arXiv subjects

Qichao Ma

Publications and source records attributed to Qichao Ma.

10 recordsLinked to original sources

arXiv2026

Graph Engineering in the Era of LLM Agents: From Individual Intelligence to System Intelligence

LLMs have evolved from language generators to autonomous agents capable of complex, long-horizon tasks. This evolution has produced paradigms including Prompt Engineering to elicit model capabilities, Context Engineering to manage information access, Harness Engineering to organize external tools and resources, and Loop Engineering to support continual reflection and self-improvement. Yet as tasks grow more complex, individual intelligence faces a fundamental limit: many tasks require heterogeneous expertise, interdependent subtasks, parallel execution, independent verification, and persistent state, exceeding any single agent's organizational capacity. Augmenting one agent's capabilities or context cannot resolve this architectural mismatch; intelligence must instead be distributed across specialized agents and organized at the system level. We call this System Intelligence: an agent system's ability to organize and coordinate multiple intelligent components into a coherent, adaptive whole pursuing a shared objective. Achieving it requires more than adding agents; it demands explicit structures to organize work, coordinate heterogeneous agents, and maintain evolving execution states. We introduce Graph Engineering, an emerging paradigm for next-generation agent systems. Unlike prior paradigms that mainly optimize individual interactions or agent-level behavior, Graph Engineering constructs explicit, dynamic, evolving graph structures representing tasks, agents, and system states. These abstractions provide a unified foundation for organizing complex objectives, orchestrating heterogeneous agents, modeling system dynamics, and enabling scalable agent evolution. We systematically review the principles, methodologies, and applications of Graph Engineering for LLM agents. Related papers, open-source data, and projects are collected at https://github.com/DEEP-JLU/Awesome-Graph-Engineering.

cs.IR↗

arXiv2026

RecGPT-Mobile-V2 Technical Report

Personalized Query prediction maps implicit behavioral signals---clicks, favorites, purchases, and post-purchase exploration---to explicit retrieval intent. On-device deployment makes this task particularly challenging: behavioral trajectories are noisy and multi-scale, multiple Queries may be valid for a single trajectory, and a uniform reasoning policy either expends unnecessary computation on simple instances or allocates insufficient capacity to complex ones. We introduce RecGPT-Mobile-V2, an end-to-end framework that treats intent quality and execution efficiency as coupled objectives within a staged design. The framework transforms heterogeneous interactions into an evidence-preserving trajectory, establishes a recommendation-native foundation through domain adaptation and supervised alignment, and applies reasoning-cost optimization only after grouped rollouts meet grounding and utility criteria. The resulting teacher is distilled into a compact student deployed with low-bit execution, structured compression, and budget-aware device--cloud routing. In an aligned CoT ablation, an evidence-focused short rationale increases ROUGE-L from 0.228 to 0.315 and Jaccard from 0.174 to 0.248, while slightly outperforming the full five-stage rationale. In the controlled RL comparison, the complete reward formulation improves Query quality from 73.2% under quality-only RL to 78.6%, lowers the hard-failure rate from 3.6% to 1.6%, and reduces the median CoT length from 62 to 14 tokens. Online retrieval analysis further indicates that the Query recall channel retrieves inventory complementary to that surfaced by established recall channels. Collectively, these findings support sufficiency-oriented rather than uniformly short reasoning: retain decision-relevant evidence and allocate additional computation only when it is likely to improve the predicted Query.

cs.IR↗

arXiv2026

Can We Perform Online RL for Image Editing without Editing Rewards?

Reinforcement learning (RL) enables direct preference optimization for image editing through editing-specific rewards, which remain less developed due to costly triplet supervision and complex task-dependent calibration. In contrast, text-to-image (T2I) generation benefits from a mature and diverse reward ecosystem spanning semantic alignment, aesthetics, realism, glyph shape, and other visual preferences. Extending this ecosystem to image editing would substantially broaden the range of visual preferences accessible to RL-based optimization, prompting the central question: \emph{Can We Perform Image Editing RL without Editing Rewards?} In this paper, we argue that the standard image editing dimensions have potential to be mapped to the T2I reward space: image quality can transfer directly, prompt following can be aligned through a description of the desired visual state, and reference consistency admits a coarse semantic conversion by encoding the source content to preserve. However, editing instructions specify relative changes, whereas T2I rewards require self-contained target descriptions; moreover, semantically valid captions from generic vision-language models may be incompatible with the frozen reward. Hence, we further introduce Lever-Edit, a two-stage framework that learns a reward-aligned captioner for counterfactual target descriptions, freezes it, and optimizes the editing policy solely with the transferred T2I reward. Experiments show competitive editing alignment and source preservation against editing-reward-based fine-tuning, while outperforming intuitive transfer baselines.

