Search

To mitigate the hallucination and knowledge deficiency in large language models (LLMs), Knowledge Graph (KG)-based Retrieval-Augmented Generation (RAG) has shown promising potential by utilizing KGs as external resource to enhance LLMs reasoning. However, existing KG-RAG approaches struggle with a trade-off between flexibility and retrieval quality. Modular methods prioritize flexibility by avoiding the use of KG-fine-tuned models during retrieval, leading to fixed retrieval strategies and suboptimal retrieval quality. Conversely, coupled methods embed KG information within models to improve retrieval quality, but at the expense of flexibility. In this paper, we propose a novel flexible modular KG-RAG framework, termed FRAG, which synergizes the advantages of both approaches. FRAG estimates the hop range of reasoning paths based solely on the query and classify it as either simple or complex. To match the complexity of the query, tailored pipelines are applied to ensure efficient and accurate reasoning path retrieval, thus fostering the final reasoning process. By using the query text instead of the KG to infer the structural information of reasoning paths and employing adaptable retrieval strategies, FRAG improves retrieval quality while maintaining flexibility. Moreover, FRAG does not require extra LLMs fine-tuning or calls, significantly boosting efficiency and conserving resources. Extensive experiments show that FRAG achieves state-of-the-art performance with high efficiency and low resource consumption.

The 3rd Workshop on Maritime Computer Vision (MaCVi) 2025 addresses maritime computer vision for Unmanned Surface Vehicles (USV) and underwater. This report offers a comprehensive overview of the findings from the challenges. We provide both statistical and qualitative analyses, evaluating trends from over 700 submissions. All datasets, evaluation code, and the leaderboard are available to the public at https://macvi.org/workshop/macvi25., Comment: Part of the MaCVi 2025 workshop

This paper presents PCDreamer, a novel method for point cloud completion. Traditional methods typically extract features from partial point clouds to predict missing regions, but the large solution space often leads to unsatisfactory results. More recent approaches have started to use images as extra guidance, effectively improving performance, but obtaining paired data of images and partial point clouds is challenging in practice. To overcome these limitations, we harness the relatively view-consistent multi-view diffusion priors within large models, to generate novel views of the desired shape. The resulting image set encodes both global and local shape cues, which is especially beneficial for shape completion. To fully exploit the priors, we have designed a shape fusion module for producing an initial complete shape from multi-modality input (\ie, images and point clouds), and a follow-up shape consolidation module to obtain the final complete shape by discarding unreliable points introduced by the inconsistency from diffusion priors. Extensive experimental results demonstrate our superior performance, especially in recovering fine details.

Large language models have achieved notable success across various domains, yet efficient inference is still limited by the quadratic computation complexity of the attention mechanism. The inference consists of prefilling and decoding phases. Although several attempts have been made to accelerate decoding, the inefficiency of the prefilling phase, especially for long-context tasks, remains a challenge. In this paper, we observe a locality in query criticality during the prefilling phase of long-context processing: adjacent query tokens tend to focus on similar subsets of the past Key-Value (KV) cache. Based on this observation, we propose CritiPrefill, a criticality-based segment-wise prefilling method. This method partitions the input sequence's queries and KV cache into segments and blocks, utilizing a segment-wise algorithm to estimate the query criticality. By pruning non-critical computations between query segments and cache blocks in the self-attention mechanism, the prefilling process can be significantly accelerated. Extensive evaluations on multiple long-context datasets show up to 2.7x speedup on Llama3-8B and 3.0x speedup on Yi-9B for 128K context length on a single A100 GPU, with minimal quality degradation.

With the advent of the big data and large language model era, zero-shot personalized rapid customization has emerged as a significant trend. In this report, we introduce Takin AudioLLM, a series of techniques and models, mainly including Takin TTS, Takin VC, and Takin Morphing, specifically designed for audiobook production. These models are capable of zero-shot speech production, generating high-quality speech that is nearly indistinguishable from real human speech and facilitating individuals to customize the speech content according to their own needs. Specifically, we first introduce Takin TTS, a neural codec language model that builds upon an enhanced neural speech codec and a multi-task training framework, capable of generating high-fidelity natural speech in a zero-shot way. For Takin VC, we advocate an effective content and timbre joint modeling approach to improve the speaker similarity, while advocating for a conditional flow matching based decoder to further enhance its naturalness and expressiveness. Last, we propose the Takin Morphing system with highly decoupled and advanced timbre and prosody modeling approaches, which enables individuals to customize speech production with their preferred timbre and prosody in a precise and controllable manner. Extensive experiments validate the effectiveness and robustness of our Takin AudioLLM series models. For detailed demos, please refer to https://everest-ai.github.io/takinaudiollm/., Comment: Technical Report; 18 pages; typos corrected, references added, demo url modified, author name modified

