Search

Matrix reordering permutes the rows and columns of a matrix to reveal meaningful visual patterns, such as blocks that represent clusters. A comprehensive collection of matrices, along with a scoring method for measuring the quality of visual patterns in these matrices, contributes to building a benchmark. This benchmark is essential for selecting or designing suitable reordering algorithms for specific tasks. In this paper, we build a matrix reordering benchmark, ReorderBench, with the goal of evaluating and improving matrix reordering techniques. This is achieved by generating a large set of representative and diverse matrices and scoring these matrices with a convolution- and entropy-based method. Our benchmark contains 2,835,000 binary matrices and 5,670,000 continuous matrices, each featuring one of four visual patterns: block, off-diagonal block, star, or band. We demonstrate the usefulness of ReorderBench through three main applications in matrix reordering: 1) evaluating different reordering algorithms, 2) creating a unified scoring model to measure the visual patterns in any matrix, and 3) developing a deep learning model for matrix reordering., Comment: Submitted to IEEE TVCG

Artwork analysis is important and fundamental skill for art appreciation, which could enrich personal aesthetic sensibility and facilitate the critical thinking ability. Understanding artworks is challenging due to its subjective nature, diverse interpretations, and complex visual elements, requiring expertise in art history, cultural background, and aesthetic theory. However, limited by the data collection and model ability, previous works for automatically analyzing artworks mainly focus on classification, retrieval, and other simple tasks, which is far from the goal of AI. To facilitate the research progress, in this paper, we step further to compose comprehensive analysis inspired by the remarkable perception and generation ability of large multimodal models. Specifically, we first propose a task of composing paragraph analysis for artworks, i.e., painting in this paper, only focusing on visual characteristics to formulate more comprehensive understanding of artworks. To support the research on formal analysis, we collect a large dataset PaintingForm, with about 19k painting images and 50k analysis paragraphs. We further introduce a superior large multimodal model for painting analysis composing, dubbed GalleryGPT, which is slightly modified and fine-tuned based on LLaVA architecture leveraging our collected data. We conduct formal analysis generation and zero-shot experiments across several datasets to assess the capacity of our model. The results show remarkable performance improvements comparing with powerful baseline LMMs, demonstrating its superb ability of art analysis and generalization. \textcolor{blue}{The codes and model are available at: https://github.com/steven640pixel/GalleryGPT., Comment: Accepted as Oral Presentation at ACM Multimedia 2024

We introduce rLLM (relationLLM), a PyTorch library designed for Relational Table Learning (RTL) with Large Language Models (LLMs). The core idea is to decompose state-of-the-art Graph Neural Networks, LLMs, and Table Neural Networks into standardized modules, to enable the fast construction of novel RTL-type models in a simple "combine, align, and co-train" manner. To illustrate the usage of rLLM, we introduce a simple RTL method named \textbf{BRIDGE}. Additionally, we present three novel relational tabular datasets (TML1M, TLF2K, and TACM12K) by enhancing classic datasets. We hope rLLM can serve as a useful and easy-to-use development framework for RTL-related tasks. Our code is available at: https://github.com/rllm-project/rllm.

Nowadays, it becomes a common practice to capture some data of sports games with devices such as GPS sensors and cameras and then use the data to perform various analyses on sports games, including tactics discovery, similar game retrieval, performance study, etc. While this practice has been conducted to many sports such as basketball and soccer, it remains largely unexplored on the billiards sports, which is mainly due to the lack of publicly available datasets. Motivated by this, we collect a dataset of billiards sports, which includes the layouts (i.e., locations) of billiards balls after performing break shots, called break shot layouts, the traces of the balls as a result of strikes (in the form of trajectories), and detailed statistics and performance indicators. We then study and develop techniques for three tasks on the collected dataset, including (1) prediction and (2) generation on the layouts data, and (3) similar billiards layout retrieval on the layouts data, which can serve different users such as coaches, players and fans. We conduct extensive experiments on the collected dataset and the results show that our methods perform effectively and efficiently., Comment: 27 pages; This paper is accepted by TKDD'2024

Head-mounted 360{\deg} displays and portable 360{\deg} cameras have significantly progressed, providing viewers a realistic and immersive experience. However, many omnidirectional videos have low frame rates that can lead to visual fatigue, and the prevailing plane frame interpolation methodologies are unsuitable for omnidirectional video interpolation because they are designed solely for traditional videos. This paper introduces the benchmark dataset, 360VFI, for Omnidirectional Video Frame Interpolation. We present a practical implementation that introduces a distortion prior from omnidirectional video into the network to modulate distortions. Specifically, we propose a pyramid distortion-sensitive feature extractor that uses the unique characteristics of equirectangular projection (ERP) format as prior information. Moreover, we devise a decoder that uses an affine transformation to further facilitate the synthesis of intermediate frames. 360VFI is the first dataset and benchmark that explores the challenge of Omnidirectional Video Frame Interpolation. Through our benchmark analysis, we present four different distortion condition scenes in the proposed 360VFI dataset to evaluate the challenges triggered by distortion during interpolation. Besides, experimental results demonstrate that Omnidirectional Video Interpolation can be effectively improved by modeling for omnidirectional distortion., Comment: This is a preprint version

Active clustering aims to boost the clustering performance by integrating human-annotated pairwise constraints through strategic querying. Conventional approaches with semi-supervised clustering schemes encounter high query costs when applied to large datasets with numerous classes. To address these limitations, we propose a novel Adaptive Active Aggregation and Splitting (A3S) framework, falling within the cluster-adjustment scheme in active clustering. A3S features strategic active clustering adjustment on the initial cluster result, which is obtained by an adaptive clustering algorithm. In particular, our cluster adjustment is inspired by the quantitative analysis of Normalized mutual information gain under the information theory framework and can provably improve the clustering quality. The proposed A3S framework significantly elevates the performance and scalability of active clustering. In extensive experiments across diverse real-world datasets, A3S achieves desired results with significantly fewer human queries compared with existing methods.

