, national security threats, and risks to public safety. To counter these challenges, various detection methodologies have emerged, with Vision Transformer (ViT)-based approaches showcasing superior performance in generality and efficiency. This survey presents a timely overview of ViT-based deepfake detection models, categorized into standalone, sequential, and parallel architectures. Furthermore, it succinctly delineates the structure and characteristics of each model. By analyzing existing research and addressing future directions, this survey aims to equip researchers with a nuanced understanding of ViT's pivotal role in deepfake detection, serving as a valuable reference for both academic and practical pursuits in this domain.

We propose to manipulate topological polaritons in optomechanical ladders consisting of an optical Su-Schrieffer-Heeger (SSH) chain and a mechanical SSH chain connected through optomechanical (interchain) interactions. We show that the topological phase diagrams are divided into six areas by four boundaries and that there are four topological phases characterized by the Berry phases. We find that a topologically nontrivial phase of the polaritons is generated by the optomechanical interaction between the optical and mechanical SSH chains even though they are both in the topologically trivial phases. Counter-intuitively, six edge states appear in one of the topological phases with only two topological nontrivial bands, and some edge states are localized near but not at the boundaries of an open-boundary ladder. Moreover, a two-dimensional Chern insulator with higher Chern numbers is simulated by introducing proper periodical adiabatic modulations of the driving amplitude and frequency. Our work not only opens a route towards topological polaritons manipulation by optomachanical interactions, but also will exert a far-reaching influence on designing topologically protected polaritonic devices., Comment: 8+15 pages, 3+7 figures

Photon blockade in weak nonlinear regime is an exciting and promising subject that has been extensively studied in the steady state. However, how to achieve dynamic blockade in a single bosonic mode with weak nonlinearity using only pulsed driving field remains unexplored. Here, we propose to optimize the parameters of the pulsed driving field to achieve dynamic blockade in a single bosonic mode with weak nonlinearity via the particle swarm optimization (PSO) algorithm. We demonstrate that both Gaussian and rectangular pulses can be used to generate dynamic photon blockade in a single bosonic mode with weak nonlinearity. Based on the Fourier series expansions of the pulsed driving field, we identify that there are many paths for two-photon excitation in the bosonic mode, even only driven by pulsed field, and the dynamic blockade in weak nonlinear regime is induced by the destructive interference between them. Our work not only highlights the effectiveness of PSO algorithm in optimizing dynamical blockade, but also opens a way to optimize the parameters for other quantum effects, such as quantum entanglement and quantum squeezing., Comment: 8 pages, 6 figures

With a high-loss resonator supplying the non-Hermiticity, we study the Energy level degeneracy and quantum state evolution in tunable coupling superconducting quantum circuit. The qubit's effective energy level and damping rate can be continually tuned in superconducting circuit, and the positions and numbers of level degenerate points are controllable. The efficient of quantum state exchange and the asymmetry of quantum state evolution can be tuned with non-hermitian and nonreciprocal coupling between two qubits. The controllable non-Hermiticity provides new insights and methods for exploring the unconventional quantum effects in superconducting quantum circuit., Comment: 7 pages, 4 figures

Photon blockade enhancement is an exciting and promising subject that has been well studied for photons in cavities. However, whether photon blockade can be enhanced in the output fields remains largely unexplored. We show that photon blockade can be greatly enhanced in the mixing output field of a nonlinear cavity and an auxiliary (linear) cavity, where no direct coupling between the nonlinear and auxiliary cavities is needed. We uncover a biquadratic scaling relation between the second-order correlation of the photons in the output field and intracavity nonlinear interaction strength, in contrast to a quadratic scaling relation for the photons in a nonlinear cavity. We identify that this scaling enhancement of photon blockade in the output field is induced by the destructive interference between two of the paths for two photons passing through the two cavities. We then extend the theory to the experimentally feasible Jaynes-Cummings model consisting of a two-level system strongly coupled to one of the two uncoupled cavities, and also predict a biquadratic scaling law in the mixing output field. Our proposed scheme is universal and can be extended to enhance blockade in other bosonic systems., Comment: 7 pages, 4 figures

Light bosons can mediate long-range forces. We show that light bosonic mediators interacting with a background medium, in particular, with the cosmic neutrino background (C$\nu$B), may induce medium-dependent masses which could effectively screen long-range forces from detection. This leads to profound implications for long-range force searches in e.g. the E\"ot-Wash, MICROSCOPE, and lunar laser-ranging (LLR) experiments. For instance, we find that when the coupling of the mediator to neutrinos is above $3\times10^{-10}$ or $5\times10^{-13}$, bounds from LLR and experiments employing the Sun as an attractor, respectively, would be entirely eliminated. Larger values of the coupling can also substantially alleviate bounds from searches conducted at shorter distances., Comment: 18 pages, 2 figures; version to appear in JHEP

