arXiv subjects

Jiachen Sun

Publications and source records attributed to Jiachen Sun.

At least 19 recordsLinked to original sources

arXiv2026

The Last Human-Written Paper: Agent-Native Research Artifacts

Scientific publication compresses a branching, iterative research process into a linear narrative, discarding the majority of what was discovered along the way. This compilation imposes two structural costs: a Storytelling Tax, where failed experiments, rejected hypotheses, and the branching exploration process are discarded to fit a linear narrative; and an Engineering Tax, where the gap between reviewer-sufficient prose and agent-sufficient specification leaves critical implementation details unwritten. Tolerable for human readers, these costs become critical when AI agents must understand, reproduce, and extend published work. We introduce the Agent-Native Research Artifact (ARA), a protocol that replaces the narrative paper with a machine-executable research package structured around four layers: scientific logic, executable code with full specifications, an exploration graph that preserves the failures compilation discards, and evidence grounding every claim in raw outputs. Three mechanisms support the ecosystem: a Live Research Manager that captures decisions and dead ends during ordinary development; an ARA Compiler that translates legacy PDFs and repos into ARAs; and an ARA-native review system that automates objective checks so human reviewers can focus on significance, novelty, and taste. On PaperBench and RE-Bench, ARA raises question-answering accuracy from 72.4% to 93.7% and reproduction success from 57.4% to 64.4%. On RE-Bench's five open-ended extension tasks, preserved failure traces in ARA accelerate progress, but can also constrain a capable agent from stepping outside the prior-run box depending on the agent's capabilities. Our code is open-sourced at https://github.com/Orchestra-Research/Agent-Native-Research-Artifact.

cs.LG↗

arXiv2026

YUHENG-OS: A Cloud-Native Space Cluster Operating System

As industry and academia continue to advance spaceborne computing and communication capabilities, the formation of cloud-native space clusters (CNSCs) has become an increasingly evident trend. This evolution progressively exposes the resource management challenges associated with coordinating fragmented and heterogeneous onboard resources while supporting large-scale and diverse space applications. However, directly transplanting mature terrestrial cloud-native cluster operating system paradigms into space is ineffective due to the fragmentation of spaceborne computing resources and satellite mobility, which collectively impose substantial challenges on resource awareness and orchestration. This article presents YUHENG-OS, a cloud-native space cluster operating system tailored for CNSCs. YUHENG-OS provides unified abstraction, awareness, and orchestration of heterogeneous spaceborne infrastructure, enabling cluster-wide task deployment and scheduling across distributed satellites. We introduce a four-layer system architecture and three key enabling technologies: modeling of heterogeneous resource demands for space tasks, fragmented heterogeneous resource awareness under network constraints, and matching of differentiated tasks with multidimensional heterogeneous resources under temporal dependency constraints. Evaluation results show that, compared with representative terrestrial cloud-native cluster operating systems exemplified by Kubernetes, YUHENG-OS achieves a substantially higher task completion ratio, with improvements of up to 98%. This advantage is primarily attributed to its ability to reduce resource awareness delay by 71%.

cs.NI↗

arXiv2026

CRISP: Compressed Reasoning via Iterative Self-Policy Distillation

Reasoning models often generate far more tokens than a task requires, which raises inference cost and can compound errors. We introduce CRISP (Compressed Reasoning via Iterative Self-Policy Distillation), an on-policy self-distillation method that teaches a model to reason more concisely by distilling its own concise behavior back into itself. The method uses a single idea: condition the same model on a "be concise" instruction to obtain teacher logits, then minimize the per-token reverse KL divergence between the student and this teacher on the student's own rollouts. It requires no ground-truth answers, no token budgets, and no difficulty estimators. The reverse-KL objective is naturally difficulty-adaptive: it compresses easy problems aggressively while preserving the reasoning steps that hard problems require. On Qwen3-14B, CRISP cuts reasoning length by up to 56% on MATH-500 and 38% on the harder AIME 2024, while improving MATH-500 accuracy by up to 3.3 points over the base model and holding AIME 2024 accuracy within about one point. This behavior generalizes across model sizes and families: Qwen3-8B shows the same compression with accuracy preserved, and DeepSeek-R1-Distill-Llama-8B improves accuracy on all five benchmarks while shortening its responses. General capabilities are preserved across all three models. Code is available at https://github.com/HJSang/OPSD_Reasoning_Compression.