How to efficiently serve Large Language Models (LLMs) has become a pressing issue because of their huge computational cost in their autoregressive generation process. To mitigate computational costs, LLMs often employ the KV Cache technique to improve the generation speed. While improving the computational efficiency, the storage requirements of the KV cache are substantial, particularly in long-context scenarios, leading to significant memory consumption. Existing KV cache eviction methods often degrade the performance of LLMs in long-context scenarios due to the information loss introduced by eviction. In this paper, we propose a novel KV cache merging approach, called KVMerger, to achieve adaptive KV cache compression for long-context tasks without significant performance degradation under constrained memory budgets. Our approach is inspired by the intriguing observation that key states exhibit high similarity at the token level within a single sequence. To facilitate merging, we develop an effective yet straightforward merging set identification algorithm to identify suitable KV states for merging. Our merging set identification algorithm stimulates the second observation that KV cache sparsity, from similarity perspective, is independent of the dataset and remains persistent at the model level. Subsequently, we propose a Gaussian kernel weighted merging algorithm to selectively merge all states within each merging set. We conduct extensive experiments to demonstrate the effectiveness of KVMerger for long-context tasks under constrained memory budgets, applying it to models including Llama2-7B-chat and Llama2-13B-chat. Using the LongBench and ZeroScroll benchmarks, we compare our method with other KV cache compression techniques, including H2O and CaM, showing that our method achieves superior performance across tasks with both 50% and 35% KV cache budgets.

Zero-shot object counting (ZOC) aims to enumerate objects in images using only the names of object classes during testing, without the need for manual annotations. However, a critical challenge in current ZOC methods lies in their inability to identify high-quality exemplars effectively. This deficiency hampers scalability across diverse classes and undermines the development of strong visual associations between the identified classes and image content. To this end, we propose the Visual Association-based Zero-shot Object Counting (VA-Count) framework. VA-Count consists of an Exemplar Enhancement Module (EEM) and a Noise Suppression Module (NSM) that synergistically refine the process of class exemplar identification while minimizing the consequences of incorrect object identification. The EEM utilizes advanced vision-language pretaining models to discover potential exemplars, ensuring the framework's adaptability to various classes. Meanwhile, the NSM employs contrastive learning to differentiate between optimal and suboptimal exemplar pairs, reducing the negative effects of erroneous exemplars. VA-Count demonstrates its effectiveness and scalability in zero-shot contexts with superior performance on two object counting datasets.

Attention is a fundamental component behind the remarkable achievements of large language models (LLMs). However, our current understanding of the attention mechanism, especially regarding how attention distributions are established, remains limited. Inspired by recent studies that explore the presence of attention sink in the initial token, which receives disproportionately large attention scores despite their lack of semantic importance, this work delves deeper into this phenomenon. We aim to provide a more profound understanding of the existence of attention sinks within LLMs and to uncover ways to enhance the achievable accuracy of LLMs by directly optimizing the attention distributions, without the need for weight finetuning. Specifically, this work begins with comprehensive visualizations of the attention distributions in LLMs during inference across various inputs and tasks. Based on these visualizations, to the best of our knowledge, we are the first to discover that (1) attention sinks occur not only at the start of sequences but also within later tokens of the input, and (2) not all attention sinks have a positive impact on the achievable accuracy of LLMs. Building upon our findings, we propose a training-free Attention Calibration Technique (ACT) that automatically optimizes the attention distributions on the fly during inference in an input-adaptive manner. Extensive experiments validate that ACT consistently enhances the accuracy of various LLMs across different applications. Specifically, ACT achieves an average improvement of up to 7.30% in accuracy across different datasets when applied to Llama-30B. Our code is available at https://github.com/GATECH-EIC/ACT.

Efficient adaption of large language models (LLMs) on edge devices is essential for applications requiring continuous and privacy-preserving adaptation and inference. However, existing tuning techniques fall short because of the high computation and memory overheads. To this end, we introduce a computation- and memory-efficient LLM tuning framework, called Edge-LLM, to facilitate affordable and effective LLM adaptation on edge devices. Specifically, Edge-LLM features three core components: (1) a layer-wise unified compression (LUC) technique to reduce the computation overhead by generating layer-wise pruning sparsity and quantization bit-width policies, (2) an adaptive layer tuning and voting scheme to reduce the memory overhead by reducing the backpropagation depth, and (3) a complementary hardware scheduling strategy to handle the irregular computation patterns introduced by LUC and adaptive layer tuning, thereby achieving efficient computation and data movements. Extensive experiments demonstrate that Edge-LLM achieves a 2.92x speed up and a 4x memory overhead reduction as compared to vanilla tuning methods with comparable task accuracy. Our code is available at https://github.com/GATECH-EIC/Edge-LLM

