Search

The creation of diverse and realistic driving scenarios has become essential to enhance perception and planning capabilities of the autonomous driving system. However, generating long-duration, surround-view consistent driving videos remains a significant challenge. To address this, we present UniMLVG, a unified framework designed to generate extended street multi-perspective videos under precise control. By integrating single- and multi-view driving videos into the training data, our approach updates cross-frame and cross-view modules across three stages with different training objectives, substantially boosting the diversity and quality of generated visual content. Additionally, we employ the explicit viewpoint modeling in multi-view video generation to effectively improve motion transition consistency. Capable of handling various input reference formats (e.g., text, images, or video), our UniMLVG generates high-quality multi-view videos according to the corresponding condition constraints such as 3D bounding boxes or frame-level text descriptions. Compared to the best models with similar capabilities, our framework achieves improvements of 21.4% in FID and 36.5% in FVD.

Crystals exhibiting glass-like and low lattice thermal conductivity ($\kappa_{\rm L}$) are not only scientifically intriguing but also practically valuable in various applications, including thermal barrier coatings, thermoelectric energy conversion, and thermal management. However, such unusual $\kappa_{\rm L}$ are typically observed only in compounds containing heavy elements, with large unit cells, or at high temperatures, primarily due to significant anharmonicity. In this study, we utilize chemical bonding principles to weaken the Ag-Ag bonds within the Ag$_6$ octahedron by introducing a ligand in the bridge position. Additionally, the weak Ag-chalcogen bonds, arising from fully filled $p$-$d$ antibonding orbitals, provide an avenue to further enhance lattice anharmonicity. We propose the incorporation of a chalcogen anion as a bridge ligand to promote phonon rattling in Ag$_6$-octahedron-based compounds. Guided by this design strategy, we theoretically identified five Ag$_6$ octahedron-based compounds, $A$Ag$_3X_2$ ($A$ = Li, Na, and K; $X$ = S and Se), which are characterized by low average atomic masses and exhibit exceptionally strong four-phonon scattering. Consequently, these compounds demonstrate ultralow thermal conductivities (0.3 $\sim$ 0.6 Wm$^{-1}$K$^{-1}$) with minimal temperature dependence (T$^{-0.1}$) across a wide temperature range. Experimental validation confirmed that the $\kappa_{\rm L}$ of NaAg$_3$S$_2$ is 0.45 Wm$^{-1}$K$^{-1}$ within the temperature range of 200 to 550 K. Our results clearly demonstrate that weak chemical bonding plays a crucial role in designing compounds with glass-like $\kappa_{\rm L}$, highlighting the effectiveness of chemical bonding engineering in achieving desired thermal transport properties.

In this paper, we introduce Hunyuan-Large, which is currently the largest open-source Transformer-based mixture of experts model, with a total of 389 billion parameters and 52 billion activation parameters, capable of handling up to 256K tokens. We conduct a thorough evaluation of Hunyuan-Large's superior performance across various benchmarks including language understanding and generation, logical reasoning, mathematical problem-solving, coding, long-context, and aggregated tasks, where it outperforms LLama3.1-70B and exhibits comparable performance when compared to the significantly larger LLama3.1-405B model. Key practice of Hunyuan-Large include large-scale synthetic data that is orders larger than in previous literature, a mixed expert routing strategy, a key-value cache compression technique, and an expert-specific learning rate strategy. Additionally, we also investigate the scaling laws and learning rate schedule of mixture of experts models, providing valuable insights and guidances for future model development and optimization. The code and checkpoints of Hunyuan-Large are released to facilitate future innovations and applications. Codes: https://github.com/Tencent/Hunyuan-Large Models: https://huggingface.co/tencent/Tencent-Hunyuan-Large, Comment: 17 pages, 4 Figures