cs.LG↗

arXiv2026

STARDIS: Strategic Scheduling and Deceptive Signaling for Satellite Intrusion Detection System Deployment

Satellite communication networks operate under stringent computational constraints and are susceptible to sophisticated cyberattacks. This paper introduces a novel defense framework that decouples security optimization into ground-based analysis and onboard real-time execution. In the long-term loop, the ground segment processes historical data to estimate key statistical parameters of the task environment. Additionally, we incorporate the time-varying characteristics of satellite wireless links to account for the dynamic communication context. In the short-term loop, the satellite employs a receding horizon optimization that models dynamic task arrivals and maximizes a utility function considering detection rates and resource costs. To counter intelligent adversaries interception, we introduce a deception mechanism using Bayesian persuasion theory. By strategically manipulating the short-term action sequences in the telemetry downlink, we mislead an external attacker's beliefs. We mathematically model the attacker's optimal response under channel uncertainty and demonstrate that our framework significantly reduces attacker utility. The approach's effectiveness is formally proven using Lyapunov theory.

cs.GT↗

arXiv2026

HulluEdit: Single-Pass Evidence-Consistent Subspace Editing for Mitigating Hallucinations in Large Vision-Language Models

Object hallucination in Large Vision-Language Models (LVLMs) significantly hinders their reliable deployment. Existing methods struggle to balance efficiency and accuracy: they often require expensive reference models and multiple forward passes, or apply static edits that risk suppressing genuine visual evidence. To address this, we introduce HulluEdit, a single-pass, reference-free intervention framework. Our core innovation is orthogonal subspace editing: we decompose the hidden states of the model into orthogonal subspaces - visual evidence, conflicting priors, and residual uncertainty - enabling selective suppression of hallucinatory patterns without interfering with visual grounding. This approach mathematically guarantees that edits applied to the prior subspace leave the visual component entirely unaffected. Extensive experiments show that HulluEdit achieves state-of-the-art hallucination reduction on benchmarks including POPE and CHAIR across diverse architectures, while preserving general capabilities on MME and maintaining efficient inference. Our method consistently outperforms contrastive decoding and static subspace editing baselines, offering a new pathway toward more trustworthy LVLMs.

cs.CV↗

arXiv2025

Heterogeneous Influence Maximization in User Recommendation

User recommendation systems enhance user engagement by encouraging users to act as inviters to interact with other users (invitees), potentially fostering information propagation. Conventional recommendation methods typically focus on modeling interaction willingness. Influence-Maximization (IM) methods focus on identifying a set of users to maximize the information propagation. However, existing methods face two significant challenges. First, recommendation methods fail to unleash the candidates' spread capability. Second, IM methods fail to account for the willingness to interact. To solve these issues, we propose two models named HeteroIR and HeteroIM. HeteroIR provides an intuitive solution to unleash the dissemination potential of user recommendation systems. HeteroIM fills the gap between the IM method and the recommendation task, improving interaction willingness and maximizing spread coverage. The HeteroIR introduces a two-stage framework to estimate the spread profits. The HeteroIM incrementally selects the most influential invitee to recommend and rerank based on the number of reverse reachable (RR) sets containing inviters and invitees. RR set denotes a set of nodes that can reach a target via propagation. Extensive experiments show that HeteroIR and HeteroIM significantly outperform the state-of-the-art baselines with the p-value < 0.05. Furthermore, we have deployed HeteroIR and HeteroIM in Tencent's online gaming platforms and gained an 8.5\% and 10\% improvement in the online A/B test, respectively. Implementation codes are available at https://github.com/socialalgo/HIM.

cs.IR↗

arXiv2025

Stance-Driven Multimodal Controlled Statement Generation: New Dataset and Task

Formulating statements that support diverse or controversial stances on specific topics is vital for platforms that enable user expression, reshape political discourse, and drive social critique and information dissemination. With the rise of Large Language Models (LLMs), controllable text generation towards specific stances has become a promising research area with applications in shaping public opinion and commercial marketing. However, current datasets often focus solely on pure texts, lacking multimodal content and effective context, particularly in the context of stance detection. In this paper, we formally define and study the new problem of stance-driven controllable content generation for tweets with text and images, where given a multimodal post (text and image/video), a model generates a stance-controlled response. To this end, we create the Multimodal Stance Generation Dataset (StanceGen2024), the first resource explicitly designed for multimodal stance-controllable text generation in political discourse. It includes posts and user comments from the 2024 U.S. presidential election, featuring text, images, videos, and stance annotations to explore how multimodal political content shapes stance expression. Furthermore, we propose a Stance-Driven Multimodal Generation (SDMG) framework that integrates weighted fusion of multimodal features and stance guidance to improve semantic consistency and stance control. We release the dataset and code (https://anonymous.4open.science/r/StanceGen-BE9D) for public use and further research.