Few-shot object detection~(FSOD), which aims to detect novel objects with limited annotated instances, has made significant progress in recent years. However, existing methods still suffer from biased representations, especially for novel classes in extremely low-shot scenarios. During fine-tuning, a novel class may exploit knowledge from similar base classes to construct its own feature distribution, leading to classification confusion and performance degradation. To address these challenges, we propose a fine-tuning based FSOD framework that utilizes semantic embeddings for better detection. In our proposed method, we align the visual features with class name embeddings and replace the linear classifier with our semantic similarity classifier. Our method trains each region proposal to converge to the corresponding class embedding. Furthermore, we introduce a multimodal feature fusion to augment the vision-language communication, enabling a novel class to draw support explicitly from well-trained similar base classes. To prevent class confusion, we propose a semantic-aware max-margin loss, which adaptively applies a margin beyond similar classes. As a result, our method allows each novel class to construct a compact feature space without being confused with similar base classes. Extensive experiments on Pascal VOC and MS COCO demonstrate the superiority of our method., Comment: Accepted by ICIP 2024

Blockchain technology has seen tremendous development over the past few years. Despite the emergence of numerous blockchain systems, they all suffer from various limitations, which can all be attributed to the fundamental issue posed by the DCS trilemma. In light of this, this work introduces a novel private blockchain system named DCS Chain. The core idea is to quantify the DCS metrics and dynamically adjust the blockchain's performance across these three dimensions, to achieve theoretically optimal system performance. Overall, our system provides a comprehensive suite of blockchain essentials, including DCS quantification, consensus protocol adjustment, and communication network simulation.

The intersection of physics-based vision and deep learning presents an exciting frontier for advancing computer vision technologies. By leveraging the principles of physics to inform and enhance deep learning models, we can develop more robust and accurate vision systems. Physics-based vision aims to invert the processes to recover scene properties such as shape, reflectance, light distribution, and medium properties from images. In recent years, deep learning has shown promising improvements for various vision tasks, and when combined with physics-based vision, these approaches can enhance the robustness and accuracy of vision systems. This technical report summarizes the outcomes of the Physics-Based Vision Meets Deep Learning (PBDL) 2024 challenge, held in CVPR 2024 workshop. The challenge consisted of eight tracks, focusing on Low-Light Enhancement and Detection as well as High Dynamic Range (HDR) Imaging. This report details the objectives, methodologies, and results of each track, highlighting the top-performing solutions and their innovative approaches., Comment: CVPR 2024 PBDL Challenges: https://pbdl-ws.github.io/pbdl2024/challenge/index.html

In this work we try to search for signals generated by ultra-heavy dark matter at the Large High Altitude Air Shower Observatory (LHAASO) data. We look for possible gamma-ray by dark matter annihilation or decay from 16 dwarf spheroidal galaxies in the field of view of LHAASO. Dwarf spheroidal galaxies are among the most promising targets for indirect detection of dark matter which have low fluxes of astrophysical $\gamma$-ray background while large amount of dark matter. By analyzing more than 700 days observational data at LHAASO, no significant dark matter signal from 1 TeV to 1 EeV is detected. Accordingly we derive the most stringent constraints on the ultra-heavy dark matter annihilation cross-section up to EeV. The constraints on the lifetime of dark matter in decay mode are also derived., Comment: 17 pages, 12 figures, accepted by PRL

Autoregressive Large Language Models (LLMs) have achieved impressive performance in language tasks but face two significant bottlenecks: (1) quadratic complexity in the attention module as the number of tokens increases, and (2) limited efficiency due to the sequential processing nature of autoregressive LLMs during generation. While linear attention and speculative decoding offer potential solutions, their applicability and synergistic potential for enhancing autoregressive LLMs remain uncertain. We conduct the first comprehensive study on the efficacy of existing linear attention methods for autoregressive LLMs, integrating them with speculative decoding. We introduce an augmentation technique for linear attention that ensures compatibility with speculative decoding, enabling more efficient training and serving of LLMs. Extensive experiments and ablation studies involving seven existing linear attention models and five encoder/decoder-based LLMs consistently validate the effectiveness of our augmented linearized LLMs. Notably, our approach achieves up to a 6.67 reduction in perplexity on the LLaMA model and up to a 2$\times$ speedup during generation compared to prior linear attention methods. Codes and models are available at https://github.com/GATECH-EIC/Linearized-LLM., Comment: Accepted by ICML 2024; 17 pages; 10 figures; 16 tables

The High Energy Photon Source (HEPS) represents a fourth-generation light source. This facility has made unprecedented advancements in accelerator technology, necessitating the development of new detectors to satisfy physical requirements such as single-photon resolution, large dynamic range, and high frame rates. Since 2016, the Institute of High Energy Physics has introduced the first user-experimental hybrid pixel detector, progressing to the fourth-generation million-pixel detector designed for challenging conditions, with the dual-threshold single-photon detector HEPS-Beijing PIXel (HEPS-BPIX) set as the next-generation target. HEPS-BPIX will employ the entirely new Application-Specific Integrated Circuit (ASIC) BP40 for pixel information readout. Data flow will be managed and controlled through readout electronics based on a two-tier Field-Programmable Gate Array (FPGA) system: the Front-End Electronics (FEE) and the Input-Output Board (IOB) handle the fan-out for 12 ASICs, and the u4FCP is tasked with processing serial data on high-speed links, transferring pixel-level data to the back-end RTM and uTCA chassis, or independently outputting through a network port, enabling remote control of the entire detector. The new HEPS-BPIX firmware has undergone a comprehensive redesign and update to meet the electronic characteristics of the new chip and to improve the overall performance of the detector. We provide an overview of the core subunits of HEPS-BPIX, emphasizing the readout system, evaluating the new hardware and firmware, and highlighting some of its innovative features and characteristics.

