arXiv subjects

Mingju Chen

Publications and source records attributed to Mingju Chen.

11 recordsLinked to original sources

arXiv2026

A-MapReduce: Executing Wide Search via Agentic MapReduce

Contemporary large language model (LLM)-based multi-agent systems exhibit systematic advantages in deep research tasks, which emphasize iterative, vertically structured information seeking. However, when confronted with wide search tasks characterized by large-scale, breadth-oriented retrieval, existing agentic frameworks, primarily designed around sequential, vertically structured reasoning, remain stuck in expansive search objectives and inefficient long-horizon execution. To bridge this gap, we propose A-MapReduce, a MapReduce paradigm-inspired multi-agent execution framework that recasts wide search as a horizontally structured retrieval problem. Concretely, A-MapReduce implements parallel processing of massive retrieval targets through task-adaptive decomposition and structured result aggregation. Meanwhile, it leverages experiential memory to drive the continual evolution of query-conditioned task allocation and recomposition, enabling progressive improvement in large-scale wide-search regimes. Extensive experiments on five agentic benchmarks demonstrate that A-MapReduce is (i) high-performing, achieving state-of-the-art performance on WideSearch and DeepWideSearch, and delivering 5.11% - 17.50% average Item F1 improvements compared with strong baselines with OpenAI o3 or Gemini 2.5 Pro backbones; (ii) cost-effective and efficient, delivering superior cost-performance trade-offs and reducing running time by 45.8\% compared to representative multi-agent baselines. The code is available at https://github.com/mingju-c/AMapReduce.

cs.MA↗

arXiv2026

TRCA: Transition-wise Rubric Credit Assignment for Long-horizon LLM Agents

Long-horizon large language model (LLM) agents are typically optimized with sparse terminal outcomes, making fine-grained credit assignment across multi-step interactions difficult. Existing approaches either rely on process evaluators, which incur annotation and inference costs, or derive step-level credit from successful trajectories. However, successful trajectories are extremely scarce during early-stage reinforcement learning, substantially weakening anchor-based methods. We propose Transition-wise Rubric Credit Assignment (TRCA), which derives step-level supervision directly from action-induced transitions without learned evaluators or successful anchors. TRCA evaluates each transition using Evidence, Execution, and Invalidity rubrics to capture task-relevant information acquisition, valid task execution, and invalid or regressive behavior. From these judgments, Foundational Rubric Reward measures local transition quality, while Breakthrough Rubric Reward tracks newly covered Evidence and Execution conditions to reward incremental task progress. Combined with terminal outcomes, these signals produce fine-grained step-level advantages for policy optimization. Experiments on ALFWorld, WebShop, and seven search-augmented question-answering benchmarks show consistent improvements over the evaluated baselines. With Qwen2.5-7B-Instruct, TRCA improves the WebShop score by 6.0%-12.6%; with Qwen2.5-3B-Instruct, it improves the average SearchQA score by 1.9%-18.3%. These results demonstrate the effectiveness of transition-wise rubric credit assignment for long-horizon tasks with sparse successful anchors.

cs.AI↗

arXiv2026

RSIFlow: Adaptive Self-Improvement Workflows with an Evolving Meta-Agent

Large language model (LLM) agents have demonstrated strong capabilities in complex tasks, including code generation, tool use, and multi-hop question answering. Existing self-improvement methods enhance agent performance through model optimization, harness evolution, and executable artifact refinement, while recursive self-improvement (RSI) further extends optimization to the improvement mechanisms themselves. However, adapting self-improvement workflows to evolving performance bottlenecks while improving workflow orchestration through execution feedback remains a key challenge. We propose \textbf{RSIFlow}, a \textbf{R}ecursive \textbf{S}elf-\textbf{I}mprovement framework that integrates adaptive work\textbf{flow} orchestration with meta-agent capability evolution. RSIFlow diagnoses performance bottlenecks from task-agent trajectories and environmental feedback to dynamically select component-specific improvement workflows. The meta-agent then distills workflow outcomes into component-specific skills and cross-component principles, evolving its own harness to improve subsequent workflow orchestration. Experiments across six benchmarks and two backbone models show that RSIFlow achieves the SOTA average performance on all benchmarks. Ablation studies further demonstrate the contributions of adaptive workflow orchestration and meta-harness evolution. Code is available at https://github.com/1HUAN1/RSI-Flow.