The Water Cherenkov Detector Array (WCDA) is one of the components of Large High Altitude Air Shower Observatory (LHAASO) and can monitor any sources over two-thirds of the sky for up to 7 hours per day with >98\% duty cycle. In this work, we report the detection of two outbursts of the Fanaroff-Riley I radio galaxy NGC 1275 that were detected by LHAASO-WCDA between November 2022 and January 2023 with statistical significance of 5.2~$\sigma$ and 8.3~$\sigma$. The observed spectral energy distribution in the range from 500 GeV to 3 TeV is fitted by a power-law with a best-fit spectral index of $\alpha=-3.37\pm0.52$ and $-3.35\pm0.29$, respectively. The outburst flux above 0.5~TeV was ($4.55\pm 4.21)\times~10^{-11}~\rm cm^{-2}~s^{-1}$ and ($3.45\pm 1.78)\times~10^{-11}~\rm cm^{-2}~s^{-1}$, corresponding to 60\%, 45\% of Crab Nebula flux. Variation analysis reveals the variability time-scale of days at the TeV energy band. A simple test by one-zone synchrotron self-Compton model reproduces the data in the gamma-ray band well., Comment: 11 pages, 8 figures, 3 tables

We report the detection of an extended very-high-energy (VHE) gamma-ray source coincident with the location of middle-aged (62.4~\rm kyr) pulsar PSR J0248+6021, by using the LHAASO-WCDA data of live 796 days and LHAASO-KM2A data of live 1216 days. A significant excess of \gray induced showers is observed both by WCDA in energy bands of 1-25~\rm TeV and KM2A in energy bands of $>$ 25~\rm TeV with 7.3 $\sigma$ and 13.5 $\sigma$, respectively. The best-fit position derived through WCDA data is R.A. = 42.06$^\circ \pm$ 0.12$^\circ$ and Dec. = 60.24$^\circ \pm $ 0.13$^\circ$ with an extension of 0.69$^\circ\pm$0.15$^\circ$ and that of the KM2A data is R.A.= 42.29$^\circ \pm $ 0.13$^\circ$ and Dec. = 60.38$^\circ \pm$ 0.07$^\circ$ with an extension of 0.37$^\circ\pm$0.07$^\circ$. No clear extended multiwavelength counterpart of this LHAASO source has been found from the radio band to the GeV band. The most plausible explanation of the VHE \gray emission is the inverse Compton process of highly relativistic electrons and positrons injected by the pulsar. These electrons/positrons are hypothesized to be either confined within the pulsar wind nebula or to have already escaped into the interstellar medium, forming a pulsar halo., Comment: 12 pages, 10 figures, Accepted by Sci. China-Phys. Mech. Astron

Large Audio-Language Models (LALMs) have unclocked audio dialogue capabilities, where audio dialogues are a direct exchange of spoken language between LALMs and humans. Recent advances, such as GPT-4o, have enabled LALMs in back-and-forth audio dialogues with humans. This progression not only underscores the potential of LALMs but also broadens their applicability across a wide range of practical scenarios supported by audio dialogues. However, given these advancements, a comprehensive benchmark to evaluate the performance of LALMs in the open-ended audio dialogue understanding remains absent currently. To address this gap, we propose an Audio Dialogue Understanding Benchmark (ADU-Bench), which consists of 4 benchmark datasets. They assess the open-ended audio dialogue ability for LALMs in 3 general scenarios, 12 skills, 9 multilingual languages, and 4 categories of ambiguity handling. Notably, we firstly propose the evaluation of ambiguity handling in audio dialogues that expresses different intentions beyond the same literal meaning of sentences, e.g., "Really!?" with different intonations. In summary, ADU-Bench includes over 20,000 open-ended audio dialogues for the assessment of LALMs. Through extensive experiments conducted on 13 LALMs, our analysis reveals that there is still considerable room for improvement in the audio dialogue understanding abilities of existing LALMs. In particular, they struggle with mathematical symbols and formulas, understanding human behavior such as roleplay, comprehending multiple languages, and handling audio dialogue ambiguities from different phonetic elements, such as intonations, pause positions, and homophones.

We report on a search for millicharged particles (mCPs) produced in cosmic ray proton atmospheric interactions using data collected during the first science run of the LUX-ZEPLIN experiment. The mCPs produced by two processes -- meson decay and proton bremsstrahlung -- are considered in this study. This search utilized a novel signature unique to liquid xenon (LXe) time projection chambers (TPCs), allowing sensitivity to mCPs with masses ranging from 10 to 1000 MeV/c$^2$ and fractional charges between 0.001 and 0.02 of the electron charge e. With an exposure of 60 live days and a 5.5 tonne fiducial mass, we observed no significant excess over background. This represents the first experimental search for atmospheric mCPs and the first search for mCPs using an underground LXe experiment.

