arXiv subjects

Junyi Shen

Publications and source records attributed to Junyi Shen.

At least 19 recordsLinked to original sources

arXiv2026

Cross-validation of six dispersion measure estimation methods for FRB 20240114A

Fast Radio Bursts (FRBs) are important cosmological probes, but their applications depend critically on accurate dispersion measure (DM) determinations. We present a systematic comparison of six DM estimation methods using 2,874 bursts from FRB20240114A, the most active repeating FRB currently known, observed by FAST during a single 4.4-hr session on 2024 March 12. This large, homogeneous sample over a short timescale, during which the propagation environment is expected to be nearly static, provides an ideal benchmark for isolating algorithmic effects on DM determination. We investigate the dependence of inter-method consistency on signal-to-noise ratio (S/N), burst morphology, and radio frequency interference (RFI). Low-S/N bursts exhibit significantly larger inter-method deviations, while single-component bursts produce highly consistent DM values across methods. In contrast, complex double- and multiple-component bursts with drifting substructures lead to substantial inter-method scattering, indicating that DM discrepancies are primarily driven by algorithmic responses to burst morphology. RFI does not significantly alter the global statistical behavior of DM deviations, but it affects density-filtering methods through morphology distortion caused by frequency-channel masking. Even after imposing strict inter-method consistency constraints, FRB20240114A still exhibits notable apparent DM fluctuations spanning $\sim$528-534~pc~cm$^{-3}$ over 15,780s. For morphologically simple bursts these variations far exceed the measurement uncertainty and, on second-to-minute timescales, cannot arise from any plausible change in the line-of-sight electron column, pointing instead to a frequency-dependent emission-time structure intrinsic to the bursts that mimics dispersion.

astro-ph.HE↗

arXiv2026

Efficient LLM Serving for Agentic Workflows: A Data Systems Perspective

Agentic workflows are composed of sequences of interdependent Large Language Model (LLM) calls, and they have become a dominant workload in modern AI systems. These workflows exhibit extensive redundancy from overlapping prompts and intermediate results due to speculative and parallel exploration. Existing LLM serving systems, such as vLLM, focus on optimizing individual inference calls and overlook cross-call dependencies, leading to significant inefficiencies. This paper rethinks LLM and agent serving from a data systems perspective and introduces Helium, a workflow-aware serving framework that models agentic workloads as query plans and treats LLM invocations as first-class operators. Helium integrates proactive caching and cache-aware scheduling to maximize reuse across prompts, KV states, and workflows. Through these techniques, Helium bridges classic query optimization principles with LLM serving, achieving up to 1.56x speedup over state-of-the-art agent serving systems on various workloads. Our results demonstrate that end-to-end optimization across workflows is essential for scalable and efficient LLM-based agents.

cs.MA↗

arXiv2026

Double: Breaking the Acceleration Limit via Double Retrieval Speculative Parallelism

Parallel Speculative Decoding (PSD) accelerates traditional Speculative Decoding (SD) by overlapping draft generation with verification. However, it remains hampered by two fundamental challenges: (1) a theoretical speedup ceiling dictated by the speed ratio between the draft and target models, and (2) high computational waste and pipeline stall due to mid-sequence token rejections of early errors. To address these limitations, we introduce \textsc{Double} (Double Retrieval Speculative Parallelism). By bridging the gap between SD and PSD, our framework resolves the Retrieval \emph{Precision-Efficiency Dilemma} through a novel synchronous mechanism. Specifically, we enable the draft model to execute iterative retrieval speculations to break the theoretical speedup limits; to alleviate rejections without rollback, the target model performs authoritative retrieval to generate multi-token guidance. \textsc{Double} is entirely training-free and lossless. Extensive experiments demonstrate state-of-the-art speedup of $\textbf{5.3}\times$ on LLaMA3.3-70B and $\textbf{2.8}\times$ on Qwen3-32B, significantly outperforming the advanced method EAGLE-3 that requires extensive model training.