Existing approaches towards anomaly detection~(AD) often rely on a substantial amount of anomaly-free data to train representation and density models. However, large anomaly-free datasets may not always be available before the inference stage; in which case an anomaly detection model must be trained with only a handful of normal samples, a.k.a. few-shot anomaly detection (FSAD). In this paper, we propose a novel methodology to address the challenge of FSAD which incorporates two important techniques. Firstly, we employ a model pre-trained on a large source dataset to initialize model weights. Secondly, to ameliorate the covariate shift between source and target domains, we adopt contrastive training to fine-tune on the few-shot target domain data. To learn suitable representations for the downstream AD task, we additionally incorporate cross-instance positive pairs to encourage a tight cluster of the normal samples, and negative pairs for better separation between normal and synthesized negative samples. We evaluate few-shot anomaly detection on on 3 controlled AD tasks and 4 real-world AD tasks to demonstrate the effectiveness of the proposed method., Comment: IEEE Transactions on Image Processing

The Primakoff process plays a crucial role in axion production in astrophysical environments and laboratories. Given the rising interest in axion physics and many on-going experimental activities, we conduct a comprehensive calculation of this process and carefully examine several aspects that have been neglected in the literature. In particular, our calculation is valid for axions with significantly large masses, which would be of importance to axion searches utilizing crystal and liquid xenon detectors. We present the most updated calculation of the Primakoff solar axion flux, with a simple parametrization that is applicable to a broad range of axion masses up to a few tens of keV. Our code is publicly available at GitHub., Comment: 26 pages, 8 figures; version accepted by JCAP; new figure added to compare various corrections; numerical code available at https://github.com/Fenyutanchan/Solar-Axion-Primakoff-Flux.git

We investigate a square-lattice architecture of superconducting transmon qubits with inter-qubit interactions mediated by inductive couplers. Therein, the inductive couling between the qubit and couplers is suggested to be designed into the gradiometer form to intigimate the flux noise orginating from the environment. Via periodically modulating the couplers,the Abelian gauge potential, termed effective magnetic flux, can be synthesized artificially, making the system an excellent platform for simulating two-dimensional topological physics. In the simplest two-dimensional model, the double (or three-leg) ladder, the staggered vortex-Meissner phase transition different from that in the two-leg ladder can be found in the single-particle ground state as the effective magnetic flux varies. Besides, the large coupling ratio between the interleg and intraleg coupling strengths also makes the chiral current resemble squeezed sinusoidal functions. If the row number is further increased, the topological band structure anticipated at massive rows begins to occur even for a relatively small number of rows (ten or so for the considered parameters). This heralds a small circuit scale to observe the topological band. The edge state in the band gap is determined by the topological Chern number and can be calculated through integrating the Berry curvature with respect to the first Brillouin zone. Besides, we present a systematic method on how to measure the topological band structure based on time- and space-domain Frourier transformation of the wave function after properly excited. The result offers an avenue for simulating two-dimensional topological physics on the state-of-the-art superconducting quantum chips.

Domain adaptation is crucial in aerial imagery, as the visual representation of these images can significantly vary based on factors such as geographic location, time, and weather conditions. Additionally, high-resolution aerial images often require substantial storage space and may not be readily accessible to the public. To address these challenges, we propose a novel Source-Free Object Detection (SFOD) method. Specifically, our approach begins with a self-training framework, which significantly enhances the performance of baseline methods. To alleviate the noisy labels in self-training, we utilize Contrastive Language-Image Pre-training (CLIP) to guide the generation of pseudo-labels, termed CLIP-guided Aggregation (CGA). By leveraging CLIP's zero-shot classification capability, we aggregate its scores with the original predicted bounding boxes, enabling us to obtain refined scores for the pseudo-labels. To validate the effectiveness of our method, we constructed two new datasets from different domains based on the DIOR dataset, named DIOR-C and DIOR-Cloudy. Experimental results demonstrate that our method outperforms other comparative algorithms. The code is available at https://github.com/Lans1ng/SFOD-RS., Comment: Accepted by IGARSS2024