Postselected weak measurement has shown significant potential for detecting small physical effects due to its unique weak-value-amplification phenomenon. Previous works suggest that Heisenberg-limit precision can be attained using only the optical coherent states. However, the measurement object is the distribution of postselection, limiting the practical applicability. Here, we demonstrate that the output photons can also reach the quantum scale by utilizing the Fisher information transfer effect. In addition, we consider the insertion of a power-recycling cavity and demonstrate its positive impact on the distribution of postselection. Our results enhance the quantum metrological advantages of the postselection strategy and broaden its application scope., Comment: 6 pages, 4figures

In this work, we perform a detailed analysis to constrain the Hu-Sawicki $f(R)$ gravity model, using cosmic shear data from three prominent Stage-III weak lensing surveys: DES-Y3, KiDS-1000, and HSC-Y3. To accurately model the nonlinear matter clustering in the analysis of cosmic shear signals, we employ FREmu, a recently developed power spectrum emulator for the $f(R)$ gravity trained on the Quijote-MG simulations. This emulator achieves precise predictions, limiting the errors to 5% on scales of $0.009h\,{\rm Mpc}^{-1} < k < 0.5h\,{\rm Mpc}^{-1}$. Our findings reveal that cosmic shear data alone impose only weak constraints on the $f(R)$ parameter $\log_{10}|f_{R_0}|$. To improve these constraints, we incorporate state-of-the-art external observations, including data from the cosmic microwave background and baryon acoustic oscillations. The inclusion of these external datasets significantly enhances the constraints, yielding an upper limit of $\log_{10}|f_{R_0}| < -4.79$ at the 95% confidence level., Comment: 9 pages, 3 figures, 1 table

Structural health monitoring (SHM) has experienced significant advancements in recent decades, accumulating massive monitoring data. Data anomalies inevitably exist in monitoring data, posing significant challenges to their effective utilization. Recently, deep learning has emerged as an efficient and effective approach for anomaly detection in bridge SHM. Despite its progress, many deep learning models require large amounts of labeled data for training. The process of labeling data, however, is labor-intensive, time-consuming, and often impractical for large-scale SHM datasets. To address these challenges, this work explores the use of self-supervised learning (SSL), an emerging paradigm that combines unsupervised pre-training and supervised fine-tuning. The SSL-based framework aims to learn from only a very small quantity of labeled data by fine-tuning, while making the best use of the vast amount of unlabeled SHM data by pre-training. Mainstream SSL methods are compared and validated on the SHM data of two in-service bridges. Comparative analysis demonstrates that SSL techniques boost data anomaly detection performance, achieving increased F1 scores compared to conventional supervised training, especially given a very limited amount of labeled data. This work manifests the effectiveness and superiority of SSL techniques on large-scale SHM data, providing an efficient tool for preliminary anomaly detection with scarce label information.

Graph-based machine learning methods are useful tools in the identification and prediction of variation in genetic data. In particular, the comprehension of phenotypic effects at the cellular level is an accelerating research area in pharmacogenomics. In this article, a novel graph theoretic approach is proposed to infer a co-occurrence network from 16S microbiome data. The approach is specialised to handle datasets containing a small number of samples. Small datasets exacerbate the significant challenges faced by biological data, which exhibit properties such as sparsity, compositionality, and complexity of interactions. Methodologies are also proposed to enrich and statistically filter the inferred networks. The utility of the proposed method lies in that it extracts an informative network from small sampled data that is not only feature-rich, but also biologically meaningful and statistically significant. Although specialised for small data sets, which are abundant, it can be generally applied to any small-sampled dataset, and can also be extended to integrate multi-omics data. The proposed methodology is tested on a data set of chickens vaccinated against and challenged by the protozoan parasite Eimeria tenella. The raw genetic reads are processed, and networks inferred to describe the ecosystems of the chicken intestines under three different stages of disease progression. Analysis of the expression of network features derive biologically intuitive conclusions from purely statistical methods. For example, there is a clear evolution in the distribution of node features in line with the progression of the disease. The distributions also reveal clusters of species interacting mutualistically and parasitically, as expected. Moreover, a specific sub-network is found to persist through all experimental conditions, representative of a persistent microbiome., Comment: 12 pages, 7 figures, under review for a special issue of ACM/IEEE TCBB journal

