arXiv subjects

Kai Zhong

Publications and source records attributed to Kai Zhong.

At least 19 recordsLinked to original sources

arXiv2026

Movable IRS-Aided ISAC Systems: Joint Beamforming and Position Optimization

Driven by intelligent reflecting surface (IRS) and movable antenna (MA) technologies, movable IRS (MIRS) has been proposed to improve the adaptability and performance of conventional IRS, enabling flexible adjustment of the IRS reflecting element positions. This paper investigates MIRS-aided integrated sensing and communication (ISAC) systems. The objective is to minimize the power required for satisfying the quality-of-service (QoS) of sensing and communication by jointly optimizing the MIRS element positions, IRS reflection coefficients, transmit beamforming, and receive filters. To balance the performance-cost trade-off, we proposed two MIRS schemes: element-wise control and array-wise control, where the positions of individual reflecting elements and arrays consisting of multiple elements are controllable, respectively. To address the joint beamforming and position optimization, a product Riemannian manifold optimization (PRMO) method is proposed, where the variables are updated over a constructed product Riemannian manifold space (PRMS) in parallel via penalty-based transformation and Riemannian Broyden-Fletcher-Goldfarb-Shanno (RBFGS) algorithm. Simulation results demonstrate that the proposed MIRS outperforms conventional IRS in power minimization with both element-wise control and array-wise control. Specifically, with different system parameters, the minimum power is achieved by the MIRS with the element-wise control scheme, while suboptimal solution and higher computational efficiency are achieved by the MIRS with array-wise control scheme.

eess.SP↗

arXiv2026

3D Primitives are a Spatial Language for VLMs

Vision-language models (VLMs) exhibit a striking paradox: they can generate executable code that reconstructs a 3D scene from geometric primitives with correct object counts, classes, and approximate positions, yet the same models fail at simpler spatial questions on the same image. We show that 3D geometric primitives (cubes, spheres, cylinders, expressed in executable code) serve as a powerful intermediate representation for spatial understanding, and exploit this through three contributions. First, we introduce \textbf{\textsc{SpatialBabel}}, a benchmark evaluating fourteen VLMs on primitive-based 3D scene reconstruction across six \emph{scene-code languages} (programming languages and declarative formats for 3D primitive scenes), revealing that a single model's object-detection F1 can vary by up to $5.7\times$ across languages. Second, we propose \textbf{Code-CoT} (Code Chain-of-Thought), a training-free inference strategy that routes spatial reasoning through primitive-based code generation. Code-CoT lifts the SpatialBabel-QA-Score by up to $+6.4$\% on primitive scenes and real-photo CV-Bench-3D accuracy by $+5.0$\% for VLMs with strong coding capabilities. Third, we propose \textbf{S$^{3}$-FT} (Self-Supervised Spatial Fine-Tuning), which self-supervisedly distills primitive spatial knowledge into general visual reasoning by parsing the model's own Three.js primitive-reconstructions into structured annotations and fine-tuning on the result, with \emph{no human labels and no teacher model}. Training on primitive images alone, S$^3$-FT improves Qwen3-VL-8B by $+4.6$ to $+8.6$\% on SpatialBabel-Primitive-QA, $+9.7$\% on CV-Bench-2D, and $+17$\% on HallusionBench; the recipe transfers across model families. These results establish geometric primitives in code as both a diagnostic and a transferable spatial vocabulary for VLMs. We will release all artifacts upon publication.

cs.CV↗

arXiv2026

LLM Agents Enable User-Governed Personalization Beyond Platform Boundaries

Personalization today is fundamentally platform-centric: services build user representations from the behavioral fragments they observe. Yet no platform can construct a complete picture of the user, as competitive incentives, legal constraints, user privacy concerns, and epistemic limits create persistent data barriers. This paper argues for a shift from platform-centric personalization to user-governed personalization, where only the user can integrate fragmented contexts across platforms and the offline world. The key asymmetry lies in data access: only users can aggregate their own cross-platform and offline information. Large language model (LLM) agents make such integration practically feasible for the first time by enabling reasoning over heterogeneous personal data and transforming users' cross-context information into actionable personalization capabilities. We provide proof-of-concept evidence that users equipped with cross-platform data exports and an off-the-shelf LLM agent can outperform single-platform personalization baselines. We conclude by outlining a research agenda for building scalable user-governed personalization systems.

