arXiv subjects

Bowen Li

Publications and source records attributed to Bowen Li.

At least 145 records · Page 8Linked to original sources

arXiv2024

Cavity QED in a High NA Resonator

From fundamental studies of light-matter interaction to applications in quantum networking and sensing, cavity quantum electrodynamics (QED) provides a platform-crossing toolbox to control interactions between atoms and photons. The coherence of such interactions is determined by the product of the single-pass atomic absorption and the number of photon round-trips. Reducing the cavity loss has enabled resonators supporting nearly 1-million optical roundtrips at the expense of severely limited optical material choices and increased alignment sensitivity. The single-pass absorption probability can be increased through the use of near-concentric, fiber or nanophotonic cavities, which reduce the mode waists at the expense of constrained optical access and exposure to surface fields. Here we present a new high numerical-aperture, lens-based resonator that pushes the single-atom-single-photon absorption probability per round trip close to its fundamental limit by reducing the mode size at the atom below a micron while keeping the atom mm-to-cm away from all optics. This resonator provides strong light-matter coupling in a cavity where the light circulates only ~ 10 times. We load a single 87Rb atom into such a cavity, observe strong coupling, demonstrate cavity-enhanced atom detection with imaging fidelity of 99.55(6) percent and survival probability of 99.89(4) percent in 130 microseconds, and leverage this new platform for a time-resolved exploration of cavity cooling. The resonator's loss-resilience paves the way to coupling of atoms to nonlinear and adaptive optical elements and provides a minimally invasive route to readout of defect centers. Introduction of intra-cavity imaging systems will enable the creation of cavity arrays compatible with Rydberg atom array computing technologies, vastly expanding the applicability of the cavity QED toolbox.

physics.atom-ph↗

arXiv2024

Zero-Query Adversarial Attack on Black-box Automatic Speech Recognition Systems

In recent years, extensive research has been conducted on the vulnerability of ASR systems, revealing that black-box adversarial example attacks pose significant threats to real-world ASR systems. However, most existing black-box attacks rely on queries to the target ASRs, which is impractical when queries are not permitted. In this paper, we propose ZQ-Attack, a transfer-based adversarial attack on ASR systems in the zero-query black-box setting. Through a comprehensive review and categorization of modern ASR technologies, we first meticulously select surrogate ASRs of diverse types to generate adversarial examples. Following this, ZQ-Attack initializes the adversarial perturbation with a scaled target command audio, rendering it relatively imperceptible while maintaining effectiveness. Subsequently, to achieve high transferability of adversarial perturbations, we propose a sequential ensemble optimization algorithm, which iteratively optimizes the adversarial perturbation on each surrogate model, leveraging collaborative information from other models. We conduct extensive experiments to evaluate ZQ-Attack. In the over-the-line setting, ZQ-Attack achieves a 100% success rate of attack (SRoA) with an average signal-to-noise ratio (SNR) of 21.91dB on 4 online speech recognition services, and attains an average SRoA of 100% and SNR of 19.67dB on 16 open-source ASRs. For commercial intelligent voice control devices, ZQ-Attack also achieves a 100% SRoA with an average SNR of 15.77dB in the over-the-air setting.

cs.CR↗

arXiv2024

Iterative Forward Tuning Boosts In-Context Learning in Language Models

Despite the advancements in in-context learning (ICL) for large language models (LLMs), current research centers on specific prompt engineering, such as demonstration selection, with the expectation that a single iteration of demonstrations processing can generalize effectively to a given test sample. However, this perspective overlooks the potential benefits derived from multiple iterations involving demonstrations, a practice aligning more closely with the iterative decision-making process exhibited by humans, who often learn through analogy. In this study, we introduce a novel two-stage framework to boost ICL in LLMs. Specifically, our framework delineates the ICL process into two distinct stages: Deep-Thinking and test stages. The Deep-Thinking stage incorporates a unique attention mechanism, i.e., iterative enhanced attention, which enables multiple rounds of information accumulation. This mechanism operates by manipulating the Key-Value matrices without training, fostering enhanced understanding capabilities in LLMs by thinking demonstrations multiple times. We evaluated Deep-Thinking across a range of benchmarks and LLMs, showing its superior performance over vanilla ICL methods and its effectiveness in challenging tasks where demonstration selection is infeasible.

