Searcharxiv⌕ Search

arXiv subjects

Xi Chen

Publications and source records attributed to Xi Chen.

At least 397 records · Page 22Linked to original sources
arXiv2025

Gemini: A Family of Highly Capable Multimodal Models

This report introduces a new family of multimodal models, Gemini, that exhibit remarkable capabilities across image, audio, video, and text understanding. The Gemini family consists of Ultra, Pro, and Nano sizes, suitable for applications ranging from complex reasoning tasks to on-device memory-constrained use-cases. Evaluation on a broad range of benchmarks shows that our most-capable Gemini Ultra model advances the state of the art in 30 of 32 of these benchmarks - notably being the first model to achieve human-expert performance on the well-studied exam benchmark MMLU, and improving the state of the art in every one of the 20 multimodal benchmarks we examined. We believe that the new capabilities of the Gemini family in cross-modal reasoning and language understanding will enable a wide variety of use cases. We discuss our approach toward post-training and deploying Gemini models responsibly to users through services including Gemini, Gemini Advanced, Google AI Studio, and Cloud Vertex AI.

cs.CL↗
arXiv2025

Scrambling in the Charging of Quantum Batteries

Exponentially fast scrambling of an initial state characterizes quantum chaotic systems. Given the importance of quickly populating higher energy levels from low-energy states in quantum battery charging protocols, this work investigates the role of quantum scrambling in quantum batteries and its effect on optimal power and charging times by means of the Sachdev-Ye-Kitaev model, a maximally-chaotic black hole physics model that has been recently proposed as a quantum battery. We adopt a bare representation with normalized bandwidths to suppress system energy dependence. To our knowledge, this is the first in-depth exploration of quantum scrambling in the context of quantum batteries. By analyzing the dynamics of out-of-time-order correlators, our findings indicate that quantum scrambling does not necessarily lead to faster charging, despite its potential for accelerating the process.

quant-ph↗
arXiv2025

Cooling process of the high-mass young stellar object G24.33+0.14 following an accretion burst

The HMYSO G24.33+0.14 (G24), has recently been observed to undergo an accretion burst since September 2019, lasting approximately two years. By utilizing 1.3 mm observational data from the NOrthern Extended Millimeter Array (NOEMA) in March 2020 and the Atacama Large Millimeter/submillimeter Array (ALMA) in September 2019, we have examined the physical environment changes in gas and dust within G24 region during the decay phase of the accretion burst. Following the burst, the continuum emission in the inner core region of G24 diminished by approximately 20%, while the emission in the outer region exhibited an increase by a factor of ~30%. This pattern indicates that the heat wave, triggered by the accretion burst, radiated outward from the core's interior to its periphery over the half-year period, with a calculated propagation speed of 0.08-0.38 times the speed of light. Moreover, the methanol emission intensity in this area has experienced a notable decline, with the rate of flux reduction correlating positively with the energy of the upper energy states. This, in conjunction with the analysis of methanol molecular line rotation temperature diagrams for different emitting regions, further substantiates that the core region of G24 cooled down, contrasted with the persistent heating in the outer region following the burst.

astro-ph.GA↗
arXiv2025

A Mysterious Connection Between Tolerant Junta Testing and Agnostically Learning Conjunctions

The main conceptual contribution of this paper is identifying a previously unnoticed connection between two central problems in computational learning theory and property testing: agnostically learning conjunctions and tolerantly testing juntas. Inspired by this connection, the main technical contribution is a pair of improved algorithms for these two problems. In more detail, - We give a distribution-free algorithm for agnostically PAC learning conjunctions over $\{\pm 1\}^n$ that runs in time $2^{\widetilde{O}(n^{1/3})}$, for constant excess error $\varepsilon$. This improves on the fastest previously published algorithm, which runs in time $2^{\widetilde{O}(n^{1/2})}$ [KKMS08]. - Building on the ideas in our agnostic conjunction learner and using significant additional technical ingredients, we give an adaptive tolerant testing algorithm for $k$-juntas that makes $2^{\widetilde{O}(k^{1/3})}$ queries, for constant "gap parameter" $\varepsilon$ between the "near" and "far" cases. This improves on the best previous results, due to [ITW21, NP24], which make $2^{\widetilde{O}(\sqrt{k})}$ queries. Since there is a known $2^{\widetildeΩ(\sqrt{k})}$ lower bound for non-adaptive tolerant junta testers, our result shows that adaptive tolerant junta testing algorithms provably outperform non-adaptive ones.