cs.CL↗

arXiv2025

FlowMesh: A Service Fabric for Composable LLM Workflows

AI deployment increasingly resembles a pipeline of data transformation, fine-tuning, and agent interactions rather than a monolithic LLM job; recent examples include RLHF/RLAIF training and agentic workflows. To cope with this shift, we propose FlowMesh, a multi-tenant service fabric that executes and optimizes these workloads as one shared service instead of isolated pipelines. It decomposes workflows into fine-grained operators with recorded lineage, enabling de-duplication of work across users and batching requests on the same hardware while preserving per-workflow provenance. A global control plane maintains a cluster-wide pool of ready operators and uses a single utility function to pick both the batch and the worker, balancing throughput, cost, and data locality on heterogeneous GPUs. The data plane is an elastic fleet of stateless workers backed by a content-addressable store, enabling rapid, automatic scale-out, safe retry after preemption, and portability across managed clusters such as Kubernetes and geo-distributed GPU marketplaces such as Vast.ai. Compared with baseline solutions, FlowMesh achieves up to 3.8x cost reduction and 2.0x lower energy usage, provides a similar or better latency profile, and remains efficient under dynamic and failure-prone conditions.

cs.DC↗

arXiv2025

Batch Query Processing and Optimization for Agentic Workflows

Large Language Models (LLMs) in agentic workflows combine multi-step reasoning, heterogeneous tool use, and collaboration across multiple specialized agents. Existing LLM serving engines optimize individual calls in isolation, while multi-agent frameworks focus on orchestration without system-level performance planning. As a result, repeated prompts, overlapping contexts, and fragmented CPU-GPU execution create substantial redundancy and poor hardware utilization, especially in batch analytics scenarios. We introduce Halo, a system that brings batch query processing and optimization into agentic LLM workflows. Halo represents each workflow as a structured query plan DAG and constructs a consolidated graph for batched queries that exposes shared computation. Guided by a cost model that jointly considers heterogeneous resource constraints, prefill and decode costs, cache reuse, and GPU placement, Halo performs plan-level optimization to minimize redundant execution. The Processor integrates adaptive batching, KV-cache sharing and migration, along with fine-grained CPU-GPU pipelining to maximize holistic hardware efficiency. Evaluation across six benchmarks shows that Halo achieves up to 3.6x speedup for batch inference and 2.6x throughput improvement under online serving, scaling to workloads of thousands of queries and complex graphs. These gains are achieved without compromising output quality. By unifying query optimization with heterogeneous LLM serving, Halo enables efficient agentic workflows in data analytics and decision-making applications.

cs.DB↗

arXiv2025

Towards Reliable Multi-Agent Systems for Marketing Applications via Reflection, Memory, and Planning

Recent advances in large language models (LLMs) enabled the development of AI agents that can plan and interact with tools to complete complex tasks. However, literature on their reliability in real-world applications remains limited. In this paper, we introduce a multi-agent framework for a marketing task: audience curation. To solve this, we introduce a framework called RAMP that iteratively plans, calls tools, verifies the output, and generates suggestions to improve the quality of the audience generated. Additionally, we equip the model with a long-term memory store, which is a knowledge base of client-specific facts and past queries. Overall, we demonstrate the use of LLM planning and memory, which increases accuracy by 28 percentage points on a set of 88 evaluation queries. Moreover, we show the impact of iterative verification and reflection on more ambiguous queries, showing progressively better recall (roughly +20 percentage points) with more verify/reflect iterations on a smaller challenge set, and higher user satisfaction. Our results provide practical insights for deploying reliable LLM-based systems in dynamic, industry-facing environments.

cs.CL↗

arXiv2025

Investigating FRB 20240114A with FAST: Morphological Classification and Drifting Rate Measurements in a Burst-Cluster Framework