cs.CL↗

arXiv2024

FedAdOb: Privacy-Preserving Federated Deep Learning with Adaptive Obfuscation

Federated learning (FL) has emerged as a collaborative approach that allows multiple clients to jointly learn a machine learning model without sharing their private data. The concern about privacy leakage, albeit demonstrated under specific conditions, has triggered numerous follow-up research in designing powerful attacking methods and effective defending mechanisms aiming to thwart these attacking methods. Nevertheless, privacy-preserving mechanisms employed in these defending methods invariably lead to compromised model performances due to a fixed obfuscation applied to private data or gradients. In this article, we, therefore, propose a novel adaptive obfuscation mechanism, coined FedAdOb, to protect private data without yielding original model performances. Technically, FedAdOb utilizes passport-based adaptive obfuscation to ensure data privacy in both horizontal and vertical federated learning settings. The privacy-preserving capabilities of FedAdOb, specifically with regard to private features and labels, are theoretically proven through Theorems 1 and 2. Furthermore, extensive experimental evaluations conducted on various datasets and network architectures demonstrate the effectiveness of FedAdOb by manifesting its superior trade-off between privacy preservation and model performance, surpassing existing methods.

cs.CR↗

arXiv2024

Decentralized AI: Permissionless LLM Inference on POKT Network

POKT Network's decentralized Remote Procedure Call (RPC) infrastructure, surpassing 740 billion requests since launching on MainNet in 2020, is well-positioned to extend into providing AI inference services with minimal design or implementation modifications. This litepaper illustrates how the network's open-source and permissionless design aligns incentives among model researchers, hardware operators, API providers and users whom we term model Sources, Suppliers, Gateways and Applications respectively. Through its Relay Mining algorithm, POKT creates a transparent marketplace where costs and earnings directly reflect cryptographically verified usage. This decentralized framework offers large model AI researchers a new avenue to disseminate their work and generate revenue without the complexities of maintaining infrastructure or building end-user products. Supply scales naturally with demand, as evidenced in recent years and the protocol's free market dynamics. POKT Gateways facilitate network growth, evolution, adoption, and quality by acting as application-facing load balancers, providing value-added features without managing LLM nodes directly. This vertically decoupled network, battle tested over several years, is set up to accelerate the adoption, operation, innovation and financialization of open-source models. It is the first mature permissionless network whose quality of service competes with centralized entities set up to provide application grade inference.

cs.DC↗

arXiv2024

On the convergence of discrete dynamic unbalanced transport models

A generalized unbalanced optimal transport distance ${\rm WB}_Λ$ on matrix-valued measures $\mathcal{M}(Ω,\mathbb{S}_+^n)$ was defined in [arXiv:2011.05845] à la Benamou-Brenier, which extends the Kantorovich-Bures and the Wasserstein-Fisher-Rao distances. In this work, we investigate the convergence properties of the discrete transport problems associated with ${\rm WB}_Λ$. We first present a convergence framework for abstract discretization. Then, we propose a specific discretization scheme that aligns with this framework, under the assumption that the initial and final distributions are absolutely continuous with respect to the Lebesgue measure. Moreover, thanks to the static formulation, we show that such an assumption can be removed for the Wasserstein-Fisher-Rao distance.

math.NA↗

arXiv2024

Before Generation, Align it! A Novel and Effective Strategy for Mitigating Hallucinations in Text-to-SQL Generation

Large Language Models (LLMs) driven by In-Context Learning (ICL) have significantly improved the performance of text-to-SQL. Previous methods generally employ a two-stage reasoning framework, namely 1) schema linking and 2) logical synthesis, making the framework not only effective but also interpretable. Despite these advancements, the inherent bad nature of the generalization of LLMs often results in hallucinations, which limits the full potential of LLMs. In this work, we first identify and categorize the common types of hallucinations at each stage in text-to-SQL. We then introduce a novel strategy, Task Alignment (TA), designed to mitigate hallucinations at each stage. TA encourages LLMs to take advantage of experiences from similar tasks rather than starting the tasks from scratch. This can help LLMs reduce the burden of generalization, thereby mitigating hallucinations effectively. We further propose TA-SQL, a text-to-SQL framework based on this strategy. The experimental results and comprehensive analysis demonstrate the effectiveness and robustness of our framework. Specifically, it enhances the performance of the GPT-4 baseline by 21.23% relatively on BIRD dev and it yields significant improvements across six models and four mainstream, complex text-to-SQL benchmarks.

cs.CL↗

arXiv2024

Linear convergence of forward-backward accelerated algorithms without knowledge of the modulus of strong convexity