Federated Learning (FL) stands to gain significant advantages from collaboratively training capacity-heterogeneous models, enabling the utilization of private data and computing power from low-capacity devices. However, the focus on personalizing capacity-heterogeneous models based on client-specific data has been limited, resulting in suboptimal local model utility, particularly for low-capacity clients. The heterogeneity in both data and device capacity poses two key challenges for model personalization: 1) accurately retaining necessary knowledge embedded within reduced submodels for each client, and 2) effectively sharing knowledge through aggregating size-varying parameters. To this end, we introduce Pa3dFL, a novel framework designed to enhance local model performance by decoupling and selectively sharing knowledge among capacity-heterogeneous models. First, we decompose each layer of the model into general and personal parameters. Then, we maintain uniform sizes for the general parameters across clients and aggregate them through direct averaging. Subsequently, we employ a hyper-network to generate size-varying personal parameters for clients using learnable embeddings. Finally, we facilitate the implicit aggregation of personal parameters by aggregating client embeddings through a self-attention module. We conducted extensive experiments on three datasets to evaluate the effectiveness of Pa3dFL. Our findings indicate that Pa3dFL consistently outperforms baseline methods across various heterogeneity settings. Moreover, Pa3dFL demonstrates competitive communication and computation efficiency compared to baseline approaches, highlighting its practicality and adaptability in adverse system conditions.

Collaborative fairness stands as an essential element in federated learning to encourage client participation by equitably distributing rewards based on individual contributions. Existing methods primarily focus on adjusting gradient allocations among clients to achieve collaborative fairness. However, they frequently overlook crucial factors such as maintaining consistency across local models and catering to the diverse requirements of high-contributing clients. This oversight inevitably decreases both fairness and model accuracy in practice. To address these issues, we propose FedSAC, a novel Federated learning framework with dynamic Submodel Allocation for Collaborative fairness, backed by a theoretical convergence guarantee. First, we present the concept of "bounded collaborative fairness (BCF)", which ensures fairness by tailoring rewards to individual clients based on their contributions. Second, to implement the BCF, we design a submodel allocation module with a theoretical guarantee of fairness. This module incentivizes high-contributing clients with high-performance submodels containing a diverse range of crucial neurons, thereby preserving consistency across local models. Third, we further develop a dynamic aggregation module to adaptively aggregate submodels, ensuring the equitable treatment of low-frequency neurons and consequently enhancing overall model accuracy. Extensive experiments conducted on three public benchmarks demonstrate that FedSAC outperforms all baseline methods in both fairness and model accuracy. We see this work as a significant step towards incentivizing broader client participation in federated learning. The source code is available at https://github.com/wangzihuixmu/FedSAC., Comment: Accepted by KDD'24

The SkatingVerse Workshop & Challenge aims to encourage research in developing novel and accurate methods for human action understanding. The SkatingVerse dataset used for the SkatingVerse Challenge has been publicly released. There are two subsets in the dataset, i.e., the training subset and testing subset. The training subsets consists of 19,993 RGB video sequences, and the testing subsets consists of 8,586 RGB video sequences. Around 10 participating teams from the globe competed in the SkatingVerse Challenge. In this paper, we provide a brief summary of the SkatingVerse Workshop & Challenge including brief introductions to the top three methods. The submission leaderboard will be reopened for researchers that are interested in the human action understanding challenge. The benchmark dataset and other information can be found at: https://skatingverse.github.io/.

We explore the bound neutrons decay into invisible particles (e.g., $n\rightarrow 3 \nu$ or $nn \rightarrow 2 \nu$) in the JUNO liquid scintillator detector. The invisible decay includes two decay modes: $ n \rightarrow { inv} $ and $ nn \rightarrow { inv} $. The invisible decays of $s$-shell neutrons in $^{12}{\rm C}$ will leave a highly excited residual nucleus. Subsequently, some de-excitation modes of the excited residual nuclei can produce a time- and space-correlated triple coincidence signal in the JUNO detector. Based on a full Monte Carlo simulation informed with the latest available data, we estimate all backgrounds, including inverse beta decay events of the reactor antineutrino $\bar{\nu}_e$, natural radioactivity, cosmogenic isotopes and neutral current interactions of atmospheric neutrinos. Pulse shape discrimination and multivariate analysis techniques are employed to further suppress backgrounds. With two years of exposure, JUNO is expected to give an order of magnitude improvement compared to the current best limits. After 10 years of data taking, the JUNO expected sensitivities at a 90% confidence level are $\tau/B( n \rightarrow { inv} ) > 5.0 \times 10^{31} \, {\rm yr}$ and $\tau/B( nn \rightarrow { inv} ) > 1.4 \times 10^{32} \, {\rm yr}$., Comment: 28 pages, 7 figures, 4 tables

Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs) by retrieving relevant memories from an external database. However, existing RAG methods typically organize all memories in a whole database, potentially limiting focus on crucial memories and introducing noise. In this paper, we introduce a multiple partition paradigm for RAG (called M-RAG), where each database partition serves as a basic unit for RAG execution. Based on this paradigm, we propose a novel framework that leverages LLMs with Multi-Agent Reinforcement Learning to optimize different language generation tasks explicitly. Through comprehensive experiments conducted on seven datasets, spanning three language generation tasks and involving three distinct language model architectures, we confirm that M-RAG consistently outperforms various baseline methods, achieving improvements of 11%, 8%, and 12% for text summarization, machine translation, and dialogue generation, respectively., Comment: This paper has been accepted by ACL 2024