cs.CL↗

arXiv2025

Minimax Rate-Optimal Inference for Individualized Quantile Treatment Effects in High-dimensional Models

The quantification of treatment effects plays an important role in a wide range of applications, including policy making and bio-pharmaceutical research. In this article, we study the quantile treatment effect (QTE) while addressing two specific types of heterogeneities: (a) personalized heterogeneity, which captures the varying treatment effects for different individuals, and (b) quantile heterogeneity, which accounts for how the impact of covariates varies across different quantile levels. A well-designed debiased estimator for the individualized quantile treatment effect (IQTE) is proposed to capture such heterogeneities effectively. We show that this estimator converges weakly to a Gaussian process as a function of the quantile levels and propose valid statistical inference methods, including the construction of confidence intervals and the development of hypothesis testing decision rules. In addition, the minimax optimality frameworks for these inference procedures are established. Specifically, we derive the minimax optimal rates for the expected length of confidence intervals and the magnitude of the detection boundary for hypothesis testing procedures, illustrating the superiority of the proposed estimator. The effectiveness of our methods is demonstrated through extensive simulations and an analysis of the National Health and Nutrition Examination Survey (NHANES) datasets.

math.ST↗

arXiv2025

Space-Based Computing Networks: Trends, Architecture, Challenges, and Key Technologies

As one of the most promising hotspots in the 6G era, space remote sensing information networks play a key and irreplaceable role in areas such as emergency response and scientific research, and are expected to foster remote sensing data processing into the next generation of killer applications. However, due to the inability to deploy ground communication stations at scale and the limited satellite-to-ground link rate, the traditional model for transmitting space data back to ground stations faces significant challenges in terms of timeliness. To address this problem, we focus on the emerging paradigm of on-orbit space data processing, which reduces the volume of transmitted data by several orders of magnitude to enable faster task response, taking the first step toward building a space-based computing network. Specifically, we propose a hierarchical space-based computing network architecture, comprising the space-based cloud constellation system, the remote sensing constellation system, the network operation control center, the orchestration data center, and the user access portal. Each component is described in detail from a system design perspective to clarify its specific role and functionality. Next, we analyze three scientific challenges: the heterogeneous resource virtualization and state information synchronization, the matching of multi-priority tasks with multidimensional resources, and the fault detection and localization under extreme conditions. Finally, we discuss key technologies to address the aforementioned challenges and highlight promising research priorities for the future.

cs.NI↗

arXiv2025

Optimizing GPT for Video Understanding: Zero-Shot Performance and Prompt Engineering

In this study, we tackle industry challenges in video content classification by exploring and optimizing GPT-based models for zero-shot classification across seven critical categories of video quality. We contribute a novel approach to improving GPT's performance through prompt optimization and policy refinement, demonstrating that simplifying complex policies significantly reduces false negatives. Additionally, we introduce a new decomposition-aggregation-based prompt engineering technique, which outperforms traditional single-prompt methods. These experiments, conducted on real industry problems, show that thoughtful prompt design can substantially enhance GPT's performance without additional finetuning, offering an effective and scalable solution for improving video classification.

cs.CV↗

arXiv2025

Spreading dynamics of information on online social networks

Social media is profoundly changing our society with its unprecedented spreading power. Due to the complexity of human behaviors and the diversity of massive messages, the information spreading dynamics are complicated, and the reported mechanisms are different and even controversial. Based on data from mainstream social media platforms, including WeChat, Weibo, and Twitter, cumulatively encompassing a total of 7.45 billion users, we uncover a ubiquitous mechanism that the information spreading dynamics are basically driven by the interplay of social reinforcement and social weakening effects. Accordingly, we propose a concise equation, which, surprisingly, can well describe all the empirical large-scale spreading trajectories. Our theory resolves a number of controversial claims and satisfactorily explains many phenomena previously observed. It also reveals that the highly clustered nature of social networks can lead to rapid and high-frequency information bursts with relatively small coverage per burst. This vital feature enables social media to have a high capacity and diversity for information dissemination, beneficial for its ecological development.