This study investigates the morphological classification and drifting rate measurement of the repeating fast radio burst (FRB) source FRB 20240114A using the Five-hundred-meter Aperture Spherical Telescope (FAST). Detected on January 14, 2024, FRB 20240114A exhibited an exceptionally high burst rate, revealing unique properties. Through observational campaigns over several months, we selected a dataset comprising 3,203 bursts (2,109 burst-clusters) during a continuous monitoring session (15,780 seconds) on March 12, 2024. Improving upon previous work, we clarify the definitions of sub-bursts, bursts and burst-clusters. Using an average dispersion measures (DM) of 529.2 pc cm$^{-3}$, we classified the burst-clusters into Downward Drifting, Upward Drifting, No Drifting, No Evidence for Drifting, Not-Clear, and Complex burst-clusters. Among the 978 burst-clusters that exhibit drifting behavior, 233 (23.82%) show upward drifting. Additionally, if 142 upward drifting single-component burst-clusters are excluded, upward drifting double- and multi-component burst-clusters still account for 10.89% of the 836 burst-clusters exhibiting drifting behavior, equating to 91 burst-clusters. Furthermore, if only upward drifting burst-clusters with consecutive time intervals (or upward drifting bursts) are considered, only 9 bursts remain. Drifting rate comparisons with other physical quantities reveal that the drifting rate increases with peak frequency for single-component burst-clusters with drifting behavior. Moreover, in single-component burst-clusters, those with upward drifting exhibit smaller effective widths, bandwidths, and fluxes than their downward drifting counterparts. A Kolmogorov-Smirnov test further indicates that upward drifting burst-clusters possess longer consecutive time intervals than downward drifting ones, suggesting distinct underlying physical mechanisms.

astro-ph.HE↗

arXiv2025

SpecBranch: Speculative Decoding via Hybrid Drafting and Rollback-Aware Branch Parallelism

Recently, speculative decoding (SD) has emerged as a promising technique to accelerate LLM inference by employing a small draft model to propose draft tokens in advance, and validating them in parallel with the large target model. However, the existing SD methods still remain fundamentally constrained by their serialized execution, which causes the mutual waiting bubbles between the draft and target models. To address this challenge, we draw inspiration from branch prediction in modern processors and propose a novel framework \textbf{SpecBranch} to unlock branch parallelism in SD. Specifically, we first take an in-depth analysis of the potential of branch parallelism in SD, and recognize that the key challenge lies in the trade-offs between parallelization and token rollback. Based on the analysis, we strategically introduce parallel speculative branches to preemptively hedge against likely rejections. Meanwhile, to enhance parallelism, we jointly orchestrate adaptive draft lengths with a hybrid combination of the implicit draft model confidence and explicit reusing of target model features. Extensive experiments across various models and benchmarks show that SpecBranch achieves over \textbf{1.8}$\times \sim$ \textbf{4.5}$\times$ speedups against the auto-regressive decoding and reduces rollback tokens by $\textbf{50}$\% for poorly aligned models, realizing its applicability for real-world deployments.

cs.DC↗

arXiv2025

Control Pneumatic Soft Bending Actuator with Online Learning Pneumatic Physical Reservoir Computing

The intrinsic nonlinearities of soft robots present significant control but simultaneously provide them with rich computational potential. Reservoir computing (RC) has shown effectiveness in online learning systems for controlling nonlinear systems such as soft actuators. Conventional RC can be extended into physical reservoir computing (PRC) by leveraging the nonlinear dynamics of soft actuators for computation. This paper introduces a PRC-based online learning framework to control the motion of a pneumatic soft bending actuator, utilizing another pneumatic soft actuator as the PRC model. Unlike conventional designs requiring two RC models, the proposed control system employs a more compact architecture with a single RC model. Additionally, the framework enables zero-shot online learning, addressing limitations of previous PRC-based control systems reliant on offline training. Simulations and experiments validated the performance of the proposed system. Experimental results indicate that the PRC model achieved superior control performance compared to a linear model, reducing the root-mean-square error (RMSE) by an average of over 37% in bending motion control tasks. The proposed PRC-based online learning control framework provides a novel approach for harnessing physical systems' inherent nonlinearities to enhance the control of soft actuators.