cs.DS↗
arXiv2025

Multilevel Monte Carlo Metamodeling for Variance Function Estimation

This work introduces a novel multilevel Monte Carlo (MLMC) metamodeling approach for variance function estimation. Although devising an efficient experimental design for simulation metamodeling can be elusive, the MLMC-based approach addresses this challenge by dynamically adjusting the number of design points and budget allocation at each level, thereby automatically creating an efficient design. Theoretical analyses show that, under mild conditions, the proposed MLMC metamodeling approach for variance function estimation can achieve superior computational efficiency compared to standard Monte Carlo metamodeling while achieving the desired level of accuracy. Additionally, this work establishes the asymptotic normality of the MLMC metamodeling estimator under certain sufficient conditions, providing valuable insights for uncertainty quantification. Finally, two MLMC metamodeling procedures are proposed for variance function estimation: one to achieve a target accuracy level and another to efficiently utilize a fixed computational budget. Numerical evaluations support the theoretical results and demonstrate the potential of the proposed approach in facilitating global sensitivity analysis.

stat.ME↗
arXiv2025

Intrinsic (Axion) Statistical Topological Insulator

Ensembles that respect symmetries on average exhibit richer topological states than those in pure states with exact symmetries, leading to the concept of average symmetry-protected topological states (ASPTs). The free-fermion counterpart of ASPT is the so-called statistical topological insulator (STI) in disordered ensembles. In this work, we demonstrate the existence of an intrinsic STI, which has no clean counterpart. Using a real space construction (topological crystal), we find an axion STI characterized by the average axion angle $\barθ=π$, protected by an average $C_4T$ symmetry with $(C_4T)^4=1$. While the exact $C_{4}T$ symmetry reverses the sign of $θ$ angle, and hence seems to protect a $\mathbb{Z}_2$ classification of $θ\!=\!0,π$, we prove that the $θ\!=\!π$ state cannot be realized in the clean limit if $(C_{4}T)^4 \!=\! 1$. Therefore, the axion STI lacks band insulator correspondence and is thus intrinsic. To illustrate this state, we construct a lattice model and numerically explore its phase diagram, identifying an axion STI phase separated from both band insulators and trivial Anderson insulators by a metallic phase, revealing the intrinsic nature of the STI. We also argue that the intrinsic STI is robust against electron-electron interactions. Our work thus provides the first intrinsic crystalline ASPT and its lattice realization.

cond-mat.mes-hall↗
arXiv2025

Online Video Understanding: OVBench and VideoChat-Online

Multimodal Large Language Models (MLLMs) have significantly progressed in offline video understanding. However, applying these models to real-world scenarios, such as autonomous driving and human-computer interaction, presents unique challenges due to the need for real-time processing of continuous online video streams. To this end, this paper presents systematic efforts from three perspectives: evaluation benchmark, model architecture, and training strategy. First, we introduce OVBench, a comprehensive question-answering benchmark designed to evaluate models' ability to perceive, memorize, and reason within online video contexts. It features 6 core task types across three temporal contexts-past, current, and future-forming 16 subtasks from diverse datasets. Second, we propose a new Pyramid Memory Bank (PMB) that effectively retains key spatiotemporal information in video streams. Third, we proposed an offline-to-online learning paradigm, designing an interleaved dialogue format for online video data and constructing an instruction-tuning dataset tailored for online video training. This framework led to the development of VideoChat-Online, a robust and efficient model for online video understanding. Despite the lower computational cost and higher efficiency, VideoChat-Online outperforms existing state-of-the-art offline and online models across popular offline video benchmarks and OVBench, demonstrating the effectiveness of our model architecture and training strategy. % Our approach surpasses existing state-of-the-art offline models Qwen2-VL 7B and online models Flash-VStream, by 4.19% and 23.7% on OVBench, respectively.