A significant milestone in modern gradient-based optimization was achieved with the development of Nesterov's accelerated gradient descent (NAG) method. This forward-backward technique has been further advanced with the introduction of its proximal generalization, commonly known as the fast iterative shrinkage-thresholding algorithm (FISTA), which enjoys widespread application in image science and engineering. Nonetheless, it remains unclear whether both NAG and FISTA exhibit linear convergence for strongly convex functions. Remarkably, these algorithms demonstrate convergence without requiring any prior knowledge of strongly convex modulus, and this intriguing characteristic has been acknowledged as an open problem in the comprehensive review [Chambolle and Pock, 2016, Appendix B]. In this paper, we address this question by utilizing the high-resolution ordinary differential equation (ODE) framework. Expanding upon the established phase-space representation, we emphasize the distinctive approach employed in crafting the Lyapunov function, which involves a dynamically adapting coefficient of kinetic energy that evolves throughout the iterations. Furthermore, we highlight that the linear convergence of both NAG and FISTA is independent of the parameter $r$. Additionally, we demonstrate that the square of the proximal subgradient norm likewise advances towards linear convergence.

math.OC↗

arXiv2024

AirShot: Efficient Few-Shot Detection for Autonomous Exploration

Few-shot object detection has drawn increasing attention in the field of robotic exploration, where robots are required to find unseen objects with a few online provided examples. Despite recent efforts have been made to yield online processing capabilities, slow inference speeds of low-powered robots fail to meet the demands of real-time detection-making them impractical for autonomous exploration. Existing methods still face performance and efficiency challenges, mainly due to unreliable features and exhaustive class loops. In this work, we propose a new paradigm AirShot, and discover that, by fully exploiting the valuable correlation map, AirShot can result in a more robust and faster few-shot object detection system, which is more applicable to robotics community. The core module Top Prediction Filter (TPF) can operate on multi-scale correlation maps in both the training and inference stages. During training, TPF supervises the generation of a more representative correlation map, while during inference, it reduces looping iterations by selecting top-ranked classes, thus cutting down on computational costs with better performance. Surprisingly, this dual functionality exhibits general effectiveness and efficiency on various off-the-shelf models. Exhaustive experiments on COCO2017, VOC2014, and SubT datasets demonstrate that TPF can significantly boost the efficacy and efficiency of most off-the-shelf models, achieving up to 36.4% precision improvements along with 56.3% faster inference speed. Code and Data are at: https://github.com/ImNotPrepared/AirShot.

cs.CV↗

arXiv2024

Understanding the PDHG Algorithm via High-Resolution Differential Equations

The least absolute shrinkage and selection operator (Lasso) is widely recognized across various fields of mathematics and engineering. Its variant, the generalized Lasso, finds extensive application in the fields of statistics, machine learning, image science, and related areas. Among the optimization techniques used to tackle this issue, saddle-point methods stand out, with the primal-dual hybrid gradient (PDHG) algorithm emerging as a particularly popular choice. However, the iterative behavior of PDHG remains poorly understood. In this paper, we employ dimensional analysis to derive a system of high-resolution ordinary differential equations (ODEs) tailored for PDHG. This system effectively captures a key feature of PDHG, the coupled $x$-correction and $y$-correction, distinguishing it from the proximal Arrow-Hurwicz algorithm. The small but essential perturbation ensures that PDHG consistently converges, bypassing the periodic behavior observed in the proximal Arrow-Hurwicz algorithm. Through Lyapunov analysis, We investigate the convergence behavior of the system of high-resolution ODEs and extend our insights to the discrete PDHG algorithm. Our analysis indicates that numerical errors resulting from the implicit scheme serve as a crucial factor affecting the convergence rate and monotonicity of PDHG, showcasing a noteworthy pattern also observed for the Alternating Direction Method of Multipliers (ADMM), as identified in [Li and Shi, 2024]. In addition, we further discover that when one component of the objective function is strongly convex, the iterative average of PDHG converges strongly at a rate $O(1/N)$, where $N$ is the number of iterations.

math.OC↗

arXiv2024

Acquiring Weak Annotations for Tumor Localization in Temporal and Volumetric Data