We provide well-founded semantics for a quantum programming language Qwhile-hp with heap manipulations, where allocation statements follow a dirty pattern, meaning that newly allocated qubits can nondeterministically assume arbitrary initial states. To thoroughly characterize heap manipulations in the quantum setting, we develop a quantum BI-style logic that includes interpretations for separating implication ($-\mkern-3mu*$) and separating conjunction ($*$). We then adopt this quantum BI-style logic as an assertion language to reason about heap-manipulated quantum programs and present a quantum separation logic which is sound and relatively complete. Finally, we apply our framework to verify the correctness of various practical quantum programs and to prove the correct usage of dirty ancilla qubits.

We prove that using global observables to train the matrix product state ansatz results in the vanishing of all partial derivatives, also known as barren plateaus, while using local observables avoids this. This ansatz is widely used in quantum machine learning for learning weakly entangled state approximations. Additionally, we empirically demonstrate that in many cases, the objective function is an inner product of almost sparse operators, highlighting the potential for classically simulating such a learning problem with few quantum resources. All our results are experimentally validated across various scenarios., Comment: 17 pages, 4 figures

The isomorphism problem for digraphs is a fundamental problem in graph theory. This problem for Cayley digraphs has been extensively investigated over the last half a century. In this paper, we consider this problem for $m$-Cayley digraphs which are generalization of Cayley digraphs. Let $m$ be a positive integer. A digraph admitting a group $G$ of automorphisms acting semiregularly on the vertices with exactly $m$ orbits is called an $m$-Cayley digraph of $G$. In particular, $1$-Cayley digraph is just the Cayley digraph. We first characterize the normalizer of $G$ in the full automorphism group of an $m$-Cayley digraph of a finite group $G$. This generalizes a similar result for Cayley digraph achieved by Godsil in 1981. Then we use this to study the isomorphisms of $m$-Cayley digraphs. The CI-property of a Cayley digraph (CI stands for `Cayley isomorphism') and the DCI-groups (whose Cayley digraphs are all CI-digraphs) are two key topics in the study of isomorphisms of Cayley digraphs. We generalize these concepts into $m$-Cayley digraphs by defining $m$CI- and $m$PCI-digraphs, and correspondingly, $m$DCI- and $m$PDCI-groups. Analogues to Babai's criterion for CI-digraphs are given for $m$CI- and $m$PCI-digraphs, respectively. With these we then classify finite $m$DCI-groups for each $m\geq 2$, and finite $m$PDCI-groups for each $m\geq 4$. Similar results are also obtained for $m$-Cayley graphs. Note that 1DCI-groups are just DCI-groups, and the classification of finite DCI-groups is a long-standing open problem that has been worked on a lot., Comment: 29

This paper presents a sophisticated reconfigurable metasurface architecture that introduces an advanced concept of flexible full-array space-time wavefront manipulation with enhanced dynamic capabilities. The practical 2-bit phase-shifting unit cell on the RIS is distinguished by its ability to maintain four stable phase states, each with ${90^ \circ }$ differences, and features an insertion loss of less than 0.6 dB across a bandwidth of 200 MHz. All reconfigurable units are equipped with meticulously designed control circuits, governed by an intelligent core composed of multiple Micro-Controller Units (MCUs), enabling rapid control response across the entire RIS array. Owing to the capability of each unit cell on the metasurface to independently switch states, the entire RIS is not limited to controlling general beams with specific directional patterns, but also generates beams with more complex structures, including multi-focus 3D spot beams and vortex beams. This development substantially broadens its applicability across various industrial wireless transmission scenarios. Moreover, by leveraging the rapid-respond space-time coding and the full-array independent programmability of the RIS prototyping operating at 10.7 GHz, we have demonstrated that: 1) The implementation of 3D spot beams scanning facilitates dynamic beam tracking and real-time communication under the indoor near-field scenario; 2) The rapid wavefront rotation of vortex beams enables precise modulation of signals within the Doppler domain, showcasing an innovative approach to wireless signal manipulation; 3) The beam steering experiments for blocking users under outdoor far-field scenarios, verifying the beamforming capability of the RIS board.