cs.CV↗
arXiv2025

Testing Juntas and Junta Subclasses with Relative Error

This papers considers the junta testing problem in a recently introduced ``relative error'' variant of the standard Boolean function property testing model. In relative-error testing we measure the distance from $f$ to $g$, where $f,g: \{0,1\}^n \to \{0,1\}$, by the ratio of $|f^{-1}(1) \triangle g^{-1}(1)|$ (the number of inputs on which $f$ and $g$ disagree) to $|f^{-1}(1)|$ (the number of satisfying assignments of $f$), and we give the testing algorithm both black-box access to $f$ and also access to independent uniform samples from $f^{-1}(1)$. Chen et al. (SODA 2025) observed that the class of $k$-juntas is $\text{poly}(2^k,1/ε)$-query testable in the relative-error model, and asked whether $\text{poly}(k,1/ε)$ queries is achievable. We answer this question affirmatively by giving a $\tilde{O}(k/ε)$-query algorithm, matching the optimal complexity achieved in the less challenging standard model. Moreover, as our main result, we show that any subclass of $k$-juntas that is closed under permuting variables is relative-error testable with a similar complexity. This gives highly efficient relative-error testing algorithms for a number of well-studied function classes, including size-$k$ decision trees, size-$k$ branching programs, and size-$k$ Boolean formulas.

cs.CC↗
arXiv2025

Relative-error testing of conjunctions and decision lists

We study the relative-error property testing model for Boolean functions that was recently introduced in the work of Chen et al. (SODA 2025). In relative-error testing, the testing algorithm gets uniform random satisfying assignments as well as black-box queries to $f$, and it must accept $f$ with high probability whenever $f$ has the property that is being tested and reject any $f$ that is relative-error far from having the property. Here the relative-error distance from $f$ to a function $g$ is measured with respect to $|f^{-1}(1)|$ rather than with respect to the entire domain size $2^n$ as in the Hamming distance measure that is used in the standard model; thus, unlike the standard model, relative-error testing allows us to study the testability of sparse Boolean functions that have few satisfying assignments. It was shown in Chen et al. (SODA 2025) that relative-error testing is at least as difficult as standard-model property testing, but for many natural and important Boolean function classes the precise relationship between the two notions is unknown. In this paper we consider the well-studied and fundamental properties of being a conjunction and being a decision list. In the relative-error setting, we give an efficient one-sided error tester for conjunctions with running time and query complexity $O(1/ε)$. Secondly, we give a two-sided relative-error $\tilde{O}$$(1/ε)$ tester for decision lists, matching the query complexity of the state-of-the-art algorithm in the standard model Bshouty (RANDOM 2020) and Diakonikolas et al. (FOCS 2007).

cs.CC↗
arXiv2025

EIDT-V: Exploiting Intersections in Diffusion Trajectories for Model-Agnostic, Zero-Shot, Training-Free Text-to-Video Generation

Zero-shot, training-free, image-based text-to-video generation is an emerging area that aims to generate videos using existing image-based diffusion models. Current methods in this space require specific architectural changes to image generation models, which limit their adaptability and scalability. In contrast to such methods, we provide a model-agnostic approach. We use intersections in diffusion trajectories, working only with the latent values. We could not obtain localized frame-wise coherence and diversity using only the intersection of trajectories. Thus, we instead use a grid-based approach. An in-context trained LLM is used to generate coherent frame-wise prompts; another is used to identify differences between frames. Based on these, we obtain a CLIP-based attention mask that controls the timing of switching the prompts for each grid cell. Earlier switching results in higher variance, while later switching results in more coherence. Therefore, our approach can ensure appropriate control between coherence and variance for the frames. Our approach results in state-of-the-art performance while being more flexible when working with diverse image-generation models. The empirical analysis using quantitative metrics and user studies confirms our model's superior temporal consistency, visual fidelity and user satisfaction, thus providing a novel way to obtain training-free, image-based text-to-video generation.