Creating large-scale and well-annotated datasets to train AI algorithms is crucial for automated tumor detection and localization. However, with limited resources, it is challenging to determine the best type of annotations when annotating massive amounts of unlabeled data. To address this issue, we focus on polyps in colonoscopy videos and pancreatic tumors in abdominal CT scans; both applications require significant effort and time for pixel-wise annotation due to the high dimensional nature of the data, involving either temporary or spatial dimensions. In this paper, we develop a new annotation strategy, termed Drag&Drop, which simplifies the annotation process to drag and drop. This annotation strategy is more efficient, particularly for temporal and volumetric imaging, than other types of weak annotations, such as per-pixel, bounding boxes, scribbles, ellipses, and points. Furthermore, to exploit our Drag&Drop annotations, we develop a novel weakly supervised learning method based on the watershed algorithm. Experimental results show that our method achieves better detection and localization performance than alternative weak annotations and, more importantly, achieves similar performance to that trained on detailed per-pixel annotations. Interestingly, we find that, with limited resources, allocating weak annotations from a diverse patient population can foster models more robust to unseen images than allocating per-pixel annotations for a small set of images. In summary, this research proposes an efficient annotation strategy for tumor detection and localization that is less accurate than per-pixel annotations but useful for creating large-scale datasets for screening tumors in various medical modalities.

cs.CV↗

arXiv2024

PirateNets: Physics-informed Deep Learning with Residual Adaptive Networks

While physics-informed neural networks (PINNs) have become a popular deep learning framework for tackling forward and inverse problems governed by partial differential equations (PDEs), their performance is known to degrade when larger and deeper neural network architectures are employed. Our study identifies that the root of this counter-intuitive behavior lies in the use of multi-layer perceptron (MLP) architectures with non-suitable initialization schemes, which result in poor trainablity for the network derivatives, and ultimately lead to an unstable minimization of the PDE residual loss. To address this, we introduce Physics-informed Residual Adaptive Networks (PirateNets), a novel architecture that is designed to facilitate stable and efficient training of deep PINN models. PirateNets leverage a novel adaptive residual connection, which allows the networks to be initialized as shallow networks that progressively deepen during training. We also show that the proposed initialization scheme allows us to encode appropriate inductive biases corresponding to a given PDE system into the network architecture. We provide comprehensive empirical evidence showing that PirateNets are easier to optimize and can gain accuracy from considerably increased depth, ultimately achieving state-of-the-art results across various benchmarks. All code and data accompanying this manuscript will be made publicly available at \url{https://github.com/PredictiveIntelligenceLab/jaxpi}.

cs.LG↗

arXiv2024

Radio Map Assisted Approach for Interference-Aware Predictive UAV Communications

Herein, an interference-aware predictive aerial-and-terrestrial communication problem is studied, where an unmanned aerial vehicle (UAV) delivers some data payload to a few nodes within a communication deadline. The first challenge is the possible interference to the ground base stations (BSs) and users possibly at unknown locations. This paper develops a radio-map-based approach to predict the channel to the receivers and the unintended nodes. Therefore, a predictive communication strategy can be optimized ahead of time to reduce the interference power and duration for the ground nodes. Such predictive optimization raises the second challenge of developing a low-complexity solution for a batch of transmission strategies over T time slots for N receivers before the flight. Mathematically, while the proposed interference-aware predictive communication problem is non-convex, it is converted into a relaxed convex problem, and solved by a novel dual-based algorithm, which is shown to achieve global optimality at asymptotically small slot duration. The proposed algorithm demonstrates orders of magnitude saving of the computational time for moderate T and N compared to several existing solvers. Simulations show that the radio-map-assisted scheme can prevent all unintended receivers with known positions from experiencing interference and significantly reduce the interference to the users at unknown locations.

eess.SP↗

arXiv2024

Structured Acyclic Nets

The concept of structured occurrence nets is an extension of that of occurrence nets which are directed acyclic graphs that represent causality and concurrency information concerning a single execution of a distributed system. The formalism of structured occurrence nets has been introduced to facilitate the portrayal and analysis of the behaviours, and in particular failures, of complex evolving systems. Such systems are composed of a large number of sub-systems which may proceed concurrently and interact with each other and with the external environment while their behaviour is subject to modification by other systems. The purpose of this paper is to provide an extension of structured occurrence nets to include models built up of acyclic nets rather than occurrence nets.