In this note we review several situations in which stochastic PDEs exhibit ergodic properties. We begin with the basic dissipative conditions, as stated by Da Prato and Zabczyk in their classical monograph. Then we describe the singular case of SPDEs with reflection. Next we move to some degenerate (and thus more demanding) settings. Namely we recall some results obtained around 2006, concerning stochastic Navier-Stokes equations with a very degenerate noise. We finish the article by handling some cases with degenerate coefficients. This includes a new result about the parabolic Anderson model in dimension $d\ge 3$, driven by a general class of noises and fairly general initial conditions. In this context, a phase transition is observed, expressed in terms of the noise intensity., Comment: Dedicated to Giuseppe Da Prato. 40 pages, no figures, 58 references

Semi-supervised domain adaptation (SSDA) aims at training a high-performance model for a target domain using few labeled target data, many unlabeled target data, and plenty of auxiliary data from a source domain. Previous works in SSDA mainly focused on learning transferable representations across domains. However, it is difficult to find a feature space where the source and target domains share the same conditional probability distribution. Additionally, there is no flexible and effective strategy extending existing unsupervised domain adaptation (UDA) approaches to SSDA settings. In order to solve the above two challenges, we propose a novel fine-tuning framework, semi-supervised transfer boosting (SS-TrBoosting). Given a well-trained deep learning-based UDA or SSDA model, we use it as the initial model, generate additional base learners by boosting, and then use all of them as an ensemble. More specifically, half of the base learners are generated by supervised domain adaptation, and half by semi-supervised learning. Furthermore, for more efficient data transmission and better data privacy protection, we propose a source data generation approach to extend SS-TrBoosting to semi-supervised source-free domain adaptation (SS-SFDA). Extensive experiments showed that SS-TrBoosting can be applied to a variety of existing UDA, SSDA and SFDA approaches to further improve their performance.

Uncertainty quantification in text-to-image (T2I) generative models is crucial for understanding model behavior and improving output reliability. In this paper, we are the first to quantify and evaluate the uncertainty of T2I models with respect to the prompt. Alongside adapting existing approaches designed to measure uncertainty in the image space, we also introduce Prompt-based UNCertainty Estimation for T2I models (PUNC), a novel method leveraging Large Vision-Language Models (LVLMs) to better address uncertainties arising from the semantics of the prompt and generated images. PUNC utilizes a LVLM to caption a generated image, and then compares the caption with the original prompt in the more semantically meaningful text space. PUNC also enables the disentanglement of both aleatoric and epistemic uncertainties via precision and recall, which image-space approaches are unable to do. Extensive experiments demonstrate that PUNC outperforms state-of-the-art uncertainty estimation techniques across various settings. Uncertainty quantification in text-to-image generation models can be used on various applications including bias detection, copyright protection, and OOD detection. We also introduce a comprehensive dataset of text prompts and generation pairs to foster further research in uncertainty quantification for generative models. Our findings illustrate that PUNC not only achieves competitive performance but also enables novel applications in evaluating and improving the trustworthiness of text-to-image models., Comment: 28 pages and 22 figures

Making multi-camera visual SLAM systems easier to set up and more robust to the environment is always one of the focuses of vision robots. Existing monocular and binocular vision SLAM systems have narrow FoV and are fragile in textureless environments with degenerated accuracy and limited robustness. Thus multi-camera SLAM systems are gaining attention because they can provide redundancy for texture degeneration with wide FoV. However, current multi-camera SLAM systems face massive data processing pressure and elaborately designed camera configurations, leading to estimation failures for arbitrarily arranged multi-camera systems. To address these problems, we propose a generic visual odometry for arbitrarily arranged multi-cameras, which can achieve metric-scale state estimation with high flexibility in the cameras' arrangement. Specifically, we first design a learning-based feature extraction and tracking framework to shift the pressure of CPU processing of multiple video streams. Then we use the rigid constraints between cameras to estimate the metric scale poses for robust SLAM system initialization. Finally, we fuse the features of the multi-cameras in the SLAM back-end to achieve robust pose estimation and online scale optimization. Additionally, multi-camera features help improve the loop detection for pose graph optimization. Experiments on KITTI-360 and MultiCamData datasets validate the robustness of our method over arbitrarily placed cameras. Compared with other stereo and multi-camera visual SLAM systems, our method obtains higher pose estimation accuracy with better generalization ability. Our codes and online demos are available at \url{https://github.com/JunhaoWang615/MCVO}, Comment: 8 pages, 8 figures

