Visual language models (VLMs) have made significant progress in image captioning tasks, yet recent studies have found they are vulnerable to backdoor attacks. Attackers can inject undetectable perturbations into the data during inference, triggering abnormal behavior and generating malicious captions. These attacks are particularly challenging to detect and defend against due to the stealthiness and cross-modal propagation of the trigger signals. In this paper, we identify two key vulnerabilities by analyzing existing attack patterns: (1) the model exhibits abnormal attention concentration on certain regions of the input image, and (2) backdoor attacks often induce semantic drift and sentence incoherence. Based on these insights, we propose Semantic Reward Defense (SRD), a reinforcement learning framework that mitigates backdoor behavior without requiring any prior knowledge of trigger patterns. SRD learns to apply discrete perturbations to sensitive contextual regions of image inputs via a deep Q-network policy, aiming to confuse attention and disrupt the activation of malicious paths. To guide policy optimization, we design a reward signal named semantic fidelity score, which jointly assesses the semantic consistency and linguistic fluency of the generated captions, encouraging the agent to achieve a robust yet faithful output. SRD offers a trigger-agnostic, policy-interpretable defense paradigm that effectively mitigates local (TrojVLM) and global (Shadowcast) backdoor attacks, reducing ASR to 3.6% and 5.6% respectively, with less than 15% average CIDEr drop on the clean inputs.

Quantum many-body problems are central to various scientific disciplines, yet their ground-state properties are intrinsically challenging to estimate. Recent advances in deep learning (DL) offer potential solutions in this field, complementing prior purely classical and quantum approaches. However, existing DL-based models typically assume access to a large-scale and noiseless labeled dataset collected by infinite sampling. This idealization raises fundamental concerns about their practical utility, especially given the limited availability of quantum hardware in the near term. To unleash the power of these DL-based models, we propose AiDE-Q (\underline{a}utomat\underline{i}c \underline{d}ata \underline{e}ngine for \underline{q}uantum property estimation), an effective framework that addresses this challenge by iteratively generating high-quality synthetic labeled datasets. Specifically, AiDE-Q utilizes a confidence-check method to assess the quality of synthetic labels and continuously improves the employed DL models with the identified high-quality synthetic dataset. To verify the effectiveness of AiDE-Q, we conduct extensive numerical simulations on a diverse set of quantum many-body and molecular systems, with up to 50 qubits. The results show that AiDE-Q enhances prediction performance for various reference learning models, with improvements of up to $14.2\\%$. Moreover, we exhibit that a basic supervised learning model integrated with AiDE-Q outperforms advanced reference models, highlighting the importance of a synthetic dataset. Our work paves the way for more efficient and practical applications of DL for quantum property estimation.

Atmospheric boundary layer detection is in urgent demand in meteorological and climate research and is closely related to human life and agricultural production. Atmospheric microwave fine-spectral sounding is a cutting-edge means to break through the regional and even global all-weather atmospheric boundary layer temperature and humidity profiling observations. In this paper, we carry out field campaigns using a unmanned aerial vehicle-based (UAV-based) fine-spectral microwave radiometer to solve key technologies of boundary layer detection and inversion, and realize the sounding of atmospheric water condensate and the study of the evolution.

Quantum Approximate Optimization Algorithm (QAOA) and its variants exhibit immense potential in tackling combinatorial optimization challenges. However, their practical realization confronts a dilemma: the requisite circuit depth for satisfactory performance is problem-specific and often exceeds the maximum capability of current quantum devices. To address this dilemma, here we first analyze the convergence behavior of QAOA, uncovering the origins of this dilemma and elucidating the intricate relationship between the employed mixer Hamiltonian, the specific problem at hand, and the permissible maximum circuit depth. Harnessing this understanding, we introduce the Mixer Generator Network (MG-Net), a unified deep learning framework adept at dynamically formulating optimal mixer Hamiltonians tailored to distinct tasks and circuit depths. Systematic simulations, encompassing Ising models and weighted Max-Cut instances with up to 64 qubits, substantiate our theoretical findings, highlighting MG-Net's superior performance in terms of both approximation ratio and efficiency.

The core of quantum machine learning is to devise quantum models with good trainability and low generalization error bounds than their classical counterparts to ensure better reliability and interpretability. Recent studies confirmed that quantum neural networks (QNNs) have the ability to achieve this goal on specific datasets. In this regard, it is of great importance to understand whether these advantages are still preserved on real-world tasks. Through systematic numerical experiments, we empirically observe that current QNNs fail to provide any benefit over classical learning models. Concretely, our results deliver two key messages. First, QNNs suffer from the severely limited effective model capacity, which incurs poor generalization on real-world datasets. Second, the trainability of QNNs is insensitive to regularization techniques, which sharply contrasts with the classical scenario. These empirical results force us to rethink the role of current QNNs and to design novel protocols for solving real-world problems with quantum advantages.

Polarimetric Microwave Imager (PMI) is planning as one of the main payload on WCOM satellite with the purposes of collecting the radiation from the Earth surface to retrieve the geophysical parameters of the land, ocean and atmosphere. PMI is passive microwave radiometer with frequencies of 6.8, 10.7, 18.7, 23.8, 37 .0 and 89GHz. The radiometers at 10.7, 18.7 and 37.0 GHz are full-polarized to measure four Stokes of the surface, whether from the ocean or the land. The other frequencies are traditional radiometer with vertical and horizontal channels. Since the third and the fourth Stokes parameter are sensitive to the orientation feature of the surface, they can be used to retrieve the wind vector, rain, snow cover, forest distribution and polar icecaps. Furthermore, with the higher resolutions at 37.0 and 89.0 GHz, the freeze-thaw on the land and rain on the ocean can be more accurately evaluated. In the paper, some primacy consideration and design on PMI are given for the application fulfilling the requirements of WCOM.

Since the first polarimetric radiometer satellite WindSat was launched on January 6, 2003 [1] and has been working normally until now, microwave polarimetry has demonstrated its ability on retrieving global wind vector [2].The first digital correlation microwave polarimeter was designed in 2001 [3], and digital technology has made a great progress during the last decade. We describe design of a digital-correlation full-polarimetric microwave radiometer system (DPMR), which operates at the same frequencies as WindSat, i.e. 6.8, 10.7, 18.7, 23.8 and 37.0 GHz. A direct digital cross-cerrelation technique is used in the system to produce four Stokes parameters at each frequency simultaneously. The main specifications are list in TABLE 1.

In this paper we present calibration and error analysis of our Ka-band fully polarimetric radiometer, the radiometer used a wideband microstrip correlator. The calibration method is based on the tri-polarimetric calibration technique of Gasiewski and Kunkee, but is extended to provide the calibration of the fourth Stokes parameter using a retarder plate. A method to determine the phase shift of the retarder plate is introduced, a calibration matrix inversion technique is described as well. The calibration uncertainties and calibration error analysis are discussed at last.