cs.CV↗
arXiv2025

A Study of Multiple Molecular Lines at the 3 mm Band toward Gas Infalling Sources

The study of multiple molecular spectral lines in gas infalling sources can provide the physical and chemical properties of these sources and help us estimate their evolutionary stages. We report line detections within the 3 mm band using the FTS wide-sideband mode of the IRAM 30 m telescope toward 20 gas-infalling sources. Using XCLASS, we identify the emission lines of up to 22 molecular species (including a few isotopologues) and one hydrogen radio recombination line in these sources. H$^{13}$CO$^+$, HCO$^+$, HCN, HNC, c-C$_3$H$_2$, and CCH lines are detected in 15 sources. We estimate the rotation temperatures and column densities of these molecular species using the LTE radiative transfer model, and compare the molecular abundances of these sources with those from nine high-mass star-forming regions reported in previous studies and with those from the chemical model. Our results suggest that G012.79-0.20, G012.87-0.22 clump A and B, and G012.96-0.23 clump A may be in the high-mass protostellar object stage, while sources with fewer detected species may be in the earlier evolutionary stage. Additionally, the CCH and c-C$_3$H$_2$ column densities in our sources reveal a linear correlation, with a ratio of N(CCH)/N(c-C$_3$H$_2$) = 89.2$\pm$5.6, which is higher than the ratios reported in the literature. When considering only sources with lower column densities, this ratio decreases to 29.0$\pm$6.1, consistent with those of diffuse clouds. Furthermore, a comparison between the N(CCH)/N(c-C$_3$H$_2$) ratio and the sources' physical parameters reveals a correlation, with sources exhibiting higher ratios tending to have higher kinetic temperatures and H$_2$ column densities.

astro-ph.GA↗
arXiv2025

Distributed Tensor Principal Component Analysis with Data Heterogeneity

As tensors become widespread in modern data analysis, Tucker low-rank Principal Component Analysis (PCA) has become essential for dimensionality reduction and structural discovery in tensor datasets. Motivated by the common scenario where large-scale tensors are distributed across diverse geographic locations, this paper investigates tensor PCA within a distributed framework where direct data pooling is impractical. We offer a comprehensive analysis of three specific scenarios in distributed Tensor PCA: a homogeneous setting in which tensors at various locations are generated from a single noise-affected model; a heterogeneous setting where tensors at different locations come from distinct models but share some principal components, aiming to improve estimation across all locations; and a targeted heterogeneous setting, designed to boost estimation accuracy at a specific location with limited samples by utilizing transferred knowledge from other sites with ample data. We introduce novel estimation methods tailored to each scenario, establish statistical guarantees, and develop distributed inference techniques to construct confidence regions. Our theoretical findings demonstrate that these distributed methods achieve sharp rates of accuracy by efficiently aggregating shared information across different tensors, while maintaining reasonable communication costs. Empirical validation through simulations and real-world data applications highlights the advantages of our approaches, particularly in managing heterogeneous tensor data.

stat.ME↗
arXiv2025

Quantum control of a random transverse Ising spin system

We consider subspace transfer within the time-dependent one-dimensional quantum transverse Ising model, with random nearest-neighbor interactions and a transverse field. We run numerical simulations using a variational approach and the numerical GRAPE (gradient-ascent pulse engineering) and dCRAB (dressed chopped random basis) quantum control algorithms.