cs.IR↗

arXiv2026

RankFlow: A Multi-Role Collaborative Reranking Workflow Utilizing Large Language Models

In an Information Retrieval (IR) system, reranking plays a critical role by sorting candidate passages according to their relevance to a specific query. This process demands a nuanced understanding of the variations among passages linked to the query. In this work, we introduce RankFlow, a multi-role reranking workflow that leverages the capabilities of Large Language Models (LLMs) and role specializations to improve reranking performance. RankFlow enlists LLMs to fulfill four distinct roles: the query Rewriter, the pseudo Answerer, the passage Summarizer, and the Reranker. This orchestrated approach enables RankFlow to: (1) accurately interpret queries, (2) draw upon LLMs' extensive pre-existing knowledge, (3) distill passages into concise versions, and (4) assess passages in a comprehensive manner, resulting in notably better reranking results. Our experimental results reveal that RankFlow outperforms existing leading approaches on widely recognized IR benchmarks, such as TREC-DL, BEIR, and NovelEval. Additionally, we investigate the individual contributions of each role in RankFlow.

cs.IR↗

arXiv2026

Fairness-Aware Beamforming for Polarimetric ISAC Systems with Polarization-Reconfigurable Antennas

Polarization diversity offers significant flexibility for enhancing integrated sensing and communications (ISAC). However, conventional dual-polarized arrays typically require dedicated radio-frequency (RF) chains for each polarization branch, leading to prohibitive hardware costs. To address this, polarization-reconfigurable (PR) antennas have emerged as a cost-effective alternative, enabling polarization flexibility with reduced hardware complexity by driving two polarization branches with a single RF chain. In this paper, we investigate fairness-aware beamforming for ISAC systems equipped with PR antennas. Specifically, we jointly optimize the transmit beamforming and PR control coefficients to maximize the minimum signal-to-interference-plus-noise ratio (SINR) for communication users and the minimum signal-to-clutter-plus-noise ratio (SCNR) for sensing targets. The resulting problem is highly nonconvex and nonsmooth due to the strong coupling among optimization variables in the max-min objective, as well as the nonconvex spherical constraints imposed by the PR antennas. To tackle this, we derive an equivalent smooth reformulation by introducing auxiliary variables and transforming the minimum operators into inequality constraints. Subsequently, we develop an exact-penalty product Riemannian manifold gradient descent (EP-PRMGD) algorithm, which integrates an exact penalty method with Riemannian optimization to guarantee convergence to a Karush-Kuhn-Tucker (KKT) point. Numerical results demonstrate that the proposed PR-enabled ISAC scheme achieves performance comparable to dual-polarized architectures while utilizing only half the RF chains, thereby validating its effectiveness in balancing fairness and hardware efficiency.

eess.SP↗

arXiv2026

SFT Doesn't Always Hurt General Capabilities: Revisiting Domain-Specific Fine-Tuning in LLMs

Supervised Fine-Tuning (SFT) on domain-specific datasets is a common approach to adapt Large Language Models (LLMs) to specialized tasks but is often believed to degrade their general capabilities. In this work, we revisit this trade-off and present both empirical and theoretical insights. First, we show that SFT does not always hurt: using a smaller learning rate can substantially mitigate general performance degradation while preserving comparable target-domain performance. We then provide a theoretical analysis that explains these phenomena and further motivates a new method, Token-Adaptive Loss Reweighting (TALR). Building on this, and recognizing that smaller learning rates alone do not fully eliminate general-performance degradation in all cases, we evaluate a range of strategies for reducing general capability loss, including L2 regularization, LoRA, model averaging, FLOW, and our proposed TALR. Experimental results demonstrate that while no method completely eliminates the trade-off, TALR consistently outperforms these baselines in balancing domain-specific gains and general capabilities. Finally, we distill our findings into practical guidelines for adapting LLMs to new domains: (i) using a small learning rate to achieve a favorable trade-off, and (ii) when a stronger balance is further desired, adopt TALR as an effective strategy.