Neutrinos are often considered as a portal to new physics beyond the Standard Model (SM) and might possess phenomenologically interesting interactions with dark matter (DM). This paper examines the cosmological imprints of DM that interacts with and is produced from SM neutrinos at temperatures below the MeV scale. We take a model-independent approach to compute the evolution of DM in this framework and present analytic results which agree well with numerical ones. Both freeze-in and freeze-out regimes are included in our analysis. Furthermore, we demonstrate that the thermal evolution of neutrinos might be substantially affected by their interaction with DM. We highlight two distinctive imprints of such DM on neutrinos: (i) a large, negative contribution to $N_{\rm eff}$, which is close to the current experimental limits and will readily be probed by future experiments; (ii) spectral distortion of the cosmic neutrino background (C$\nu$B) due to DM annihilating into neutrinos, a potentially important effect for the ongoing experimental efforts to detect C$\nu$B., Comment: 24 pages, 7 figures

We propose and prove two theorems for determining the number of dark modes in linear two-component quantum networks composed of two types of bosonic modes. This is achieved by diagonalizing the two sub-networks of the same type of modes, mapping the networks to either a standard or a thick arrowhead matrix, and analyzing the linear dependence and independence between the column vectors associated with degenerate normal modes in the coupling matrix. We confirm the two theorems by checking the simultaneous ground-state cooling of the mechanical modes in linearized optomechanical networks. These results also work for linear fermionic networks and other networks described by quadratic coupled-mode Hamiltonian. The present method can be extended to study the dark-state effect in driven atom systems and to construct large decoherence-free subspaces for processing quantum information. This work will initiate the studies on dynamical, transport, and statistical properties of linear networks with decoupled subspaces., Comment: 43 pages, 17 figures

The success of large language models has inspired the computer vision community to explore image segmentation foundation model that is able to zero/few-shot generalize through prompt engineering. Segment-Anything(SAM), among others, is the state-of-the-art image segmentation foundation model demonstrating strong zero/few-shot generalization. Despite the success, recent studies reveal the weakness of SAM under strong distribution shift. In particular, SAM performs awkwardly on corrupted natural images, camouflaged images, medical images, etc. Motivated by the observations, we aim to develop a self-training based strategy to adapt SAM to target distribution. Given the unique challenges of large source dataset, high computation cost and incorrect pseudo label, we propose a weakly supervised self-training architecture with anchor regularization and low-rank finetuning to improve the robustness and computation efficiency of adaptation. We validate the effectiveness on 5 types of downstream segmentation tasks including natural clean/corrupted images, medical images, camouflaged images and robotic images. Our proposed method is task-agnostic in nature and outperforms pre-trained SAM and state-of-the-art domain adaptation methods on almost all downstream tasks with the same testing prompt inputs., Comment: 20 pages, 12 figures

Wash-in leptogenesis is a powerful mechanism to generate the baryon asymmetry of the Universe that treats right-handed-neutrino interactions on the same footing as electroweak sphaleron processes: as mere spectator processes acting on the background of chemical potentials in the Standard Model plasma. Successful wash-in leptogenesis requires this chemical background to be CP-violating, which can be achieved by violating any of the more than ten global charges that are conserved in the Standard Model at very high temperatures. In this paper, we demonstrate that the primordial charge asymmetries required for wash-in leptogenesis can be readily produced by evaporating primordial black holes (PBHs). Our argument is based on the fact that the Hawking radiation emitted by PBHs contains more or less any state in the particle spectrum. Therefore, if heavy states with CP-violating decays are present in the ultraviolet, PBH evaporation will unavoidably lead to the production of these states. We illustrate this scenario by means of a simple toy model where PBH evaporation leads to the production of heavy particles that we call asymmetrons and whose decay results in a primordial charge asymmetry for right-handed electrons, which in turn sets the initial conditions for wash-in leptogenesis. We focus on the parameter region where the decay of the initial thermal asymmetron abundance occurs long before PBH evaporation and only results in a negligible primordial charge asymmetry. PBH evaporation at later times then serves as a mechanism to resurrect the asymmetron abundance and ensure the successful generation of the baryon asymmetry after all. We conclude that PBHs can act as asymmetry-producing machines that grant access to whatever CP-violating physics may be present in the ultraviolet, rekindling it at lower energies where it can be reprocessed into a baryon asymmetry by right-handed neutrinos., Comment: 11 pages, 4 figures, analyses on two-to-two scattering added, matches the PLB verion