We propose a new quantum computing formalism named Pauli quantum computing. In this formalism, we use the Pauli basis $I$ and $X$ on the non-diagonal blocks of density matrices to encode information and treat them as the computational basis $|0\rangle$ and $|1\rangle$ in standard quantum computing. There are significant differences between Pauli quantum computing and standard quantum computing from the achievable operations to the meaning of measurements, resulting in novel features and comparative advantages for certain tasks. We will give three examples in particular. First, we show how to design Lindbladians to realize imaginary time evolutions and prepare stabilizer ground states in Pauli quantum computing. These stabilizer states can characterize the coherence in the steady subspace of Lindbladians. Second, for quantum amplitudes of the form $\langle +|^{\otimes n}U|0\rangle^{\otimes n}$ with $U$ composed of $\{H,S,T,\text{CNOT}\}$, as long as the number of Hadamard gates in the unitary circuit $U$ is sub-linear $\mathit{o}(n)$, the gate (time) complexity of estimating such amplitudes using Pauli quantum computing formalism can be exponentially reduced compared with the standard formalism ($\mathcal{O}(\epsilon^{-1})$ to $\mathcal{O}(2^{-(n-\mathit{o}(n))/2}\epsilon^{-1})$). Third, given access to a searching oracle under the Pauli encoding picture manifested as a quantum channel, which mimics the phase oracle in Grover's algorithm, the searching problem can be solved with $\mathcal{O}(n)$ scaling for the query complexity and $\mathcal{O}(\text{poly}(n))$ scaling for the time complexity. While so, how to construct such an oracle is highly non-trivial and unlikely efficient due to the hardness of the problem., Comment: 8+3 pages, 1 figure

A regular bipartite graph $\Gamma$ is called semisymmetric if its full automorphism group $\mathrm{Aut}(\Gamma)$ acts transitively on the edge set but not on the vertex set. For a subgroup $G$ of $\mathrm{Aut}(\Gamma)$ that stabilizes the biparts of $\Gamma$, we say that $\Gamma$ is $G$-biprimitive if $G$ acts primitively on each part. In this paper, we first provide a method to construct infinite families of biprimitive semisymmetric graphs admitting almost simple groups. With the aid of this result, a classification of $G$-biprimitive semisymmetric graphs is obtained for $G=\mathrm{A}_n$ or $\mathrm{S}_n$. In pursuit of this goal, we determine all pairs of maximal subgroups of $\mathrm{A}_n$ or $\mathrm{S}_n$ with the same order and all pairs of almost simple groups of the same order., Comment: Your kind suggestions and insights are most warmly welcomed!

The end-to-end autonomous driving paradigm has recently attracted lots of attention due to its scalability. However, existing methods are constrained by the limited scale of real-world data, which hinders a comprehensive exploration of the scaling laws associated with end-to-end autonomous driving. To address this issue, we collected substantial data from various driving scenarios and behaviors and conducted an extensive study on the scaling laws of existing imitation learning-based end-to-end autonomous driving paradigms. Specifically, approximately 4 million demonstrations from 23 different scenario types were gathered, amounting to over 30,000 hours of driving demonstrations. We performed open-loop evaluations and closed-loop simulation evaluations in 1,400 diverse driving demonstrations (1,300 for open-loop and 100 for closed-loop) under stringent assessment conditions. Through experimental analysis, we discovered that (1) the performance of the driving model exhibits a power-law relationship with the amount of training data; (2) a small increase in the quantity of long-tailed data can significantly improve the performance for the corresponding scenarios; (3) appropriate scaling of data enables the model to achieve combinatorial generalization in novel scenes and actions. Our results highlight the critical role of data scaling in improving the generalizability of models across diverse autonomous driving scenarios, assuring safe deployment in the real world. Project repository: https://github.com/ucaszyp/Driving-Scaling-Law