cs.CV↗

arXiv2026

RecGPT-Mobile: On-Device Large Language Models for User Intent Understanding in Taobao Feed Recommendation

Predicting a user's next search query from recent interaction behaviors is a critical problem in modern e-commerce systems, particularly in scenarios where user intent evolves rapidly. Large Language Models (LLMs) offer strong semantic reasoning capabilities and have recently been adopted to enhance training data construction for next-query prediction. However, due to resource constraints on mobile devices, existing applications are deployed on cloud servers, resulting in high inference costs. In this paper, we propose RecGPT-Mobile, a framework that designs a lightweight LLM-based intent understanding agent to improve recommendation quality in mobile e-commerce scenarios. By deploying LLMs directly on mobile devices, our approach can capture evolving interests of users more quickly and adjust the recommendation results in real time. Extensive offline analyses and online experiments demonstrate that our method significantly improves the accuracy of recommendation results, laying a practical path for LLM deployment in production-scale recommendation systems on mobile devices, as well as a scalable solution for integrating LLMs into real-world next-query prediction systems.

cs.IR↗

arXiv2026

Inner-Probe: Discovering Copyright-related Data Generation in LLM Architecture

Large Language Models (LLMs) utilize extensive knowledge databases and show powerful text generation ability. However, their reliance on high-quality copyrighted datasets raises concerns about copyright infringements in generated texts. Current research often employs prompt engineering or semantic classifiers to identify copyrighted content, but these approaches have two significant limitations: (1) Challenging to identify which specific subdataset (e.g., works from particular authors) influences an LLM's output. (2) Treating the entire training database as copyrighted, hence overlooking the inclusion of non-copyrighted training data. We propose Inner-Probe, a lightweight framework designed to evaluate the influence of copyrighted sub-datasets on LLM-generated texts. Unlike traditional methods relying solely on text, we discover that the results of multi-head attention (MHA) during LLM output generation provide more effective information. Thus, Inner-Probe performs sub-dataset contribution analysis using a lightweight LSTM based network trained on MHA results in a supervised manner. Harnessing such a prior, Inner-Probe enables non-copyrighted text detection through a concatenated global projector trained with unsupervised contrastive learning. Inner-Probe demonstrates 3x improved efficiency compared to semantic model training in sub-dataset contribution analysis on Books3, achieves 15.04% - 58.7% higher accuracy over baselines on the Pile, and delivers a 0.104 increase in AUC for non-copyrighted data filtering.

cs.CL↗

arXiv2025

Faster and Stronger: When ANN-SNN Conversion Meets Parallel Spiking Calculation

Spiking Neural Network (SNN), as a brain-inspired and energy-efficient network, is currently facing the pivotal challenge of exploring a suitable and efficient learning framework. The predominant training methodologies, namely Spatial-Temporal Back-propagation (STBP) and ANN-SNN Conversion, are encumbered by substantial training overhead or pronounced inference latency, which impedes the advancement of SNNs in scaling to larger networks and navigating intricate application domains. In this work, we propose a novel parallel conversion learning framework, which establishes a mathematical mapping relationship between each time-step of the parallel spiking neurons and the cumulative spike firing rate. We theoretically validate the lossless and sorting properties of the conversion process, as well as pointing out the optimal shifting distance for each step. Furthermore, by integrating the above framework with the distribution-aware error calibration technique, we can achieve efficient conversion towards more general activation functions or training-free circumstance. Extensive experiments have confirmed the significant performance advantages of our method for various conversion cases under ultra-low time latency. To our best knowledge, this is the first work which jointly utilizes parallel spiking calculation and ANN-SNN Conversion, providing a highly promising approach for SNN supervised training. Code is available at https://github.com/hzc1208/Parallel_Conversion.

cs.NE↗

arXiv2024

PE-Planner: A Performance-Enhanced Quadrotor Motion Planner for Autonomous Flight in Complex and Dynamic Environments

The role of a motion planner is pivotal in quadrotor applications, yet existing methods often struggle to adapt to complex environments, limiting their ability to achieve fast, safe, and robust flight. In this letter, we introduce a performance-enhanced quadrotor motion planner designed for autonomous flight in complex environments including dense obstacles, dynamic obstacles, and unknown disturbances. The global planner generates an initial trajectory through kinodynamic path searching and refines it using B-spline trajectory optimization. Subsequently, the local planner takes into account the quadrotor dynamics, estimated disturbance, global reference trajectory, control cost, time cost, and safety constraints to generate real-time control inputs, utilizing the framework of model predictive contouring control. Both simulations and real-world experiments corroborate the heightened robustness, safety, and speed of the proposed motion planner. Additionally, our motion planner achieves flights at more than 6.8 m/s in a challenging and complex racing scenario.