physics.soc-ph↗

arXiv2024

Cocoon: Robust Multi-Modal Perception with Uncertainty-Aware Sensor Fusion

An important paradigm in 3D object detection is the use of multiple modalities to enhance accuracy in both normal and challenging conditions, particularly for long-tail scenarios. To address this, recent studies have explored two directions of adaptive approaches: MoE-based adaptive fusion, which struggles with uncertainties arising from distinct object configurations, and late fusion for output-level adaptive fusion, which relies on separate detection pipelines and limits comprehensive understanding. In this work, we introduce Cocoon, an object- and feature-level uncertainty-aware fusion framework. The key innovation lies in uncertainty quantification for heterogeneous representations, enabling fair comparison across modalities through the introduction of a feature aligner and a learnable surrogate ground truth, termed feature impression. We also define a training objective to ensure that their relationship provides a valid metric for uncertainty quantification. Cocoon consistently outperforms existing static and adaptive methods in both normal and challenging conditions, including those with natural and artificial corruptions. Furthermore, we show the validity and efficacy of our uncertainty metric across diverse datasets.

cs.CV↗

arXiv2024

ASFT: Aligned Supervised Fine-Tuning through Absolute Likelihood

Direct Preference Optimization (DPO) is a method for enhancing model performance by directly optimizing for the preferences or rankings of outcomes, instead of traditional loss functions. This approach has proven effective in aligning Large Language Models (LLMs) with human preferences. Despite its widespread use across various tasks, DPO has been criticized for its sensitivity to the effectiveness of Supervised Fine-Tuning (SFT) and its limitations in enabling models to learn human-preferred responses, leading to less satisfactory performance. To address these limitations, we propose Aligned Supervised Fine-Tuning (ASFT), an effective approach that better aligns LLMs with pair-wise datasets by optimizing absolute likelihood for each response, rather than using the Bradley-Terry model, and eliminates the need for a reference model. Through theoretical gradient analysis, we demonstrate that ASFT mitigates the issue where the DPO loss function decreases the probability of generating human-dispreferred data at a faster rate than it increases the probability of producing preferred data. Additionally, we compare ASFT to DPO and its latest variants, such as the single-step approach ORPO, using the latest instruction-tuned model Llama3, which has been fine-tuned on UltraFeedback and HH-RLHF. We evaluated performance on instruction-following benchmarks like MT-Bench and traditional text generation metrics such as BLEU-4 and ROUGE-L. Extensive experiments demonstrate that ASFT is an effective alignment approach, consistently outperforming existing methods.

cs.LG↗

arXiv2024

ELFS: Label-Free Coreset Selection with Proxy Training Dynamics

High-quality human-annotated data is crucial for modern deep learning pipelines, yet the human annotation process is both costly and time-consuming. Given a constrained human labeling budget, selecting an informative and representative data subset for labeling can significantly reduce human annotation effort. Well-performing state-of-the-art (SOTA) coreset selection methods require ground truth labels over the whole dataset, failing to reduce the human labeling burden. Meanwhile, SOTA label-free coreset selection methods deliver inferior performance due to poor geometry-based difficulty scores. In this paper, we introduce ELFS (Effective Label-Free Coreset Selection), a novel label-free coreset selection method. ELFS significantly improves label-free coreset selection by addressing two challenges: 1) ELFS utilizes deep clustering to estimate training dynamics-based data difficulty scores without ground truth labels; 2) Pseudo-labels introduce a distribution shift in the data difficulty scores, and we propose a simple but effective double-end pruning method to mitigate bias on calculated scores. We evaluate ELFS on four vision benchmarks and show that, given the same vision encoder, ELFS consistently outperforms SOTA label-free baselines. For instance, when using SwAV as the encoder, ELFS outperforms D2 by up to 10.2% in accuracy on ImageNet-1K. We make our code publicly available on GitHub.