cs.CL↗

arXiv2026

Cooperative Double IRS aided Secure Communication for MIMO-OFDM Systems

Cooperative double intelligent reflecting surface (double-IRS) has emerged as a promising approach for enhancing physical layer security (PLS) in MIMO systems. However, existing studies are limited to narrowband scenarios and fail to address wideband MIMO-OFDM. In this regime, frequency-flat IRS phases and cascaded IRS links cause severe coupling, rendering narrowband designs inapplicable. To overcome this challenge, we introduce cooperative double-IRS-assisted wideband MIMO-OFDM and propose an efficient manifold-based solution. By regarding the power and constant modulus constraints as Riemannian manifolds, we reformulate the non-convex secrecy sum rate maximization as an unconstrained optimization on a product manifold. Building on this formulation, we further develop a product Riemannian gradient descent (PRGD) algorithm with guaranteed stationary convergence. Simulation results demonstrate that the proposed scheme effectively resolves the OFDM coupling issue and achieves significant secrecy rate gains, outperforming single-IRS and distributed multi-IRS benchmarks by 32.0% and 22.3%, respectively.

eess.SP↗

arXiv2025

Secure Analog Beamforming for Multi-user MISO Systems with Movable Antennas

Movable antennas (MAs) represent a novel approach that enables flexible adjustments to antenna positions, effectively altering the channel environment and thereby enhancing the performance of wireless communication systems. However, conventional MA implementations often adopt fully digital beamforming (FDB), which requires a dedicated RF chain for each antenna. This requirement significantly increase hardware costs, making such systems impractical for multi-antenna deployments. To address this, hardware-efficient analog beamforming (AB) offers a cost-effective alternative. This paper investigates the physical layer security (PLS) in an MA-enabled multiple-input single-output (MISO) communication system with an emphasis on AB. In this scenario, an MA-enabled transmitter with AB broadcasts common confidential information to a group of legitimate receivers, while a number of eavesdroppers overhear the transmission and attempt to intercept the information. Our objective is to maximize the multicast secrecy rate (MSR) by jointly optimizing the phase shifts of the AB and the positions of the MAs, subject to constraints on the movement area of the MAs and the constant modulus (CM) property of the analog phase shifters. This MSR maximization problem is highly challenging, as we have formally proven it to be NP-hard. To solve it efficiently, we propose a penalty constrained product manifold (PCPM) framework. Specifically, we first reformulate the position constraints as a penalty function, enabling unconstrained optimization on a product manifold space (PMS), and then propose a parallel conjugate gradient descent algorithm to efficiently update the variables. Simulation results demonstrate that MA-enabled systems with AB can achieve a well-balanced performance in terms of MSR and hardware costs.

eess.SP↗

arXiv2025

Secure Analog Beamforming Design for Wireless Communication Systems With Movable Antennas

Movable antennas (MA) are a novel technology that allows for the flexible adjustment of antenna positions within a specified region, thereby enhancing the performance of wireless communication systems. In this paper, we explore the use of MA to improve physical layer security in an analog beamforming (AB) communication system. Our goal is to maximize the secrecy rate by jointly optimizing the transmit AB and MA position, subject to constant modulus (CM) constraints on the AB and position constraints for the MA. The resulting problem is non-convex, and we propose a penalty product manifold (PPM) method to solve it efficiently. Specifically, we convert the inequality constraints related to MA position into a penalty function using smoothing techniques, thereby reformulating the problem as an unconstrained optimization on the product manifold space (PMS). We then derive a parallel conjugate gradient descent (PCGD) algorithm to update both the AB and MA position on the PMS. This method is efficient, providing an analytical solution at each step and ensuring convergence to a KKT point. Simulation results show that the MA system achieves a higher secrecy rate than systems with fixed-position antennas.