cs.RO↗

arXiv2024

DINO-X: A Unified Vision Model for Open-World Object Detection and Understanding

In this paper, we introduce DINO-X, which is a unified object-centric vision model developed by IDEA Research with the best open-world object detection performance to date. DINO-X employs the same Transformer-based encoder-decoder architecture as Grounding DINO 1.5 to pursue an object-level representation for open-world object understanding. To make long-tailed object detection easy, DINO-X extends its input options to support text prompt, visual prompt, and customized prompt. With such flexible prompt options, we develop a universal object prompt to support prompt-free open-world detection, making it possible to detect anything in an image without requiring users to provide any prompt. To enhance the model's core grounding capability, we have constructed a large-scale dataset with over 100 million high-quality grounding samples, referred to as Grounding-100M, for advancing the model's open-vocabulary detection performance. Pre-training on such a large-scale grounding dataset leads to a foundational object-level representation, which enables DINO-X to integrate multiple perception heads to simultaneously support multiple object perception and understanding tasks, including detection, segmentation, pose estimation, object captioning, object-based QA, etc. Experimental results demonstrate the superior performance of DINO-X. Specifically, the DINO-X Pro model achieves 56.0 AP, 59.8 AP, and 52.4 AP on the COCO, LVIS-minival, and LVIS-val zero-shot object detection benchmarks, respectively. Notably, it scores 63.3 AP and 56.5 AP on the rare classes of LVIS-minival and LVIS-val benchmarks, improving the previous SOTA performance by 5.8 AP and 5.0 AP. Such a result underscores its significantly improved capacity for recognizing long-tailed objects.

cs.CV↗

arXiv2024

Improving Initial Transients of Online Learning Echo State Network Control System with Feedback Adjustments

Echo state networks (ESNs) have become increasingly popular in online learning control systems due to their ease of training. However, online learning ESN controllers often suffer from slow convergence during the initial transient phase. Existing solutions, such as prior training, control mode switching, and incorporating plant dynamic approximations, have notable drawbacks, including undermining the system's online learning property or relying on prior knowledge of the controlled system. This work proposes a simple yet effective approach to address the slow initial convergence of online learning ESN control systems by integrating a feedback proportional-derivative (P-D) controller. Simulation results demonstrate that the proposed control system achieves rapid convergence during the initial transient phase and shows strong robustness against changes in the controlled system's dynamics and variations in the online learning model's hyperparameters. We show that the feedback controller accelerates convergence by guiding the online learning ESN to operate within a data range well-suited for learning. This study offers practical benefits for engineers aiming to implement online learning ESN control systems with fast convergence and easy deployment.

eess.SY↗

arXiv2024

Control Pneumatic Soft Bending Actuator with Feedforward Hysteresis Compensation by Pneumatic Physical Reservoir Computing

The nonlinearities of soft robots bring control challenges like hysteresis but also provide them with computational capacities. This paper introduces a fuzzy pneumatic physical reservoir computing (FPRC) model for feedforward hysteresis compensation in motion tracking control of soft actuators. Our method utilizes a pneumatic bending actuator as a physical reservoir with nonlinear computing capacities to control another pneumatic bending actuator. The FPRC model employs a Takagi-Sugeno (T-S) fuzzy logic to process outputs from the physical reservoir. The proposed FPRC model shows equivalent training performance to an Echo State Network (ESN) model, whereas it exhibits better test accuracies with significantly reduced execution time. Experiments validate the FPRC model's effectiveness in controlling the bending motion of a pneumatic soft actuator with open-loop and closed-loop control system setups. The proposed FPRC model's robustness against environmental disturbances has also been experimentally verified. To the authors' knowledge, this is the first implementation of a physical system in the feedforward hysteresis compensation model for controlling soft actuators. This study is expected to advance physical reservoir computing in nonlinear control applications and extend the feedforward hysteresis compensation methods for controlling soft actuators.

cs.RO↗

arXiv2023

Investigation on a Novel Length-Based Local Linear Subdivision Strategy for Triangular Meshes

Triangular meshes are a widely used representation in the field of 3D modeling. In this paper, we present a novel approach for edge length-based linear subdivision on triangular meshes, along with two auxiliary techniques. We conduct a comprehensive comparison of different subdivision methods in terms of computational capabilities and mesh-enhancing abilities. Our proposed approach demonstrates improved computational efficiency and generates fewer elements with higher quality compared to existing methods. The improvement in computational efficiency and mesh augmentation capability of our method is further enhanced when working with the two auxiliary techniques presented in this paper. Our novel strategy represents a significant contribution to the field and has important implications for local mesh refinement, computer-aided design, and isotropic remeshing.

cs.GR↗

arXiv2023

Two Degree of Freedom Adaptive Control for Hysteresis Compensation of Pneumatic Continuum Bending Actuator