cs.RO↗

arXiv2023

SRL-ORCA: A Socially Aware Multi-Agent Mapless Navigation Algorithm In Complex Dynamic Scenes

For real-world navigation, it is important to endow robots with the capabilities to navigate safely and efficiently in a complex environment with both dynamic and non-convex static obstacles. However, achieving path-finding in non-convex complex environments without maps as well as enabling multiple robots to follow social rules for obstacle avoidance remains challenging problems. In this letter, we propose a socially aware robot mapless navigation algorithm, namely Safe Reinforcement Learning-Optimal Reciprocal Collision Avoidance (SRL-ORCA). This is a multi-agent safe reinforcement learning algorithm by using ORCA as an external knowledge to provide a safety guarantee. This algorithm further introduces traffic norms of human society to improve social comfort and achieve cooperative avoidance by following human social customs. The result of experiments shows that SRL-ORCA learns strategies to obey specific traffic rules. Compared to DRL, SRL-ORCA shows a significant improvement in navigation success rate in different complex scenarios mixed with the application of the same training network. SRL-ORCA is able to cope with non-convex obstacle environments without falling into local minimal regions and has a 14.1\% improvement in path quality (i.e., the average time to target) compared to ORCA. Videos are available at https://youtu.be/huhXfCDkGws.

cs.RO↗

arXiv2023

Exponential Consensus of Multiple Agents over Dynamic Network Topology: Controllability, Connectivity, and Compactness

This paper investigates the problem of securing exponentially fast consensus (exponential consensus for short) for identical agents with finite-dimensional linear system dynamics over dynamic network topology. Our aim is to find the weakest possible conditions that guarantee exponentially fast consensus using a Lyapunov function consisting of a sum of terms of the same functional form. We first investigate necessary conditions, starting by examining the system (both agent and network) parameters. It is found that controllability of the linear agents is necessary for reaching consensus. Then, to work out necessary conditions incorporating the network topology, we construct a set of Laplacian matrix-valued functions. The precompactness of this set of functions is shown to be a significant generalization of existing assumptions on network topology, including the common assumption that the edge weights are bounded piecewise constant functions or continuous functions. With the aid of such a precompactness assumption and restricting the Lyapunov function to one consisting of a sum of terms of the same functional form, we prove that a joint $(δ, T)$-connectivity condition on the network topology is necessary for exponential consensus. Finally, we investigate how the above two ``necessities'' work together to guarantee exponential consensus. To partially address this problem, we define a synchronization index to characterize the interplay between agent parameters and network topology. Based on this notion, it is shown that by designing a proper feedback matrix and under the precompactness assumption, exponential consensus can be reached globally and uniformly if the joint $(δ,T)$-connectivity and controllability conditions are satisfied, and the synchronization index is not less than one.

eess.SY↗

arXiv2018

On Synchronization of Dynamical Systems over Directed Switching Topologies: An Algebraic and Geometric Perspective

In this paper, we aim to investigate the synchronization problem of dynamical systems, which can be of generic linear or Lipschitz nonlinear type, communicating over directed switching network topologies. A mild connectivity assumption on the switching topologies is imposed, which allows them to be directed and jointly connected. We propose a novel analysis framework from both algebraic and geometric perspectives to justify the attractiveness of the synchronization manifold. Specifically, it is proven that the complementary space of the synchronization manifold can be spanned by certain subspaces. These subspaces can be the eigenspaces of the nonzero eigenvalues of Laplacian matrices in linear case. They can also be subspaces in which the projection of the nonlinear self-dynamics still retains the Lipschitz property. This allows to project the states of the dynamical systems into these subspaces and transform the synchronization problem under consideration equivalently into a convergence one of the projected states in each subspace. Then, assuming the joint connectivity condition on the communication topologies, we are able to work out a simple yet effective and unified convergence analysis for both types of dynamical systems. More specifically, for partial-state coupled generic linear systems, it is proven that synchronization can be reached if an extra condition, which is easy to verify in several cases, on the system dynamics is satisfied. For Lipschitz-type nonlinear systems with positive-definite inner coupling matrix, synchronization is realized if the coupling strength is strong enough to stabilize the evolution of the projected states in each subspace under certain conditions.

eess.SY↗