Large Language Models have excelled in various fields but encounter challenges in memory and time efficiency due to the expanding Key-Value (KV) cache required for long-sequence inference. Recent efforts try to reduce KV cache size to a given memory budget by evicting vast non-critical cache elements during runtime, while preserving generation quality. Our revisiting of current eviction methods reveals that they fundamentally minimize an upper bound of the $L_1$ eviction loss between the pre- and post-eviction outputs of multi-head self-attention mechanisms. Moreover, our analysis indicates that the common practices of uniformly assigning budgets across attention heads harm their post-eviction generation quality. In light of these findings, we propose a simple yet effective adaptive budget allocation algorithm. This algorithm not only optimizes the theoretical loss upper bound but also reduces the $L_1$ eviction loss in practice by aligning with the varied characteristics across different heads. By integrating this algorithm into two state-of-the-art methods, we demonstrate the effectiveness of using adaptive budget allocation to optimize KV cache eviction. Extensive evaluations on 16 datasets and the Needle-in-a-Haystack test confirm significant performance improvements across various tasks.

Materials with flat bands can serve as a promising platform to investigate strongly interacting phenomena. However, experimental realization of ideal flat bands is mostly limited to artificial lattices or moir\'e systems. Here we report a general way to construct one-dimensional (1D) flat bands in phosphorene nanoribbons (PNRs) with pentagonal nature: penta-hexa-PNRs and penta-dodeca-PNRs, wherein the corresponding 1D flat bands are directly verified by using angle-resolved photoemission spectroscopy. We confirm that the observed 1D flat bands originate from the electronic 1D zigzag and Lieb lattices, respectively, as revealed by the combination of bond-resolved scanning tunneling microscopy, scanning tunneling spectroscopy, tight-binding models, and first-principles calculations. Our study demonstrates a general way to construct 1D flat bands in 1D solid materials system, which provides a robust platform to explore strongly interacting phases of matter., Comment: 13 pages, 4 figures

High-resolution observations with GRAVITY-VLTI instrument have provided abundant information about the flares in Sgr A*, the supermassive black hole in our Galactic center, including the time-dependent location of the centroid (a "hot spot"), the light curve, and polarization. Yuan et al. (2009) proposed a "coronal mass ejection" model to explain the flares and their association with the plasma ejection. The key idea is that magnetic reconnection in the accretion flow produces the flares and results in the formation and ejection of flux ropes. The dynamical process proposed in the model has been confirmed by three-dimensional GRMHD simulations in a later work. Based on this scenario, in our previous works the radiation of the flux rope has been calculated analytically and compared to the observations. In the present paper, we develop the model by directly using numerical simulation data to interpret observations. We first identify flux ropes formed due to reconnection from the data. By assuming that electrons are accelerated in the reconnection current sheet and flow into the flux rope and emit their radiation there, we have calculated the time-dependent energy distribution of electrons after phenomenologically considering their injection due to reconnection acceleration, radiative and adiabatic cooling. The radiation of these electrons is calculated using the ray-tracing approach. The trajectory of the hot spot, the radiation light curve during the flare, and the polarization are calculated. These results are compared with the GRAVITY observations and good consistencies are found., Comment: 15 pages, 16 figures

In this paper, we propose an environment sensing-aided beam prediction model for smart factory that can be transferred from given environments to a new environment. In particular, we first design a pre-training model that predicts the optimal beam by sensing the present environmental information. When encountering a new environment, it generally requires collecting a large amount of new training data to retrain the model, whose cost severely impedes the application of the designed pre-training model. Therefore, we next design a transfer learning strategy that fine-tunes the pre-trained model by limited labeled data of the new environment. Simulation results show that when the pre-trained model is fine-tuned by 30\% of labeled data from the new environment, the Top-10 beam prediction accuracy reaches 94\%. Moreover, compared with the way to completely re-training the prediction model, the amount of training data and the time cost of the proposed transfer learning strategy reduce 70\% and 75\% respectively.