Right-handed neutrinos ($\nu_{R}$) offer an intriguing portal to new physics in hidden sectors where dark matter (DM) may reside. In this work, we delve into the simplest hidden sector involving only a real scalar exclusively coupled to $\nu_{R}$, referred to as the $\nu_{R}$-philic scalar. We investigate the viability of the $\nu_{R}$-philic scalar to serve as a DM candidate, under the constraint that the coupling of $\nu_{R}$ to the standard model is determined by the seesaw relation and is responsible for the observed DM abundance. By analyzing the DM decay channels and solving Boltzmann equations, we identify the viable parameter space. In particular, our study reveals a lower bound ($2.6\times10^{5}$ GeV) on the mass of $\nu_{R}$ for the $\nu_{R}$-philic scalar to be DM. The DM mass may vary from sub-MeV to sub-GeV. Within the viable parameter space, monochromatic neutrino lines from DM decay can be an important signal for DM indirect detection., Comment: 24 pages, 5 figures, Lyman-alpha bounds added, matches the JCAP published version

Annotating remote sensing images (RSIs) presents a notable challenge due to its labor-intensive nature. Semi-supervised object detection (SSOD) methods tackle this issue by generating pseudo-labels for the unlabeled data, assuming that all classes found in the unlabeled dataset are also represented in the labeled data. However, real-world situations introduce the possibility of out-of-distribution (OOD) samples being mixed with in-distribution (ID) samples within the unlabeled dataset. In this paper, we delve into techniques for conducting SSOD directly on uncurated unlabeled data, which is termed Open-Set Semi-Supervised Object Detection (OSSOD). Our approach commences by employing labeled in-distribution data to dynamically construct a class-wise feature bank (CFB) that captures features specific to each class. Subsequently, we compare the features of predicted object bounding boxes with the corresponding entries in the CFB to calculate OOD scores. We design an adaptive threshold based on the statistical properties of the CFB, allowing us to filter out OOD samples effectively. The effectiveness of our proposed method is substantiated through extensive experiments on two widely used remote sensing object detection datasets: DIOR and DOTA. These experiments showcase the superior performance and efficacy of our approach for OSSOD on RSIs.

Test-Time Adaptation aims to adapt source domain model to testing data at inference stage with success demonstrated in adapting to unseen corruptions. However, these attempts may fail under more challenging real-world scenarios. Existing works mainly consider real-world test-time adaptation under non-i.i.d. data stream and continual domain shift. In this work, we first complement the existing real-world TTA protocol with a globally class imbalanced testing set. We demonstrate that combining all settings together poses new challenges to existing methods. We argue the failure of state-of-the-art methods is first caused by indiscriminately adapting normalization layers to imbalanced testing data. To remedy this shortcoming, we propose a balanced batchnorm layer to swap out the regular batchnorm at inference stage. The new batchnorm layer is capable of adapting without biasing towards majority classes. We are further inspired by the success of self-training~(ST) in learning from unlabeled data and adapt ST for test-time adaptation. However, ST alone is prone to over adaption which is responsible for the poor performance under continual domain shift. Hence, we propose to improve self-training under continual domain shift by regularizing model updates with an anchored loss. The final TTA model, termed as TRIBE, is built upon a tri-net architecture with balanced batchnorm layers. We evaluate TRIBE on four datasets representing real-world TTA settings. TRIBE consistently achieves the state-of-the-art performance across multiple evaluation protocols. The code is available at \url{https://github.com/Gorilla-Lab-SCUT/TRIBE}., Comment: 23 pages, 7 figures and 22 tables

As a promising approach for optical nonreciprocity without magnetic materials, optomechanically induced nonreciprocity has great potential for all-optical controllable isolators and circulators on chips. However, as a very important issue in practical applications, the bandwidth for nonreciprocal transmission with high isolation has not been fully investigated yet. In this study we review the nonreciprocity in a Brillouin optomechanical system with single cavity and point out the challenge in achieving broad bandwidth with high isolation. To overcome this challenge, we propose a one dimensional optomechanical array to realize broadband optical nonreciprocity via nonreciprocal band structure. We exploit nonreciprocal band structure by the stimulated Brillouin scattering induced transparency with directional optical pumping, and show that it is possible to demonstrate optical nonreciprocity with both broad bandwidth and high isolation. Such Brillouin optomechanical lattices with nonreciprocal band structure, offer an avenue to explore nonreciprocal collective effects in different electromagnetic and mechanical frequency regimes, such as nonreciprocal topological photonic and phononic phases., Comment: 10 pages, 6 figures