Quantum spin liquid has massive many spin entanglement in the ground state, we can evaluate it by the entanglement entropy, but the latter can not be observed directly by experiment. In this manuscript, we try to characterize its topological properties by the geometric phase. However the usual adiabatic or non-adiabatic geometric phase can not appear in the density matrix of entanglement entropy, so we extend it to the sub-geometric phase which can exist in the density matrix and have influence on the entanglement entropy, spin correlation function as well as other physical observable. We will demonstrate that the imaginary part of sub-geometric phase will deviate the resonance peak by an amount concerning with this phase and affect the energy level crossing, while the real part of sub-geometric phase will determine the stability of initial state, it may provide a complement on the selection rule of quantum transition., Comment: 19pages

The KM2A is the largest sub-array of the Large High Altitude Air Shower Observatory (LHAASO). It consists of 5216 electromagnetic particle detectors (EDs) and 1188 muon detectors (MDs). The data recorded by the EDs and MDs are used to reconstruct primary information of cosmic ray and gamma-ray showers. This information is used for physical analysis in gamma-ray astronomy and cosmic ray physics. To ensure the reliability of the LHAASO-KM2A data, a three-level quality control system has been established. It is used to monitor the status of detector units, stability of reconstructed parameters and the performance of the array based on observations of the Crab Nebula and Moon shadow. This paper will introduce the control system and its application on the LHAASO-KM2A data collected from August 2021 to July 2023. During this period, the pointing and angular resolution of the array were stable. From the observations of the Moon shadow and Crab Nebula, the results achieved using the two methods are consistent with each other. According to the observation of the Crab Nebula at energies from 25 TeV to 100 TeV, the time averaged pointing errors are estimated to be $-0.003^{\circ} \pm 0.005^{\circ}$ and $0.001^{\circ} \pm 0.006^{\circ}$ in the R.A. and Dec directions, respectively., Comment: 15 pages, 9 figures

Based on the adiabatic geometric phase concerning with density matrix[1] , we extend it to the sub-geometric phase in the non-adiabatic case. It is found that whatever the real part or imaginary part of the sub-geometric phase can play an important role in quantum transition. The imaginary part of sub-geometric phase can deviate the resonance peak in the quantum transition, which may bring modification on the level crossing, while the real part of sub-geometric phase will determine the stability of initial state according to the linear stability analysis theory, which can be regarded as somewhat complement on the selection rule of quantum transition. Finally, we illustrate them by two examples: one is the system with time-dependent perturbation, the other is a two-level system. It indicates that both the real and imaginary parts of sub-geometric phase have influence on quantum transition., Comment: 0 figures, 16pages

The first source catalog of Large High Altitude Air Shower Observatory reported the detection of a very-high-energy gamma ray source, 1LHAASO J1219+2915. In this paper a further detailed study of the spectral and temporal behavior of this point-like source have been carried. The best-fit position of the TeV source ($\rm{RA}=185.05^{\circ}\pm0.04^{\circ}$, $\rm{Dec}=29.25^{\circ}\pm0.03^{\circ}$) is compatible with NGC 4278 within $\sim0.03$ degree. Variation analysis shows an indication of the variability at a few months level in the TeV band, which is consistent with low frequency observations. Based on these observations, we report the detection of TeV $\gamma$-ray emissions from this low-luminosity AGN NGC 4278. The observations by LHAASO-WCDA during active period has a significance level of 8.8\,$\sigma$ with best-fit photon spectral index $\varGamma=2.56\pm0.14$ and a flux $f_{1-10\,\rm{TeV}}=(7.0\pm1.1_{\rm{sta}}\pm0.35_{\rm{syst}})\times10^{-13}\,\rm{photons\,cm^{-2}\,s^{-1}}$, or approximately $5\%$ of the Crab Nebula. The discovery of VHE from NGC 4278 indicates that the compact, weak radio jet can efficiently accelerate particles and emit TeV photons., Comment: 11 pages, 5 figures

Large Language Models (LLMs) have achieved remarkable progress in code generation. It now becomes crucial to identify whether the code is AI-generated and to determine the specific model used, particularly for purposes such as protecting Intellectual Property (IP) in industry and preventing cheating in programming exercises. To this end, several attempts have been made to insert watermarks into machine-generated code. However, existing approaches are limited to inserting only a single bit of information or overly depending on particular code patterns. In this paper, we introduce CodeIP, a novel multi-bit watermarking technique that embeds additional information to preserve crucial provenance details, such as the vendor ID of an LLM, thereby safeguarding the IPs of LLMs in code generation. Furthermore, to ensure the syntactical correctness of the generated code, we propose constraining the sampling process for predicting the next token by training a type predictor. Experiments conducted on a real-world dataset across five programming languages demonstrate the effectiveness of CodeIP in watermarking LLMs for code generation while maintaining the syntactical correctness of code., Comment: 15 pages, 13 figures