Model-checking techniques have been extended to analyze quantum programs and communication protocols represented as quantum Markov chains, an extension of classical Markov chains. To specify qualitative temporal properties, a subspace-based quantum temporal logic is used, which is built on Birkhoff-von Neumann atomic propositions. These propositions determine whether a quantum state is within a subspace of the entire state space. In this paper, we propose the measurement-based linear-time temporal logic MLTL to check quantitative properties. MLTL builds upon classical linear-time temporal logic (LTL) but introduces quantum atomic propositions that reason about the probability distribution after measuring a quantum state. To facilitate verification, we extend the symbolic dynamics-based techniques for stochastic matrices described by Agrawal et al. (JACM 2015) to handle more general quantum linear operators (super-operators) through eigenvalue analysis. This extension enables the development of an efficient algorithm for approximately model checking a quantum Markov chain against an MLTL formula. To demonstrate the utility of our model-checking algorithm, we use it to simultaneously verify linear-time properties of both quantum and classical random walks. Through this verification, we confirm the previously established advantages discovered by Ambainis et al. (STOC 2001) of quantum walks over classical random walks and discover new phenomena unique to quantum walks.

The Low-Gain Avalanche Diode (LGAD) is a new silicon detector and holds wide application prospects in particle physics experiments due to its excellent timing resolution. The LGAD with a pixel size of 1.3 mm $\times$ 1.3 mm was used to construct a High Granularity Timing Detector (HGTD) in ATLAS experiments to solve the pile-up problem. Meanwhile, the Circular Electron Positron Collider (CEPC) also proposes detectors using the LGAD. However, pixel LGAD exhibits higher readout electronics density and cost, which somewhat limits the application of LGADs. To decrease the readout electronics density, the Institute of High Energy Physics (IHEP) of the Chinese Academy of Sciences has designed strip LGADs with larger areas. These strip LGADs are all 19 mm in length but with different widths of 1.0 mm, 0.5 mm, and 0.3 mm. This article provides a detailed introduction to the design parameters of these strip LGADs and tests their electrical characteristics, including leakage current, break-down voltage, depletion capacitance, etc. The timing resolution and signal-to-noise ratio of the three strip LGAD sensors were investigated using a beta source test system. The position resolution parallel to the strip direction was tested and analyzed for the first time using a pico-second laser test system. Tests have demonstrated that the timing resolution of strip LGADs can reach about 37.5 ps, and position resolution parallel to the strip direction is better than 1 mm.

Parameterised quantum circuits (PQCs) hold great promise for demonstrating quantum advantages in practical applications of quantum computation. Examples of successful applications include the variational quantum eigensolver, the quantum approximate optimisation algorithm, and quantum machine learning. However, before executing PQCs on real quantum devices, they undergo compilation and optimisation procedures. Given the inherent error-proneness of these processes, it becomes crucial to verify the equivalence between the original PQC and its compiled or optimised version. Unfortunately, most existing quantum circuit verifiers cannot directly handle parameterised quantum circuits; instead, they require parameter substitution to perform verification. In this paper, we address the critical challenge of equivalence checking for PQCs. We propose a novel compact representation for PQCs based on tensor decision diagrams. Leveraging this representation, we present an algorithm for verifying PQC equivalence without the need for instantiation. Our approach ensures both effectiveness and efficiency, as confirmed by experimental evaluations. The decision-diagram representations offer a powerful tool for analysing and verifying parameterised quantum circuits, bridging the gap between theoretical models and practical implementations.

This article explores how to drive intelligent iot monitoring and control through cloud computing and machine learning. As iot and the cloud continue to generate large and diverse amounts of data as sensor devices in the network, the collected data is sent to the cloud for statistical analysis, prediction, and data analysis to achieve business objectives. However, because the cloud computing model is limited by distance, it can be problematic in environments where the quality of the Internet connection is not ideal for critical operations. Therefore, edge computing, as a distributed computing architecture, moves the location of processing applications, data and services from the central node of the network to the logical edge node of the network to reduce the dependence on cloud processing and analysis of data, and achieve near-end data processing and analysis. The combination of iot and edge computing can reduce latency, improve efficiency, and enhance security, thereby driving the development of intelligent systems. The paper also introduces the development of iot monitoring and control technology, the application of edge computing in iot monitoring and control, and the role of machine learning in data analysis and fault detection. Finally, the application and effect of intelligent Internet of Things monitoring and control system in industry, agriculture, medical and other fields are demonstrated through practical cases and experimental studies.