cs.CV↗

arXiv2024

Safeguarding Vision-Language Models Against Patched Visual Prompt Injectors

Large language models have become increasingly prominent, also signaling a shift towards multimodality as the next frontier in artificial intelligence, where their embeddings are harnessed as prompts to generate textual content. Vision-language models (VLMs) stand at the forefront of this advancement, offering innovative ways to combine visual and textual data for enhanced understanding and interaction. However, this integration also enlarges the attack surface. Patch-based adversarial attack is considered the most realistic threat model in physical vision applications, as demonstrated in many existing literature. In this paper, we propose to address patched visual prompt injection, where adversaries exploit adversarial patches to generate target content in VLMs. Our investigation reveals that patched adversarial prompts exhibit sensitivity to pixel-wise randomization, a trait that remains robust even against adaptive attacks designed to counteract such defenses. Leveraging this insight, we introduce SmoothVLM, a defense mechanism rooted in smoothing techniques, specifically tailored to protect VLMs from the threat of patched visual prompt injectors. Our framework significantly lowers the attack success rate to a range between 0% and 5.0% on two leading VLMs, while achieving around 67.3% to 95.0% context recovery of the benign images, demonstrating a balance between security and usability.

cs.CV↗

arXiv2024

Reconstructing the evolution history of networked complex systems

The evolution processes of complex systems carry key information in the systems' functional properties. Applying machine learning algorithms, we demonstrate that the historical formation process of various networked complex systems can be extracted, including protein-protein interaction, ecology, and social network systems. The recovered evolution process has demonstrations of immense scientific values, such as interpreting the evolution of protein-protein interaction network, facilitating structure prediction, and particularly revealing the key co-evolution features of network structures such as preferential attachment, community structure, local clustering, degree-degree correlation that could not be explained collectively by previous theories. Intriguingly, we discover that for large networks, if the performance of the machine learning model is slightly better than a random guess on the pairwise order of links, reliable restoration of the overall network formation process can be achieved. This suggests that evolution history restoration is generally highly feasible on empirical networks.

physics.soc-ph↗

arXiv2023

Dolphins: Multimodal Language Model for Driving

The quest for fully autonomous vehicles (AVs) capable of navigating complex real-world scenarios with human-like understanding and responsiveness. In this paper, we introduce Dolphins, a novel vision-language model architected to imbibe human-like abilities as a conversational driving assistant. Dolphins is adept at processing multimodal inputs comprising video (or image) data, text instructions, and historical control signals to generate informed outputs corresponding to the provided instructions. Building upon the open-sourced pretrained Vision-Language Model, OpenFlamingo, we first enhance Dolphins's reasoning capabilities through an innovative Grounded Chain of Thought (GCoT) process. Then we tailored Dolphins to the driving domain by constructing driving-specific instruction data and conducting instruction tuning. Through the utilization of the BDD-X dataset, we designed and consolidated four distinct AV tasks into Dolphins to foster a holistic understanding of intricate driving scenarios. As a result, the distinctive features of Dolphins are characterized into two dimensions: (1) the ability to provide a comprehensive understanding of complex and long-tailed open-world driving scenarios and solve a spectrum of AV tasks, and (2) the emergence of human-like capabilities including gradient-free instant adaptation via in-context learning and error recovery via reflection.

cs.CV↗

arXiv2023

CALICO: Self-Supervised Camera-LiDAR Contrastive Pre-training for BEV Perception

Perception is crucial in the realm of autonomous driving systems, where bird's eye view (BEV)-based architectures have recently reached state-of-the-art performance. The desirability of self-supervised representation learning stems from the expensive and laborious process of annotating 2D and 3D data. Although previous research has investigated pretraining methods for both LiDAR and camera-based 3D object detection, a unified pretraining framework for multimodal BEV perception is missing. In this study, we introduce CALICO, a novel framework that applies contrastive objectives to both LiDAR and camera backbones. Specifically, CALICO incorporates two stages: point-region contrast (PRC) and region-aware distillation (RAD). PRC better balances the region- and scene-level representation learning on the LiDAR modality and offers significant performance improvement compared to existing methods. RAD effectively achieves contrastive distillation on our self-trained teacher model. CALICO's efficacy is substantiated by extensive evaluations on 3D object detection and BEV map segmentation tasks, where it delivers significant performance improvements. Notably, CALICO outperforms the baseline method by 10.5% and 8.6% on NDS and mAP. Moreover, CALICO boosts the robustness of multimodal 3D object detection against adversarial attacks and corruption. Additionally, our framework can be tailored to different backbones and heads, positioning it as a promising approach for multimodal BEV perception.

cs.CV↗