Text to Motion aims to generate human motions from texts. Existing settings assume that texts include action labels, which limits flexibility in practical scenarios. This paper extends this task with a more realistic assumption that the texts are arbitrary. Specifically, in our setting, arbitrary texts include existing action texts composed of action labels and introduce scene texts without explicit action labels. To address this practical issue, we extend the action texts in the HUMANML3D dataset by incorporating additional scene texts, thereby creating a new dataset, HUMANML3D++. Concurrently, we propose a simple framework that extracts action representations from arbitrary texts using a Large Language Model (LLM) and subsequently generates motions. Furthermore, we enhance the existing evaluation methodologies to address their inadequacies. Extensive experiments are conducted under different application scenarios to validate the effectiveness of the proposed framework on existing and proposed datasets. The results indicate that Text to Motion in this realistic setting is very challenging, fostering new research in this practical direction. Our dataset and code will be released., Comment: Updated errors in author information

In this paper, we plan to show an eigenvalue algorithm for block Hessenberg matrices by using the idea of non-commutative integrable systems and matrix-valued orthogonal polynomials. We introduce adjacent families of matrix-valued $\theta$-deformed bi-orthogonal polynomials, and derive corresponding discrete non-commutative hungry Toda lattice from discrete spectral transformations for polynomials. It is shown that this discrete system can be used as a pre-precessing algorithm for block Hessenberg matrices. Besides, some convergence analysis and numerical examples of this algorithm are presented., Comment: 24 pages, 2 figures. Comments are welcome

The Langevin sampling method relies on an accurate score matching while the existing massive multiple-input multiple output (MIMO) Langevin detection involves an inevitable singular value decomposition (SVD) to calculate the posterior score. In this work, a massive MIMO sampling detection strategy that leverages the denoising diffusion model is proposed to narrow the gap between the given iterative detector and the maximum likelihood (ML) detection in an SVD-free manner. Specifically, the proposed score-based sampling detection strategy, denoted as approximate diffusion detection (ADD), is applicable to a wide range of iterative detection methods, and therefore entails a considerable potential in their performance improvement by multiple sampling attempts. On the other hand, the ADD scheme manages to bypass the channel SVD by introducing a reliable iterative detector to produce a sample from the approximate posterior, so that further Langevin sampling is tractable. Customized by the conjugated gradient descent algorithm as an instance, the proposed sampling scheme outperforms the existing score-based detector in terms of a better complexity-performance trade-off., Comment: 6 pages, 4 figures, already accepted by the 20th International Wireless Communications and Mobile Computing Conference (IWCMC 2024)

Adapting Foundation Models (FMs) for downstream tasks through Federated Learning (FL) emerges a promising strategy for protecting data privacy and valuable FMs. Existing methods fine-tune FM by allocating sub-FM to clients in FL, however, leading to suboptimal performance due to insufficient tuning and inevitable error accumulations of gradients. In this paper, we propose Federated Proxy Fine-Tuning (FedPFT), a novel method enhancing FMs adaptation in downstream tasks through FL by two key modules. First, the sub-FM construction module employs a layer-wise compression approach, facilitating comprehensive FM fine-tuning across all layers by emphasizing those crucial neurons. Second, the sub-FM alignment module conducts a two-step distillations-layer-level and neuron-level-before and during FL fine-tuning respectively, to reduce error of gradient by accurately aligning sub-FM with FM under theoretical guarantees. Experimental results on seven commonly used datasets (i.e., four text and three vision) demonstrate the superiority of FedPFT., Comment: Accepted by IJCAI'24

Ascribed as $\rho\rho$ and $K^* \bar{K}^*$ molecular states, respectively, iso-scalar $f_0(1500)$ and $f_0(1710)$ states are expected to have iso-vector partners, potentially identified as $a_0(1450)$ and $a_0(1710)$. The predicted dominant decay modes for these two $a_0$ resonances are $a_0(1450) \to \omega\pi\pi$ and $a_0(1710) \to \omega\pi\pi,\,\phi\pi\pi$. We estimate cross sections for two-photon production of these four resonances within the hadronic molecular picture, and demonstrate that SuperKEKB's luminosity is sufficient for their observation and more precise parameter measurements.

KM2A is one of the main sub-arrays of LHAASO, working on gamma ray astronomy and cosmic ray physics at energies above 10 TeV. Detector simulation is the important foundation for estimating detector performance and data analysis. It is a big challenge to simulate the KM2A detector in the framework of Geant4 due to the need to track numerous photons from a large number of detector units (>6000) with large altitude difference (30 m) and huge coverage (1.3 km^2). In this paper, the design of the KM2A simulation code G4KM2A based on Geant4 is introduced. The process of G4KM2A is optimized mainly in memory consumption to avoid memory overffow. Some simpliffcations are used to signiffcantly speed up the execution of G4KM2A. The running time is reduced by at least 30 times compared to full detector simulation. The particle distributions and the core/angle resolution comparison between simulation and experimental data of the full KM2A array are also presented, which show good agreement.

Parameter-efficient finetuning (PEFT) methods seek to adapt large neural models via updates to a small number of weights. However, much prior interpretability work has shown that representations encode rich semantic information, suggesting that editing representations might be a more powerful alternative. We pursue this hypothesis by developing a family of Representation Finetuning (ReFT) methods. ReFT methods operate on a frozen base model and learn task-specific interventions on hidden representations. We define a strong instance of the ReFT family, Low-rank Linear Subspace ReFT (LoReFT), and we identify an ablation of this method that trades some performance for increased efficiency. Both are drop-in replacements for existing PEFTs and learn interventions that are 15x--65x more parameter-efficient than LoRA. We showcase LoReFT on eight commonsense reasoning tasks, four arithmetic reasoning tasks, instruction-tuning, and GLUE. In all these evaluations, our ReFTs deliver the best balance of efficiency and performance, and almost always outperform state-of-the-art PEFTs. We release a generic ReFT training library publicly at https://github.com/stanfordnlp/pyreft., Comment: preprint