Aerosol forecasting is essential for air quality warnings, health risk assessment, and climate change mitigation. However, it is more complex than weather forecasting due to the intricate interactions between aerosol physicochemical processes and atmospheric dynamics, resulting in significant uncertainty and high computational costs. Here, we develop an artificial intelligence-driven global aerosol-meteorology forecasting system (AI-GAMFS), which provides reliable 5-day, 3-hourly forecasts of aerosol optical components and surface concentrations at a 0.5{\deg} x 0.625{\deg} resolution. AI-GAMFS combines Vision Transformer and U-Net in a backbone network, robustly capturing the complex aerosol-meteorology interactions via global attention and spatiotemporal encoding. Trained on 42 years of advanced aerosol reanalysis data and initialized with GEOS Forward Processing (GEOS-FP) analyses, AI-GAMFS delivers operational 5-day forecasts in one minute. It outperforms the Copernicus Atmosphere Monitoring Service (CAMS) global forecasting system, GEOS-FP forecasts, and several regional dust forecasting systems in forecasting most aerosol variables including aerosol optical depth and dust components. Our results mark a significant step forward in leveraging AI to refine physics-based aerosol forecasting, facilitating more accurate global warnings for aerosol pollution events, such as dust storms and wildfires., Comment: 37 pages, 14 figures

Learning to forecast the trajectories of intelligent agents like pedestrians has caught more researchers' attention. Despite researchers' efforts, it remains a challenge to accurately account for social interactions among agents when forecasting, and in particular, to simulate such social modifications to future trajectories in an explainable and decoupled way. Inspired by the resonance phenomenon of vibration systems, we propose the Resonance (short for Re) model to forecast pedestrian trajectories as co-vibrations, and regard that social interactions are associated with spectral properties of agents' trajectories. It forecasts future trajectories as three distinct vibration terms to represent agents' future plans from different perspectives in a decoupled way. Also, agents' social interactions and how they modify scheduled trajectories will be considered in a resonance-like manner by learning the similarities of their trajectory spectrums. Experiments on multiple datasets, whether pedestrian or vehicle, have verified the usefulness of our method both quantitatively and qualitatively.

This paper discusses a nonlinear integral equation arising from a class of time-consistent portfolio selection problem. We propose a unified framework requiring minimal assumptions, such as right-continuity of market coefficients and square-integrability of the market price of risk. Our main contribution is proving the existence and uniqueness of the square-integrable solution for the integral equation under mild conditions. Illustrative applications include the mean-variance portfolio selection and the utility maximization with random risk aversion.

Topological entities based on bulk-boundary correspondence are ubiquitous, from conventional to higher-order topological insulators, where the protected states are typically localized at the outer boundaries (edges or corners). A less explored scenario involves protected states that are localized at the inner boundaries, sharing the same energy as the bulk states. Here, we propose and demonstrate what we refer to as the bulk-hole correspondence - a relation between the inner robust boundary modes (RBMs) and the existence of multiple "holes" in singular flatband lattices, mediated by the immovable discontinuity of the bulk Bloch wavefunctions. We find that the number of independent flatband states always equals the sum of the number of independent compact localized states and the number of nontrivial inner RBMs, as captured by the Betti number that also counts the hole number from topological data analysis. This correspondence is universal for singular flatband lattices, regardless of the lattice shape and the hole shape. Using laser-written Kagome lattices as a platform, we experimentally observe such inner RBMs, demonstrating their real-space topological nature and robustness. Our results may extend to other singular flatband systems beyond photonics, including non-Euclidean lattices, providing a new approach for understanding nontrivial flatband states and topology in hole-bearing lattice systems., Comment: 12 pages, 4 figures

