Large language models (LLMs) still struggle with multi-hop reasoning over knowledgegraphs (KGs), and we identify a previously overlooked structural reason for this difficulty: Transformer attention heads naturally specialize in distinct semantic relations across reasoning stages, forming a hop-aligned relay pattern.This key finding suggests that multi-hop reasoning is inherently multi-view, yet existing KG-based retrieval-augmented generation (KG-RAG) systems collapse all reasoning hops into a single representation, flat embedding space, suppressing this implicit structure and causing noisy or drifted path exploration.We introduce ParallaxRAG, a symmetric multi-view framework that decouples queries and KGs into aligned, head-specific semantic spaces.By enforcing relational diversity across multiple heads while constraining weakly related paths, ParallaxRAG constructs more accurate, cleaner subgraphs and guides LLMs through grounded, hop-wise reasoning.On WebQSP and CWQ, it achieves state-of-the-art retrieval and QA performance, substantially reduces hallucination, and generalizes strongly to the biomedical BioASQ benchmark.Our implementation is available at https://github.com/ LucaLiu1313/ParallaxRAG.

In this paper, we propose a dynamic agile reconfigurable intelligent surface antenna (DARISA) array integrated into multi-input multi-output (MIMO) transceivers. Each DARISA comprises a number of metasurface elements activated simultaneously via a parallel feed network. The proposed system enables rapid and intelligent phase response adjustments for each metasurface element within a single symbol duration, facilitating a dynamic agile adjustment of phase response (DAAPR) strategy. By analyzing the theoretical degrees of freedom (DoF) of the DARISA MIMO system under the DAAPR framework, we derive an explicit relationship between DoF and critical system parameters, including agility frequentness (i.e., the number of phase adjustments of metasurface elements during one symbol period), cluster angular spread of wireless channels, DARISA array size, and the number of transmit/receive DARISAs. The DoF result reveals a significant conclusion: when the number of receive DARISAs is smaller than that of transmit DARISAs, the DAAPR strategy of the DARISA MIMO enhances the overall system DoF. Furthermore, relying on DoF alone to measure channel capacity is insufficient, so we analyze the effective DoF (EDoF) that reflects the impacts of the DoF and channel matrix singular value distribution on capacity. We show channel capacity monotonically increases with EDoF, and optimize the agile phase responses of metasurface elements by using fractional programming (FP) and semidefinite relaxation (SDR) algorithms to maximize the EDoF. Simulations validate the theoretical DoF gains and reveal that increasing agility frequentness, metasurface element density, and phase quantization accuracy can enhance the EDoF. Additionally, densely deployed elements can compensate for the loss in communication performance caused by lower phase quantization accuracy.

This paper investigates an intelligent reflecting surface (IRS) aided green multiple-user downlink communication system. In contrast to the existing works that deploy the IRS in a fixed location, the location of the IRS is taken as an optimization variable to minimize the total transmit power by jointly optimizing the location of the IRS, transmit beamformers at the base station (BS), and IRS phase shifts. We point out a critical conclusion that before and after IRS deployment, the channel state information (CSI) of all the communication terminals is different, so an offline-online hybrid-CSI optimization framework is proposed to solve the problem. In the offline stage, we optimize the IRS location with only the statistical CSI (S-CSI) so the ergodic quality of service (QoS) constraints have to be considered, and universal lower bounds associated only with the location variable are derived to decouple all variables. In the online stage, all the instantaneous-CSI (I-CSI) are available. To solve this non-convex problem, an alternating optimization framework is developed. We propose a Riemannian Manifold (RM) algorithm to optimize the IRS phase shifts. Simulation results validate that the proposed algorithm is convergent and effective, and show that the location deployment of IRS is crucial for green communication.

In this paper, we propose a robust secrecy transmission scheme for intelligent reflecting surface (IRS) aided communication systems. Different from all the existing works where IRS has already been deployed at a fixed location, we take the location of IRS as a variable to maximize the secrecy rate (SR) under the outage probability constraint by jointly optimizing the location of IRS, transmit beamformer and IRS phase shifts with imperfect channel state information (CSI) of Eve, where we consider two cases: a) the location of Eve is known; b) only a suspicious area of Eve is available. We show a critical observation that CSI models are different before and after IRS deployment, thus the optimization problem could be decomposed and solved via a two-stage framework. For case a), in the first stage, universal upper bounds of outage probabilities only related to the location of IRS are derived which can be optimized via successive convex approximation (SCA) method. In the second stage, we develop an alternative optimization (AO) algorithm to optimize beamformer and phase shifts iteratively. For case b), we propose a Max-Min SR scheme based on two-stage framework, where the location of IRS is optimized based on the worst location of Eve. Simulation results indicate the importance of the location of IRS optimization.