Solar antineutrinos are absent in the standard solar model prediction. Consequently, solar antineutrino searches emerge as a powerful tool to probe new physics capable of converting neutrinos into antineutrinos. In this study, we highlight that neutrino self-interactions, recently gaining considerable attention due to their cosmological and astrophysical implications, can lead to significant solar antineutrino production. We systematically explore various types of four-fermion effective operators and light scalar mediators for neutrino self-interactions. By estimating the energy spectra and event rates of solar antineutrinos at prospective neutrino detectors such as JUNO, Hyper-Kamiokande, and THEIA, we reveal that solar antineutrino searches can impose stringent constraints on neutrino self-interactions and probe the parameter space favored by the Hubble tension., Comment: 21 pages, 5 figures, the revised version accepted by JCAP

Optomechanical systems with parametric coupling between optical (photon) and mechanical (phonon) modes provide a useful platform to realize various magnetic-free nonreciprocal devices, such as isolators, circulators, and directional amplifiers. However, nonreciprocal router with multiaccess channels has not been extensively studied yet. Here, we propose a nonreciprocal router with one transmitter, one receiver, and two output terminals, based on an optomechanical plaquette composing of two optical modes and two mechanical modes. The time-reversal symmetry of the system is broken via synthetic magnetism induced by driving the two optical modes with phase-correlated laser fields. The prerequisites for nonreciprocal routing are obtained both analytically and numerically, and the robustness of the nonreciprocity is demonstrated numerically. Multi-terminal nonreciprocal router in optomechanical plaquette provides a useful quantum node for development of quantum network information security and realization of quantum secure communication., Comment: 12 pages, 5 figures

Generalizing deep learning models to unknown target domain distribution with low latency has motivated research into test-time training/adaptation (TTT/TTA). Existing approaches often focus on improving test-time training performance under well-curated target domain data. As figured out in this work, many state-of-the-art methods fail to maintain the performance when the target domain is contaminated with strong out-of-distribution (OOD) data, a.k.a. open-world test-time training (OWTTT). The failure is mainly due to the inability to distinguish strong OOD samples from regular weak OOD samples. To improve the robustness of OWTTT we first develop an adaptive strong OOD pruning which improves the efficacy of the self-training TTT method. We further propose a way to dynamically expand the prototypes to represent strong OOD samples for an improved weak/strong OOD data separation. Finally, we regularize self-training with distribution alignment and the combination yields the state-of-the-art performance on 5 OWTTT benchmarks. The code is available at https://github.com/Yushu-Li/OWTTT., Comment: Accepted at ICCV 2023 (Oral)

We investigate the impact of new light particles, carrying significant energy in the early universe after neutrino decoupling, on the cosmological effective relativistic neutrino species, $N_{{\rm eff}}$. If the light particles are produced from decoupled neutrinos, $N_{{\rm eff}}$ is predominantly modified through the dilution-resistant effect. This effect arises because the energy stored in the mass of new particles is less diluted than the photon and neutrino energy as the universe expands. Our study comprehensively explores this effect, deriving $N_{{\rm eff}}$ constraints on the couplings of light mediators with neutrinos, encompassing both scalar and vector mediators. We find that the dilution-resistant effect can increase $N_{{\rm eff}}$ by 0.118 and 0.242 for scalar and vector mediators, respectively. These values can be readily reached by forthcoming CMB experiments. Upon reaching these levels, future $N_{{\rm eff}}$ constraints on the couplings will be improved by many orders of magnitude., Comment: 25 pages, 6 figures and 2 tables, version accepted by JHEP

After being produced as electron neutrinos ($\nu_e$), solar neutrinos partially change their flavor to $\nu_{\mu}$ and $\nu_{\tau}$ en route to Earth. Although the flavor ratio of the $\nu_e$ flux to the total flux has been well measured, the $\nu_{\mu}:\nu_{\tau}$ composition has not yet been experimentally probed. In this work we investigate the potential of the next-generation experiments for measuring the $\nu_{\mu}:\nu_{\tau}$ flavor ratio by utilizing flavor-dependent radiative corrections in the cross sections for $\nu_{\mu}$ and $\nu_{\tau}$ scattering. Since the transition probabilities of $\nu_e$ to $\nu_\mu$ and $\nu_\tau$ depend on the leptonic CP phase, we also investigate the sensitivity to the CP phase and show that a statistical significance of $\sim1 \sigma$ could be reached through precision measurements of solar neutrino spectra., Comment: 6 pages, 5 figures

Neutrinos as almost massless particles could mediate long-range forces, known as neutrino forces. In this talk, I will introduce some theoretical aspects of neutrino forces, including why the potential of a neutrino force has the $1/r^{5}$ form and how it may vary under different circumstances. Experimental probes and possible implications for cosmology are also briefly discussed., Comment: 6 pages, 4 figures, Contribution to the 2023 Electroweak session of the 57th Rencontres de Moriond