Understanding and anticipating human movement has become more critical and challenging in diverse applications such as autonomous driving and surveillance. The complex interactions brought by different relations between agents are a crucial reason that poses challenges to this task. Researchers have put much effort into designing a system using rule-based or data-based models to extract and validate the patterns between pedestrian trajectories and these interactions, which has not been adequately addressed yet. Inspired by how humans perceive social interactions with different level of relations to themself, this work proposes the GrouP ConCeption (short for GPCC) model composed of the Group method, which categorizes nearby agents into either group members or non-group members based on a long-term distance kernel function, and the Conception module, which perceives both visual and acoustic information surrounding the target agent. Evaluated across multiple datasets, the GPCC model demonstrates significant improvements in trajectory prediction accuracy, validating its effectiveness in modeling both social and individual dynamics. The qualitative analysis also indicates that the GPCC framework successfully leverages grouping and perception cues human-like intuitively to validate the proposed model's explainability in pedestrian trajectory forecasting., Comment: 15 pages, 10 figures, submitted to CVPR 2025

In many domains, the most successful AI models tend to be the largest, indeed often too large to be handled by AI players with limited computational resources. To mitigate this, a number of compression methods have been developed, including methods that prune the network down to high sparsity whilst retaining performance. The best-performing pruning techniques are often those that use second-order curvature information (such as an estimate of the Fisher information matrix) to score the importance of each weight and to predict the optimal compensation for weight deletion. However, these methods are difficult to scale to high-dimensional parameter spaces without making heavy approximations. Here, we propose the FishLeg surgeon (FLS), a new second-order pruning method based on the Fisher-Legendre (FishLeg) optimizer. At the heart of FishLeg is a meta-learning approach to amortising the action of the inverse FIM, which brings a number of advantages. Firstly, the parameterisation enables the use of flexible tensor factorisation techniques to improve computational and memory efficiency without sacrificing much accuracy, alleviating challenges associated with scalability of most second-order pruning methods. Secondly, directly estimating the inverse FIM leads to less sensitivity to the amplification of stochasticity during inversion, thereby resulting in more precise estimates. Thirdly, our approach also allows for progressive assimilation of the curvature into the parameterisation. In the gradual pruning regime, this results in a more efficient estimate refinement as opposed to re-estimation. We find that FishLeg achieves higher or comparable performance against two common baselines in the area, most notably in the high sparsity regime when considering a ResNet18 model on CIFAR-10 (84% accuracy at 95% sparsity vs 60% for OBS) and TinyIM (53% accuracy at 80% sparsity vs 48% for OBS)., Comment: Published in NeurIPS 2024 - Neural Compression Workshop, 13 pages, 6 figures

In the realm of large-scale point cloud registration, designing a compact symbolic representation is crucial for efficiently processing vast amounts of data, ensuring registration robustness against significant viewpoint variations and occlusions. This paper introduces a novel point cloud registration method, i.e., QuadricsReg, which leverages concise quadrics primitives to represent scenes and utilizes their geometric characteristics to establish correspondences for 6-DoF transformation estimation. As a symbolic feature, the quadric representation fully captures the primary geometric characteristics of scenes, which can efficiently handle the complexity of large-scale point clouds. The intrinsic characteristics of quadrics, such as types and scales, are employed to initialize correspondences. Then we build a multi-level compatibility graph set to find the correspondences using the maximum clique on the geometric consistency between quadrics. Finally, we estimate the 6-DoF transformation using the quadric correspondences, which is further optimized based on the quadric degeneracy-aware distance in a factor graph, ensuring high registration accuracy and robustness against degenerate structures. We test on 5 public datasets and the self-collected heterogeneous dataset across different LiDAR sensors and robot platforms. The exceptional registration success rates and minimal registration errors demonstrate the effectiveness of QuadricsReg in large-scale point cloud registration scenarios. Furthermore, the real-world registration testing on our self-collected heterogeneous dataset shows the robustness and generalization ability of QuadricsReg on different LiDAR sensors and robot platforms. The codes and demos will be released at \url{https://levenberg.github.io/QuadricsReg}., Comment: 25 pages, 17 figures

Large language models (LLMs) have made significant strides at code generation through improved model design, training, and chain-of-thought. However, prompt-level optimizations remain an important yet under-explored aspect of LLMs for coding. This work focuses on the few-shot examples present in most code generation prompts, offering a systematic study on whether few-shot examples improve LLM's coding capabilities, which few-shot examples have the largest impact, and how to select impactful examples. Our work offers 2 approaches for selecting few-shot examples, a model-free method, CODEEXEMPLAR-FREE, and a model-based method, CODEEXEMPLAR-BASED. The 2 methods offer a trade-off between improved performance and reliance on training data and interpretability. Both methods significantly improve CodeLlama's coding ability across the popular HumanEval+ coding benchmark. In summary, our work provides valuable insights into how to pick few-shot examples in code generation prompts to improve LLM code generation capabilities.