eess.SP↗

arXiv2025

Joint Beamforming and Antenna Position Optimization for IRS-Aided Multi-User Movable Antenna Systems

Intelligent reflecting surface (IRS) and movable antenna (MA) technologies have been proposed to enhance wireless communications by creating favorable channel conditions. This paper investigates the joint beamforming and antenna position optimization for MA-enabled IRS (MA-IRS)-aided multi-user multiple-input single-output (MU-MISO) communication systems, where the MA-IRS is deployed to aid the communication between the MA-enabled base station (BS) and user equipment (UE). In contrast to conventional fixed position antenna (FPA)-enabled IRS (FPA-IRS), the positions of the reflecting elements of the MA-IRS can be controlled to enhances the wireless channel. To verify the system's effectiveness and optimize its performance, we formulate a sum-rate maximization problem with a minimum rate threshold constraint for the MU-MISO communication. To tackle the non-convex problem, a product Riemannian manifold optimization (PRMO) method is proposed for the joint optimization of the beamforming and MA positions. Specifically, a product Riemannian manifold space (PRMS) is constructed and the corresponding Riemannian gradient is derived for updating the variables, and the Riemannian exact penalty (REP) method and a Riemannian Broyden-Fletcher-Goldfarb-Shanno (RBFGS) algorithm is exploited to obtain a feasible solution over the PRMS. Simulation results demonstrate that compared with the conventional FPA-IRS-aided communications, the reflecting elements of the MA-IRS can move to the positions with higher channel gain, thus enhancing the system performance. Furthermore, it is shown that optimizing the positions of the reflecting elements brings higher performance gain than controlling the phase shifts of the IRS, and integrating MA with IRS leads to higher performance gains compared to integrating MA with BS.

eess.SP↗

arXiv2025

APEER: Automatic Prompt Engineering Enhances Large Language Model Reranking

Large Language Models (LLMs) have significantly enhanced Information Retrieval (IR) across various modules, such as reranking. Despite impressive performance, current zero-shot relevance ranking with LLMs heavily relies on human prompt engineering. Existing automatic prompt engineering algorithms primarily focus on language modeling and classification tasks, leaving the domain of IR, particularly reranking, underexplored. Directly applying current prompt engineering algorithms to relevance ranking is challenging due to the integration of query and long passage pairs in the input, where the ranking complexity surpasses classification tasks. To reduce human effort and unlock the potential of prompt optimization in reranking, we introduce a novel automatic prompt engineering algorithm named APEER. APEER iteratively generates refined prompts through feedback and preference optimization. Extensive experiments with four LLMs and ten datasets demonstrate the substantial performance improvement of APEER over existing state-of-the-art (SoTA) manual prompts. Furthermore, we find that the prompts generated by APEER exhibit better transferability across diverse tasks and LLMs.

cs.AI↗

arXiv2024

FOSS: A Self-Learned Doctor for Query Optimizer

Various works have utilized deep learning to address the query optimization problem in database system. They either learn to construct plans from scratch in a bottom-up manner or steer the plan generation behavior of traditional optimizer using hints. While these methods have achieved some success, they face challenges in either low training efficiency or limited plan search space. To address these challenges, we introduce FOSS, a novel framework for query optimization based on deep reinforcement learning. FOSS initiates optimization from the original plan generated by a traditional optimizer and incrementally refines suboptimal nodes of the plan through a sequence of actions. Additionally, we devise an asymmetric advantage model to evaluate the advantage between two plans. We integrate it with a traditional optimizer to form a simulated environment. Leveraging this simulated environment, FOSS can bootstrap itself to rapidly generate a large amount of high-quality simulated experiences. FOSS then learns from these experiences to improve its optimization capability. We evaluate the performance of FOSS on Join Order Benchmark, TPC-DS, and Stack Overflow. The experimental results demonstrate that FOSS outperforms the state-of-the-art methods in terms of latency performance. Compared to PostgreSQL, FOSS achieves speedup ranging from 1.15x to 8.33x in total latency across different benchmarks.