In this paper, we consider robust joint access point (AP) clustering and beamforming design with imperfect channel state information (CSI) in cell-free systems. Specifically, we jointly optimize AP clustering and beamforming with imperfect CSI to simultaneously maximize the worst-case sum rate and minimize the number of AP clustering under power constraint and the sparsity constraint of AP clustering. By transformations, the semi-infinite constraints caused by the imperfect CSI are converted into more tractable forms for facilitating a computationally efficient unsupervised deep learning algorithm. In addition, to further reduce the computational complexity, a computationally effective unsupervised deep learning algorithm is proposed to implement robust joint AP clustering and beamforming design with imperfect CSI in cell-free systems. Numerical results demonstrate that the proposed unsupervised deep learning algorithm achieves a higher worst-case sum rate under a smaller number of AP clustering with computational efficiency., Comment: 13 pages, 11 figures. The paper has been submitted to IEEE Transactions on Wireless Communications

Byzantine consensus protocols are essential in blockchain technology. The widely recognized HotStuff protocol uses cryptographic measures for efficient view changes and reduced communication complexity. Recently, the main authors of HotStuff introduced an advanced iteration named HotStuff-2. This paper aims to compare the principles and analyze the effectiveness of both protocols, hoping to depict their key differences and assess the potential enhancements offered by HotStuff-2.

Large Language Models (LLMs) have shown remarkable progress in automated code generation. Yet, LLM-generated code may contain errors in API usage, class, data structure, or missing project-specific information. As much of this project-specific context cannot fit into the prompts of LLMs, we must find ways to allow the model to explore the project-level code context. We present CoCoGen, a new code generation approach that uses compiler feedback to improve the LLM-generated code. CoCoGen first leverages static analysis to identify mismatches between the generated code and the project's context. It then iteratively aligns and fixes the identified errors using information extracted from the code repository. We integrate CoCoGen with two representative LLMs, i.e., GPT-3.5-Turbo and Code Llama (13B), and apply it to Python code generation. Experimental results show that CoCoGen significantly improves the vanilla LLMs by over 80% in generating code dependent on the project context and consistently outperforms the existing retrieval-based code generation baselines.

We present the measurements of all-particle energy spectrum and mean logarithmic mass of cosmic rays in the energy range of 0.3-30 PeV using data collected from LHAASO-KM2A between September 2021 and December 2022, which is based on a nearly composition-independent energy reconstruction method, achieving unprecedented accuracy. Our analysis reveals the position of the knee at $3.67 \pm 0.05 \pm 0.15$ PeV. Below the knee, the spectral index is found to be -$2.7413 \pm 0.0004 \pm 0.0050$, while above the knee, it is -$3.128 \pm 0.005 \pm 0.027$, with the sharpness of the transition measured with a statistical error of 2%. The mean logarithmic mass of cosmic rays is almost heavier than helium in the whole measured energy range. It decreases from 1.7 at 0.3 PeV to 1.3 at 3 PeV, representing a 24% decline following a power law with an index of -$0.1200 \pm 0.0003 \pm 0.0341$. This is equivalent to an increase in abundance of light components. Above the knee, the mean logarithmic mass exhibits a power law trend towards heavier components, which is reversal to the behavior observed in the all-particle energy spectrum. Additionally, the knee position and the change in power-law index are approximately the same. These findings suggest that the knee observed in the all-particle spectrum corresponds to the knee of the light component, rather than the medium-heavy components., Comment: 8 pages, 3 figures

Isolated interacting quantum systems generally thermalize, yet there are several counterexamples for the breakdown of ergodicity, such as many-body localization and quantum scars. Recently, ergodicity breaking has been observed in systems subjected to linear potentials, termed Stark many-body localization. This phenomenon is closely associated with Hilbert-space fragmentation, characterized by a strong dependence of dynamics on initial conditions. Here, we experimentally explore initial-state dependent dynamics using a ladder-type superconducting processor with up to 24 qubits, which enables precise control of the qubit frequency and initial state preparation. In systems with linear potentials, we observe distinct non-equilibrium dynamics for initial states with the same quantum numbers and energy, but with varying domain wall numbers. This distinction becomes increasingly pronounced as the system size grows, in contrast with disordered interacting systems. Our results provide convincing experimental evidence of the fragmentation in Stark systems, enriching our understanding of the weak breakdown of ergodicity., Comment: main text: 7 pages, 4 figures; supplementary: 13 pages, 14 figures

Interventions on model-internal states are fundamental operations in many areas of AI, including model editing, steering, robustness, and interpretability. To facilitate such research, we introduce $\textbf{pyvene}$, an open-source Python library that supports customizable interventions on a range of different PyTorch modules. $\textbf{pyvene}$ supports complex intervention schemes with an intuitive configuration format, and its interventions can be static or include trainable parameters. We show how $\textbf{pyvene}$ provides a unified and extensible framework for performing interventions on neural models and sharing the intervened upon models with others. We illustrate the power of the library via interpretability analyses using causal abstraction and knowledge localization. We publish our library through Python Package Index (PyPI) and provide code, documentation, and tutorials at https://github.com/stanfordnlp/pyvene., Comment: 8 pages, 3 figures