cs.DC↗

arXiv2024

Understanding the ADMM Algorithm via High-Resolution Differential Equations

In the fields of statistics, machine learning, image science, and related areas, there is an increasing demand for decentralized collection or storage of large-scale datasets, as well as distributed solution methods. To tackle this challenge, the alternating direction method of multipliers (ADMM) has emerged as a widely used approach, particularly well-suited to distributed convex optimization. However, the iterative behavior of ADMM has not been well understood. In this paper, we employ dimensional analysis to derive a system of high-resolution ordinary differential equations (ODEs) for ADMM. This system captures an important characteristic of ADMM, called the $λ$-correction, which causes the trajectory of ADMM to deviate from the constrained hyperplane. To explore the convergence behavior of the system of high-resolution ODEs, we utilize Lyapunov analysis and extend our findings to the discrete ADMM algorithm. Through this analysis, we identify that the numerical error resulting from the implicit scheme is a crucial factor that affects the convergence rate and monotonicity in the discrete ADMM algorithm. In addition, we further discover that if one component of the objective function is assumed to be strongly convex, the iterative average of ADMM converges strongly with a rate $O(1/N)$, where $N$ is the number of iterations.

math.OC↗

arXiv2023

Optimisation-Based Multi-Modal Semantic Image Editing

Image editing affords increased control over the aesthetics and content of generated images. Pre-existing works focus predominantly on text-based instructions to achieve desired image modifications, which limit edit precision and accuracy. In this work, we propose an inference-time editing optimisation, designed to extend beyond textual edits to accommodate multiple editing instruction types (e.g. spatial layout-based; pose, scribbles, edge maps). We propose to disentangle the editing task into two competing subtasks: successful local image modifications and global content consistency preservation, where subtasks are guided through two dedicated loss functions. By allowing to adjust the influence of each loss function, we build a flexible editing solution that can be adjusted to user preferences. We evaluate our method using text, pose and scribble edit conditions, and highlight our ability to achieve complex edits, through both qualitative and quantitative experiments.

cs.CV↗

arXiv2023

Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLs

Text-to-SQL parsing, which aims at converting natural language instructions into executable SQLs, has gained increasing attention in recent years. In particular, Codex and ChatGPT have shown impressive results in this task. However, most of the prevalent benchmarks, i.e., Spider, and WikiSQL, focus on database schema with few rows of database contents leaving the gap between academic study and real-world applications. To mitigate this gap, we present Bird, a big benchmark for large-scale database grounded in text-to-SQL tasks, containing 12,751 pairs of text-to-SQL data and 95 databases with a total size of 33.4 GB, spanning 37 professional domains. Our emphasis on database values highlights the new challenges of dirty database contents, external knowledge between NL questions and database contents, and SQL efficiency, particularly in the context of massive databases. To solve these problems, text-to-SQL models must feature database value comprehension in addition to semantic parsing. The experimental results demonstrate the significance of database values in generating accurate text-to-SQLs for big databases. Furthermore, even the most effective text-to-SQL models, i.e. ChatGPT, only achieves 40.08% in execution accuracy, which is still far from the human result of 92.96%, proving that challenges still stand. Besides, we also provide an efficiency analysis to offer insights into generating text-to-efficient-SQLs that are beneficial to industries. We believe that BIRD will contribute to advancing real-world applications of text-to-SQL research. The leaderboard and source code are available: https://bird-bench.github.io/.

cs.CL↗

arXiv2023

ImDiffusion: Imputed Diffusion Models for Multivariate Time Series Anomaly Detection

Anomaly detection in multivariate time series data is of paramount importance for ensuring the efficient operation of large-scale systems across diverse domains. However, accurately detecting anomalies in such data poses significant challenges. Existing approaches, including forecasting and reconstruction-based methods, struggle to address these challenges effectively. To overcome these limitations, we propose a novel anomaly detection framework named ImDiffusion, which combines time series imputation and diffusion models to achieve accurate and robust anomaly detection. The imputation-based approach employed by ImDiffusion leverages the information from neighboring values in the time series, enabling precise modeling of temporal and inter-correlated dependencies, reducing uncertainty in the data, thereby enhancing the robustness of the anomaly detection process. ImDiffusion further leverages diffusion models as time series imputers to accurately capturing complex dependencies. We leverage the step-by-step denoised outputs generated during the inference process to serve as valuable signals for anomaly prediction, resulting in improved accuracy and robustness of the detection process. We evaluate the performance of ImDiffusion via extensive experiments on benchmark datasets. The results demonstrate that our proposed framework significantly outperforms state-of-the-art approaches in terms of detection accuracy and timeliness. ImDiffusion is further integrated into the real production system in Microsoft and observe a remarkable 11.4% increase in detection F1 score compared to the legacy approach. To the best of our knowledge, ImDiffusion represents a pioneering approach that combines imputation-based techniques with time series anomaly detection, while introducing the novel use of diffusion models to the field.

cs.LG↗