cs.DB↗

arXiv2022

Enterprise-Scale Search: Accelerating Inference for Sparse Extreme Multi-Label Ranking Trees

Tree-based models underpin many modern semantic search engines and recommender systems due to their sub-linear inference times. In industrial applications, these models operate at extreme scales, where every bit of performance is critical. Memory constraints at extreme scales also require that models be sparse, hence tree-based models are often back-ended by sparse matrix algebra routines. However, there are currently no sparse matrix techniques specifically designed for the sparsity structure one encounters in tree-based models for extreme multi-label ranking/classification (XMR/XMC) problems. To address this issue, we present the masked sparse chunk multiplication (MSCM) technique, a sparse matrix technique specifically tailored to XMR trees. MSCM is easy to implement, embarrassingly parallelizable, and offers a significant performance boost to any existing tree inference pipeline at no cost. We perform a comprehensive study of MSCM applied to several different sparse inference schemes and benchmark our methods on a general purpose extreme multi-label ranking framework. We observe that MSCM gives consistently dramatic speedups across both the online and batch inference settings, single- and multi-threaded settings, and on many different tree models and datasets. To demonstrate its utility in industrial applications, we apply MSCM to an enterprise-scale semantic product search problem with 100 million products and achieve sub-millisecond latency of 0.88 ms per query on a single thread -- an 8x reduction in latency over vanilla inference techniques. The MSCM technique requires absolutely no sacrifices to model accuracy as it gives exactly the same results as standard sparse matrix techniques. Therefore, we believe that MSCM will enable users of XMR trees to save a substantial amount of compute resources in their inference pipelines at very little cost.

cs.LG↗

arXiv2022

PECOS: Prediction for Enormous and Correlated Output Spaces

Many large-scale applications amount to finding relevant results from an enormous output space of potential candidates. For example, finding the best matching product from a large catalog or suggesting related search phrases on a search engine. The size of the output space for these problems can range from millions to billions, and can even be infinite in some applications. Moreover, training data is often limited for the long-tail items in the output space. Fortunately, items in the output space are often correlated thereby presenting an opportunity to alleviate the data sparsity issue. In this paper, we propose the Prediction for Enormous and Correlated Output Spaces (PECOS) framework, a versatile and modular machine learning framework for solving prediction problems for very large output spaces, and apply it to the eXtreme Multilabel Ranking (XMR) problem: given an input instance, find and rank the most relevant items from an enormous but fixed and finite output space. We propose a three phase framework for PECOS: (i) in the first phase, PECOS organizes the output space using a semantic indexing scheme, (ii) in the second phase, PECOS uses the indexing to narrow down the output space by orders of magnitude using a machine learned matching scheme, and (iii) in the third phase, PECOS ranks the matched items using a final ranking scheme. The versatility and modularity of PECOS allows for easy plug-and-play of various choices for the indexing, matching, and ranking phases. We also develop very fast inference procedures which allow us to perform XMR predictions in real time; for example, inference takes less than 1 millisecond per input on the dataset with 2.8 million labels. The PECOS software is available at https://libpecos.org.

cs.LG↗

arXiv2021

Doppler velocity-based algorithm for Clustering and Velocity Estimation of moving objects

We propose a Doppler velocity-based cluster and velocity estimation algorithm based on the characteristics of FMCW LiDAR which achieves highly accurate, single-scan, and real-time motion state detection and velocity estimation. We prove the continuity of the Doppler velocity on the same object. Based on this principle, we achieve the distinction between moving objects and stationary background via region growing clustering algorithm. The obtained stationary background will be used to estimate the velocity of the FMCW LiDAR by the least-squares method. Then we estimate the velocity of the moving objects using the estimated LiDAR velocity and the Doppler velocity of moving objects obtained by clustering. To ensure real-time processing, we set the appropriate least-squares parameters. Meanwhile, to verify the effectiveness of the algorithm, we create the FMCW LiDAR model on the autonomous driving simulation platform CARLA for spawning data. The results show that our algorithm can process at least a 4.5million points and estimate the velocity of 150 moving objects per second under the arithmetic power of the Ryzen 3600x CPU, with a motion state detection accuracy of over 99% and estimated velocity accuracy of 0.1 m/s.