A Markov model is a widely used tool for modeling sequences of events from a finite state-space and hence can be employed to identify the transition probabilities across treatments based on treatment sequence data. To understand how patient-level covariates impact these treatment transitions, the transition probabilities are modeled as a function of patient covariates. This approach enables the visualization of the effect of patient-level covariates on the treatment transitions across patient visits. The proposed method automatically estimates the entries of the transition matrix with smaller numbers of empirical transitions as constant; the user can set desired cutoff of the number of empirical transition counts required for a particular transition probability to be estimated as a function of covariates. Firstly, this strategy automatically enforces the final estimated transition matrix to contain zeros at the locations corresponding to zero empirical transition counts, avoiding further complicated model constructs to handle sparsity, in an efficient manner. Secondly, it restricts estimation of transition probabilities as a function of covariates, when the number of empirical transitions is particularly small, thus avoiding the identifiability issue which might arise due to the p>n scenario when estimating each transition probability as a function of patient covariates. To optimize the multi-modal likelihood, a parallelized scalable global optimization routine is also developed. The proposed method is applied to understand how the transitions across disease modifying therapies (DMTs) in Multiple Sclerosis (MS) patients are influenced by patient-level demographic and clinical phenotypes.

Vehicle-to-everything (V2X) technologies offer a promising paradigm to mitigate the limitations of constrained observability in single-vehicle systems. Prior work primarily focuses on single-frame cooperative perception, which fuses agents' information across different spatial locations but ignores temporal cues and temporal tasks (e.g., temporal perception and prediction). In this paper, we focus on temporal perception and prediction tasks in V2X scenarios and design one-step and multi-step communication strategies (when to transmit) as well as examine their integration with three fusion strategies - early, late, and intermediate (what to transmit), providing comprehensive benchmarks with various fusion models (how to fuse). Furthermore, we propose V2XPnP, a novel intermediate fusion framework within one-step communication for end-to-end perception and prediction. Our framework employs a unified Transformer-based architecture to effectively model complex spatiotemporal relationships across temporal per-frame, spatial per-agent, and high-definition map. Moreover, we introduce the V2XPnP Sequential Dataset that supports all V2X cooperation modes and addresses the limitations of existing real-world datasets, which are restricted to single-frame or single-mode cooperation. Extensive experiments demonstrate our framework outperforms state-of-the-art methods in both perception and prediction tasks., Comment: Website link: https://mobility-lab.seas.ucla.edu/v2xpnp/

In human social systems, debates are often seen as a means to resolve differences of opinion. However, in reality, debates frequently incur significant communication costs, especially when dealing with stubborn opponents. Inspired by this phenomenon, this paper examines the impact of malicious agents on the evolution of normal agents' opinions from the perspective of opinion evolution cost, and proposes corresponding solutions for the scenario in which malicious agents hold different opinions in multi-agent systems(MASs). First, the paper analyzes the negative impact of malicious agents on the opinion evolution process, revealing the evolutionary cost they bring, which provides the theoretical foundation for the proposed solution. Next, based on the process of opinion evolution, a strategy is introduced where agents dynamically adjust trust values during the opinion evolution process, gradually isolating malicious agents and achieving this even when malicious agents are in the majority. Additionally, an evolution rate adjustment mechanism is introduced, allowing the system to flexibly regulate the evolution process in complex situations, effectively achieving the trade-off between opinion evolution rate and cost. Extensive numerical simulations demonstrate that the algorithm can effectively isolate the negative influence of malicious agents and achieve a balance between opinion evolution costs and convergence speed., Comment: 20 pages, 12 figures, 2 tables. This work has been submitted to the IEEE for possible publication