Soft robotics, with their inherent flexibility and infinite degrees of freedom (DoF), offer promising advancements in human-machine interfaces. Particularly, pneumatic artificial muscles (PAMs) and pneumatic bending actuators have been fundamental in driving this evolution, capitalizing on their mimetic nature to natural muscle movements. However, with the versatility of these actuators comes the intricate challenge of hysteresis - a nonlinear phenomenon that hampers precise positioning, especially pronounced in pneumatic actuators due to gas compressibility. In this study, we introduce a novel 2-DoF adaptive control for precise bending tracking using a pneumatic continuum actuator. Notably, our control method integrates adaptability into both the feedback and the feedforward element, enhancing trajectory tracking in the presence of profound nonlinear effects. Comparative analysis with existing approaches underscores the superior tracking accuracy of our proposed strategy. This work discusses a new way of simple yet effective control designs for soft actuators with hysteresis properties.

cs.RO↗

arXiv2023

Trajectory Tracking Control of Dual-PAM Soft Actuator with Hysteresis Compensator

Soft robotics is a swiftly evolving field. Pneumatic actuators are suitable for driving soft robots because of their superior performance. However, their control is challenging due to the hysteresis characteristics. In response to this challenge, we propose an adaptive control method to compensate for the hysteresis of soft actuators. Employing a novel dual pneumatic artificial muscle (PAM) bending actuator, the innovative control approach abates hysteresis effects by dynamically modulating gains within a traditional PID controller corresponding to the predicted variation of the reference trajectory. Through experimental evaluation, we found that the proposed control method outperforms its conventional counterparts regarding tracking accuracy and response speed. Our work reveals a new direction for advancing model-free control in soft actuators.

cs.RO↗

arXiv2023

ChatHaruhi: Reviving Anime Character in Reality via Large Language Model

Role-playing chatbots built on large language models have drawn interest, but better techniques are needed to enable mimicking specific fictional characters. We propose an algorithm that controls language models via an improved prompt and memories of the character extracted from scripts. We construct ChatHaruhi, a dataset covering 32 Chinese / English TV / anime characters with over 54k simulated dialogues. Both automatic and human evaluations show our approach improves role-playing ability over baselines. Code and data are available at https://github.com/LC1332/Chat-Haruhi-Suzumiya .

cs.CL↗

arXiv2022

Size Does Matter: An Experimental Study of Anxiety in Virtual Reality

The emotional response of users induced by VR scenarios has become a topic of interest, however, whether changing the size of objects in VR scenes induces different levels of anxiety remains a question to be studied. In this study, we conducted an experiment to initially reveal how the size of a large object in a VR environment affects changes in participants' (N = 38) anxiety level and heart rate. To holistically quantify the size of large objects in the VR visual field, we used the omnidirectional field of view occupancy (OFVO) criterion for the first time to represent the dimension of the object in the participant's entire field of view. The results showed that the participants' heartbeat and anxiety while viewing the large objects were positively and significantly correlated to OFVO. These study reveals that the increase of object size in VR environments is accompanied by a higher degree of user's anxiety.

cs.MM↗

arXiv2022

CLS: Cross Labeling Supervision for Semi-Supervised Learning

It is well known that the success of deep neural networks is greatly attributed to large-scale labeled datasets. However, it can be extremely time-consuming and laborious to collect sufficient high-quality labeled data in most practical applications. Semi-supervised learning (SSL) provides an effective solution to reduce the cost of labeling by simultaneously leveraging both labeled and unlabeled data. In this work, we present Cross Labeling Supervision (CLS), a framework that generalizes the typical pseudo-labeling process. Based on FixMatch, where a pseudo label is generated from a weakly-augmented sample to teach the prediction on a strong augmentation of the same input sample, CLS allows the creation of both pseudo and complementary labels to support both positive and negative learning. To mitigate the confirmation bias of self-labeling and boost the tolerance to false labels, two different initialized networks with the same structure are trained simultaneously. Each network utilizes high-confidence labels from the other network as additional supervision signals. During the label generation phase, adaptive sample weights are assigned to artificial labels according to their prediction confidence. The sample weight plays two roles: quantify the generated labels' quality and reduce the disruption of inaccurate labels on network training. Experimental results on the semi-supervised classification task show that our framework outperforms existing approaches by large margins on the CIFAR-10 and CIFAR-100 datasets.

cs.CV↗