quant-ph↗
arXiv2025

Debate-Feedback: A Multi-Agent Framework for Efficient Legal Judgment Prediction

The use of AI in legal analysis and prediction (LegalAI) has gained widespread attention, with past research focusing on retrieval-based methods and fine-tuning large models. However, these approaches often require large datasets and underutilize the capabilities of modern large language models (LLMs). In this paper, inspired by the debate phase of real courtroom trials, we propose a novel legal judgment prediction model based on the Debate-Feedback architecture, which integrates LLM multi-agent debate and reliability evaluation models. Unlike traditional methods, our model achieves significant improvements in efficiency by minimizing the need for large historical datasets, thus offering a lightweight yet robust solution. Comparative experiments show that it outperforms several general-purpose and domain-specific legal models, offering a dynamic reasoning process and a promising direction for future LegalAI research.

cs.MA↗
arXiv2025

EnvGS: Modeling View-Dependent Appearance with Environment Gaussian

Reconstructing complex reflections in real-world scenes from 2D images is essential for achieving photorealistic novel view synthesis. Existing methods that utilize environment maps to model reflections from distant lighting often struggle with high-frequency reflection details and fail to account for near-field reflections. In this work, we introduce EnvGS, a novel approach that employs a set of Gaussian primitives as an explicit 3D representation for capturing reflections of environments. These environment Gaussian primitives are incorporated with base Gaussian primitives to model the appearance of the whole scene. To efficiently render these environment Gaussian primitives, we developed a ray-tracing-based renderer that leverages the GPU's RT core for fast rendering. This allows us to jointly optimize our model for high-quality reconstruction while maintaining real-time rendering speeds. Results from multiple real-world and synthetic datasets demonstrate that our method produces significantly more detailed reflections, achieving the best rendering quality in real-time novel view synthesis. The code is available at https://zju3dv.github.io/envgs.

cs.CV↗
arXiv2025

DeCompress: Denoising via Neural Compression

Learning-based denoising algorithms achieve state-of-the-art performance across various denoising tasks. However, training such models relies on access to large training datasets consisting of clean and noisy image pairs. On the other hand, in many imaging applications, such as microscopy, collecting ground truth images is often infeasible. To address this challenge, researchers have recently developed algorithms that can be trained without requiring access to ground truth data. However, training such models remains computationally challenging and still requires access to large noisy training samples. In this work, inspired by compression-based denoising and recent advances in neural compression, we propose a new compression-based denoising algorithm, which we name DeCompress, that i) does not require access to ground truth images, ii) does not require access to large training dataset - only a single noisy image is sufficient, iii) is robust to overfitting, and iv) achieves superior performance compared with zero-shot or unsupervised learning-based denoisers.

eess.IV↗
arXiv2025

Gemini Robotics: Bringing AI into the Physical World

Recent advancements in large multimodal models have led to the emergence of remarkable generalist capabilities in digital domains, yet their translation to physical agents such as robots remains a significant challenge. This report introduces a new family of AI models purposefully designed for robotics and built upon the foundation of Gemini 2.0. We present Gemini Robotics, an advanced Vision-Language-Action (VLA) generalist model capable of directly controlling robots. Gemini Robotics executes smooth and reactive movements to tackle a wide range of complex manipulation tasks while also being robust to variations in object types and positions, handling unseen environments as well as following diverse, open vocabulary instructions. We show that with additional fine-tuning, Gemini Robotics can be specialized to new capabilities including solving long-horizon, highly dexterous tasks, learning new short-horizon tasks from as few as 100 demonstrations and adapting to completely novel robot embodiments. This is made possible because Gemini Robotics builds on top of the Gemini Robotics-ER model, the second model we introduce in this work. Gemini Robotics-ER (Embodied Reasoning) extends Gemini's multimodal reasoning capabilities into the physical world, with enhanced spatial and temporal understanding. This enables capabilities relevant to robotics including object detection, pointing, trajectory and grasp prediction, as well as multi-view correspondence and 3D bounding box predictions. We show how this novel combination can support a variety of robotics applications. We also discuss and address important safety considerations related to this new class of robotics foundation models. The Gemini Robotics family marks a substantial step towards developing general-purpose robots that realizes AI's potential in the physical world.

cs.RO↗