Intelligent reflecting surface (IRS) has been proposed as a very promising technique for beyond 5G and 6G communications. In this paper, we apply IRS to enhance the secure transmission of secondary user in a multi-input multioutput (MIMO) cognitive radio (CR) wiretap channel. Since the study of secure IRS-assisted CR communication is still an open problem, all the existing numerical solutions for enhancing secure communications in non-CR setting as well as non-secure communications in CR setting in the literature fail to this work due to the complicated structure of objective functions as well as the constraints. Therefore, to maximize the secrecy rate of secondary user, an efficient alternating optimization (AO) algorithm is proposed to jointly optimize the transmit covariance at base station and phase shift coefficients at IRS. Simulation results show that our proposed algorithm have fast monotonic convergence as well as better performance on enhancing the secrecy rate than the benchmark schemes.

In this letter, improving the security of an intelligent reflecting surface (IRS) assisted multiple-input single-output (MISO) communication system is studied. Different from the ideal assumption in existing literatures that full eavesdropper's (Eve's) channel state information (CSI) is available, we consider a more practical scenario without Eve's CSI. To enhance the security of this system given a total transmit power at transmitter (Alice), we propose a joint beamforming and jamming approach, in which a minimum transmit power is firstly optimized at Alice so as to meet the quality of service (QoS) at legitimate user (Bob), and then artificial noise (AN) is emitted to jam the eavesdropper by using the residual power at Alice. Two efficient algorithms exploiting oblique manifold (OM) and minorization-maximization (MM) algorithms, respectively, are developed for solving the resulting non-convex optimization problem. Simulation results have been provided to validate the performance and convergence of the proposed algorithms.

Hashing has been widely used in large-scale image retrieval based on approximate nearest neighbor search. Most learning-to-hashing methods adopt a two-stage algorithm to generate binary codes. First, original images are mapped into continuous visual features. Then, binary codes are generated by quantization step or separate projection. Nevertheless, these methods are sensitive to quantization operation, i.e., thresholding. To explicitly address this issue, this study proposes a novel feature quantization scheme with a loopy recurrent neural network, called loopy residual hashing, for the purpose of high accuracy in image retrieval. Instead of one-off thresholding-based feature binarization, the proposed approach performs an iterative threshold-then-approximate operation, which calculates the quantization residual after each thresholding step and then imitates another round of binarization to further approximate the coding residual. The resulting sequences of binary codes possess higher representation accuracy and extensive experiments on image retrieval demonstrate its superior discriminative capability over the prior art. In the meantime, theoretical approximation error analysis is given.

Hashing is a widely adopted method based on an approximate nearest neighbor search and is used in large-scale image retrieval tasks. Conventional learning-based hashing algorithms employ end-to-end representation learning, which is a one-off technique. Because of the tradeoff between efficiency and performance, conventional learning-based hashing methods must sacrifice code length to improve performance, which increases their computational complexity. To improve the efficiency of binary codes, motivated by the “nonsalient-to-salient” attention scheme of humans, we propose a recursive hashing mechanism that maps progressively expanded salient regions to a series of binary codes. These salient regions are generated by a conventional saliency model based on bottom-up saliency-driven attention and a semantic-guided saliency model based on top-down task-driven attention. After obtaining a series of salient regions, we perform long-range temporal modeling of salient regions using a graph-based recurrent deep network to obtain more refined representative features. The later output nodes inherit aggregated information from all previous nodes and extract discriminative features from more salient regions. Therefore, this network possesses more significant information and satisfactory scalability. The proposed recursive hashing neural network, optimized by a triplet ranking loss, is end-to-end trainable. Extensive experimental results from several image retrieval benchmarks show the scalability of our method and demonstrate its strong performance compared with state-of-the-art methods.

This paper proposes a novel recursive hashing scheme, in contrast to conventional "one-off" based hashing algorithms. Inspired by human's "nonsalient-to-salient" perception path, the proposed hashing scheme generates a series of binary codes based on progressively expanded salient regions. Built on a recurrent deep network, i.e., LSTM structure, the binary codes generated from later output nodes naturally inherit information aggregated from previously codes while explore novel information from the extended salient region, and therefore it possesses good scalability property. The proposed deep hashing network is trained via minimizing a triplet ranking loss, which is end-to-end trainable. Extensive experimental results on several image retrieval benchmarks demonstrate good performance gain over state-of-the-art image retrieval methods and its scalability property.