cs.RO↗

arXiv2021

Exploring the Potential of Low-bit Training of Convolutional Neural Networks

In this work, we propose a low-bit training framework for convolutional neural networks, which is built around a novel multi-level scaling (MLS) tensor format. Our framework focuses on reducing the energy consumption of convolution operations by quantizing all the convolution operands to low bit-width format. Specifically, we propose the MLS tensor format, in which the element-wise bit-width can be largely reduced. Then, we describe the dynamic quantization and the low-bit tensor convolution arithmetic to leverage the MLS tensor format efficiently. Experiments show that our framework achieves a superior trade-off between the accuracy and the bit-width than previous low-bit training frameworks. For training a variety of models on CIFAR-10, using 1-bit mantissa and 2-bit exponent is adequate to keep the accuracy loss within $1\%$. And on larger datasets like ImageNet, using 4-bit mantissa and 2-bit exponent is adequate to keep the accuracy loss within $1\%$. Through the energy consumption simulation of the computing units, we can estimate that training a variety of models with our framework could achieve $8.3\sim10.2\times$ and $1.9\sim2.3\times$ higher energy efficiency than training with full-precision and 8-bit floating-point arithmetic, respectively.

cs.LG↗

arXiv2021

Extreme Multi-label Learning for Semantic Matching in Product Search

We consider the problem of semantic matching in product search: given a customer query, retrieve all semantically related products from a huge catalog of size 100 million, or more. Because of large catalog spaces and real-time latency constraints, semantic matching algorithms not only desire high recall but also need to have low latency. Conventional lexical matching approaches (e.g., Okapi-BM25) exploit inverted indices to achieve fast inference time, but fail to capture behavioral signals between queries and products. In contrast, embedding-based models learn semantic representations from customer behavior data, but the performance is often limited by shallow neural encoders due to latency constraints. Semantic product search can be viewed as an eXtreme Multi-label Classification (XMC) problem, where customer queries are input instances and products are output labels. In this paper, we aim to improve semantic product search by using tree-based XMC models where inference time complexity is logarithmic in the number of products. We consider hierarchical linear models with n-gram features for fast real-time inference. Quantitatively, our method maintains a low latency of 1.25 milliseconds per query and achieves a 65% improvement of Recall@100 (60.9% v.s. 36.8%) over a competing embedding-based DSSM model. Our model is robust to weight pruning with varying thresholds, which can flexibly meet different system requirements for online deployments. Qualitatively, our method can retrieve products that are complementary to existing product search system and add diversity to the match set.

cs.IR↗

arXiv2021

BoolNet: Minimizing The Energy Consumption of Binary Neural Networks

Recent works on Binary Neural Networks (BNNs) have made promising progress in narrowing the accuracy gap of BNNs to their 32-bit counterparts. However, the accuracy gains are often based on specialized model designs using additional 32-bit components. Furthermore, almost all previous BNNs use 32-bit for feature maps and the shortcuts enclosing the corresponding binary convolution blocks, which helps to effectively maintain the accuracy, but is not friendly to hardware accelerators with limited memory, energy, and computing resources. Thus, we raise the following question: How can accuracy and energy consumption be balanced in a BNN network design? We extensively study this fundamental problem in this work and propose a novel BNN architecture without most commonly used 32-bit components: \textit{BoolNet}. Experimental results on ImageNet demonstrate that BoolNet can achieve 4.6x energy reduction coupled with 1.2\% higher accuracy than the commonly used BNN architecture Bi-RealNet. Code and trained models are available at: https://github.com/hpi-xnor/BoolNet.

cs.LG↗