Context: Software of different functional categories, such as text processing vs. networking, has different profiles in terms of metrics like security and updates. Using popularity to compare e.g. Java vs. Python libraries might give a skewed perspective, as the categories of the most popular software vary from one ecosystem to the next. How can one compare libraries datasets across software ecosystems, when not even the category names are uniform among them? Objective: We study how to generate a language-agnostic categorisation of software by functional purpose, that enables cross-ecosystem studies of libraries datasets. This provides the functional fingerprint information needed for software metrics comparisons. Method: We designed and implemented a human-guided protocol to categorise libraries from software ecosystems. Category names mirror PyPI Topic classifiers, but the protocol is generic and can be applied to any ecosystem. We demonstrate it by categorising 256 Java/Maven libraries with severe security vulnerabilities. Results: The protocol allows three or more people to categorise any number of libraries. The categorisation produced is functional-oriented and language-agnostic. The Java/Maven dataset demonstration resulted in a majority of Internet-oriented libraries, coherent with its selection by severe vulnerabilities. To allow replication and updates, we make the dataset and the protocol individual steps available as open data. Conclusions: Libraries categorisation by functional purpose is feasible with our protocol, which produced the fingerprint of a 256-libraries Java dataset. While this was labour intensive, humans excel in the required inference tasks, so full automation of the process is not envisioned. However, results can provide the ground truth needed for machine learning in large-scale cross-ecosystem empirical studies., Comment: Submitted to the journal "Information and Software Technology"

To enable large-scale and efficient deployment of artificial intelligence (AI), the combination of AI and edge computing has spawned Edge Intelligence, which leverages the computing and communication capabilities of end devices and edge servers to process data closer to where it is generated. A key technology for edge intelligence is the privacy-protecting machine learning paradigm known as Federated Learning (FL), which enables data owners to train models without having to transfer raw data to third-party servers. However, FL networks are expected to involve thousands of heterogeneous distributed devices. As a result, communication efficiency remains a key bottleneck. To reduce node failures and device exits, a Hierarchical Federated Learning (HFL) framework is proposed, where a designated cluster leader supports the data owner through intermediate model aggregation. Therefore, based on the improvement of edge server resource utilization, this paper can effectively make up for the limitation of cache capacity. In order to mitigate the impact of soft clicks on the quality of user experience (QoE), the authors model the user QoE as a comprehensive system cost. To solve the formulaic problem, the authors propose a decentralized caching algorithm with federated deep reinforcement learning (DRL) and federated learning (FL), where multiple agents learn and make decisions independently

Some high-z active galactic nuclei (AGNs) are found to reside in extreme star-forming galaxies, such as hyper-luminous infrared galaxies (HyLIRGs), with AGN-removed $L_{\rm{IR}}$ of $>10^{13} L_{\rm{\odot}}$. In this paper, we report NOEMA observations of six apparent starburst HyLIRGs associated with optical quasars at $z\sim2-3$ in the Stripe 82 field, to study their dust and molecular CO properties. Five out of the six candidates are detected with CO(4-3) or CO(5-4) emission, and four in 2mm dust continuum. Based on the linewidth-$L'_{\rm{CO(1-0)}}$ diagnostics, we find that four galaxies are likely unlensed or weakly lensed sources. The molecular gas mass is in the range of $\mu M_{\rm{H_2}} \sim0.8-9.7\times10^{10} M_{\odot}$ (with $\alpha = 0.8 M_{\odot} (\rm{K km s^{-1} pc^2})^{-1}$ and $\mu$ is the unknown possible gravitational magnification factor). We fit their SEDs, after including the observed 2mm fluxes and upper limits, and estimate their apparent (uncorrected for possible lensing effect) star formation rates ($\mu$SFRs) to be $\sim400-2500$ $M_{\rm{\odot}} \rm{yr^{-1}}$ with depletion time of $\sim20-110$ Myr. We notice interesting offsets, of $\sim10-40$ kpc spatially or $\sim1000-2000$ km s$^{-1}$ spectroscopically, between the optical quasar and the mm continuum or CO emissions. The observed velocity shift is likely related to the blueshifted broad-emission-line region of quasars, though mergers or recoiling black holes are also possible causes, which can explain the spatial offset and the high intrinsic SFRs in the HyLIRG-quasar systems., Comment: 22 pages, 6 figures, 7 tables. Accepted for publication in The Astrophysical Journal