As live streaming services skyrocket, Crowdsourced Cloud-edge service Platforms (CCPs) have surfaced as pivotal intermediaries catering to the mounting demand. Despite the role of stream scheduling to CCPs' Quality of Service (QoS) and throughput, conventional optimization strategies struggle to enhancing CCPs' revenue, primarily due to the intricate relationship between resource utilization and revenue. Additionally, the substantial scale of CCPs magnifies the difficulties of time-intensive scheduling. To tackle these challenges, we propose Seer, a proactive revenue-aware scheduling system for live streaming services in CCPs. The design of Seer is motivated by meticulous measurements of real-world CCPs environments, which allows us to achieve accurate revenue modeling and overcome three key obstacles that hinder the integration of prediction and optimal scheduling. Utilizing an innovative Pre-schedule-Execute-Re-schedule paradigm and flexible scheduling modes, Seer achieves efficient revenue-optimized scheduling in CCPs. Extensive evaluations demonstrate Seer's superiority over competitors in terms of revenue, utilization, and anomaly penalty mitigation, boosting CCPs revenue by 147% and expediting scheduling $3.4 \times$ faster.

In the rapidly evolving landscape of information retrieval, search engines strive to provide more personalized and relevant results to users. Query suggestion systems play a crucial role in achieving this goal by assisting users in formulating effective queries. However, existing query suggestion systems mainly rely on textual inputs, potentially limiting user search experiences for querying images. In this paper, we introduce a novel Multimodal Query Suggestion (MMQS) task, which aims to generate query suggestions based on user query images to improve the intentionality and diversity of search results. We present the RL4Sugg framework, leveraging the power of Large Language Models (LLMs) with Multi-Agent Reinforcement Learning from Human Feedback to optimize the generation process. Through comprehensive experiments, we validate the effectiveness of RL4Sugg, demonstrating a 18% improvement compared to the best existing approach. Moreover, the MMQS has been transferred into real-world search engine products, which yield enhanced user engagement. Our research advances query suggestion systems and provides a new perspective on multimodal information retrieval., Comment: This paper has been accepted by WWW 2024

Masked autoencoding and generative pretraining have achieved remarkable success in computer vision and natural language processing, and more recently, they have been extended to the point cloud domain. Nevertheless, existing point cloud models suffer from the issue of information leakage due to the pre-sampling of center points, which leads to trivial proxy tasks for the models. These approaches primarily focus on local feature reconstruction, limiting their ability to capture global patterns within point clouds. In this paper, we argue that the reduced difficulty of pretext tasks hampers the model's capacity to learn expressive representations. To address these limitations, we introduce a novel solution called the Differentiable Center Sampling Network (DCS-Net). It tackles the information leakage problem by incorporating both global feature reconstruction and local feature reconstruction as non-trivial proxy tasks, enabling simultaneous learning of both the global and local patterns within point cloud. Experimental results demonstrate that our method enhances the expressive capacity of existing point cloud models and effectively addresses the issue of information leakage.

Coreset selection seeks to choose a subset of crucial training samples for efficient learning. It has gained traction in deep learning, particularly with the surge in training dataset sizes. Sample selection hinges on two main aspects: a sample's representation in enhancing performance and the role of sample diversity in averting overfitting. Existing methods typically measure both the representation and diversity of data based on similarity metrics, such as L2-norm. They have capably tackled representation via distribution matching guided by the similarities of features, gradients, or other information between data. However, the results of effectively diverse sample selection are mired in sub-optimality. This is because the similarity metrics usually simply aggregate dimension similarities without acknowledging disparities among the dimensions that significantly contribute to the final similarity. As a result, they fall short of adequately capturing diversity. To address this, we propose a feature-based diversity constraint, compelling the chosen subset to exhibit maximum diversity. Our key lies in the introduction of a novel Contributing Dimension Structure (CDS) metric. Different from similarity metrics that measure the overall similarity of high-dimensional features, our CDS metric considers not only the reduction of redundancy in feature dimensions, but also the difference between dimensions that contribute significantly to the final similarity. We reveal that existing methods tend to favor samples with similar CDS, leading to a reduced variety of CDS types within the coreset and subsequently hindering model performance. In response, we enhance the performance of five classical selection methods by integrating the CDS constraint. Our experiments on three datasets demonstrate the general effectiveness of the proposed method in boosting existing methods., Comment: 13 pages,11 figures, to be published in AAAI2024

To reduce the complexity of software, Developers manually simplify program (known as developer-induced program simplification in this paper) to reduce its code size yet preserving its functionality but manual simplification is time-consuming and error-prone. To reduce manual effort, rule-based approaches (e.g., refactoring) and deletion-based approaches (e.g., delta debugging) can be potentially applied to automate developer-induced program simplification. However, as there is little study on how developers simplify programs in Open-source Software (OSS) projects, it is unclear whether these approaches can be effectively used for developer-induced program simplification. Hence, we present the first study of developer-induced program simplification in OSS projects, focusing on the types of program transformations used, the motivations behind simplifications, and the set of program transformations covered by existing refactoring types. Our study of 382 pull requests from 296 projects reveals that there exist gaps in applying existing approaches for automating developer-induced program simplification. and outlines the criteria for designing automatic program simplification techniques. Inspired by our study and to reduce the manual effort in developer-induced program simplification, we propose SimpT5, a tool that can automatically produce simplified programs (semantically-equivalent programs with reduced source lines of code). SimpT5 is trained based on our collected dataset of 92,485 simplified programs with two heuristics: (1) simplified line localization that encodes lines changed in simplified programs, and (2)checkers that measure the quality of generated programs. Our evaluation shows that SimpT5 are more effective than prior approaches in automating developer-induced program simplification.