cs.AI↗

arXiv2026

AGAR: a reinforcement learning substrate for LLM program evolution

Given a task and an evaluator, a language model can rewrite a candidate program while a search loop decides which rewrites survive, offering a practical route to algorithm discovery. But that loop is governed by five constants set by hand: which parent to select, how hard to mutate, how to keep diversity, what to remember, and a scalar score that never says which part of the program earned it. Reinforcement learning already has an estimator for each. The obstacle is that program evolution is not usually written down as a decision process. We formalize it as a Markov decision process whose action is the modular prefix the model is conditioned on, rather than the program it emits. Credit assignment, value estimation, adaptive exploration, and experience memory can then attach to distinct components. AGAR (Algorithm Generation As RL) provides the resulting substrate: any estimator can be replaced or switched off without changing the controller, making the transfer auditable one mechanism at a time, with no gradient training of the backend model. Across 19 tasks, two backends, and three seeds under one harness, AGAR improves on the stronger of two published baselines on most tasks, with gains concentrated in the competitive-programming family. The formalization also yields a checkable reading of prior work: these systems are implicitly zero-discount, not by choice, but because fitness is exogenous to an individual rather than a return over successors, leaving a discount factor nothing to act on.

cs.AI↗

arXiv2026

RLDISCOVER: LLM-driven co-evolution of reinforcement learning algorithms

LLM-guided program evolution has enabled discoveries in mathematics and computational optimization, raising the prospect of reinforcement learning (RL) algorithms that self-evolve to improve how agents learn. However, realizing this prospect faces two obstacles. Joint search over coupled algorithmic components is difficult to scale: simultaneous changes can disrupt learning, while isolated changes overlook their dependencies. Evaluating candidate algorithms also requires costly training, with fitness remaining uncertain across random seeds. We introduce RLDiscover, a framework for the self-evolution of model-free deep RL algorithms. Progressive Co-Evolution advances from targeted component edits to joint evolution, while Progressive Probabilistic Evaluation balances search breadth and evaluation fidelity through staged training and repeated evaluation. Experiments across SAC, PPO, and DQN on four benchmark suites show substantial improvements in mean return, with per-family median gains of 32%-84% and a peak return ratio of approximately 363x over a near-zero baseline. These gains include transitions from failed learning to successful task completion, and improvements persist when evolution starts from stronger open-source implementations. On measured SAC locomotion runs, evaluation uses approximately one-fifteenth the estimated compute required to fully evaluate the same candidate pool. Remarkably, independent searches repeatedly discover interpretable combinations of adaptive robust losses, progress-dependent value targets, and running statistics, with selected programs transferring to unseen tasks. These findings point toward a broader role for self-evolution in AI: discovering interpretable algorithms that improve how agents learn.

cs.AI↗

arXiv2026

Beyond Timestamps: Decision-Aligned On-Policy Distillation for Long-Horizon Agents

Reinforcement learning with verifiable rewards (RLVR) often relies on sparse outcome rewards, providing coarse supervision for long-horizon agents. On-policy self-distillation (OPSD) complements this signal with dense privileged feedback. However, we identify \emph{Decision--Timestamp Mismatch}: privileged guidance may be misaligned with the student's functional decision because the corresponding decision can occur at a different timestep, while the student's decision itself may span multiple timesteps rather than being tied to a single timestamp. Thus, timestamp-local supervision can misalign both the context and the temporal scope of credit. To address this mismatch, we introduce \textsc{AlignOPSD}, following the principle of aligning supervision before assigning credit. Decision-Aligned Supervision Rectification re-scores the same student-sampled response in functionally matched contexts across sibling rollouts to calibrate local teacher evidence. Semi-Markov Hierarchical Credit Assignment then derives variable-duration decision spans from correspondence changes and uses rectified evidence to allocate outcome-grounded credit across spans and their constituent turns. We evaluate \textsc{AlignOPSD} with Qwen2.5-3B and Qwen2.5-7B on ALFWorld, WebShop, and Search-QA against representative baselines. \textsc{AlignOPSD} outperforms both GRPO and StepOPSD across all eight backbone--aggregate-metric comparisons, improving on GRPO by 5.5--8.7 \% and ranking first in six. Additional analyzes examine the two alignment stages and hyperparameter sensitivity between tasks. Our code is avaliable at https://github.com/mingju-c/Align-OPSD

cs.LG↗

arXiv2026

Centering Drives Normalization Gains: Price-Offset Nuisances in Cross-Sectional Return Prediction

Cross-sectional return prediction from raw intraday bars is sensitive to each instrument price level, an additive nuisance under a return-ranking hypothesis. We test whether removing this offset, rather than rescaling amplitudes or changing the encoder, explains gains on a point-in-time CSI~300 five-minute panel. We evaluate eight parameter-matched encoders with and without RevIN normalization; a parameter-free ladder then separates identity, scale-only, centering, last-value referencing, differencing, and standardization across all fields and restricted channels. Centering drives the reliable effect, while scale-only normalization does not help. All eight paired effects are positive and survive Holm correction on raw rank IC, after style residualization, and after further residualizing on short-term reversal. Among six stronger encoders, gains of 0.0376-0.0567 exceed the 0.0109 spread of normalized IC (0.0830-0.0939). Price-only standardization retains 93--101% of the all-field gain. These results place the main effect in transformed price-channel offset removal rather than amplitude scaling or encoder choice.

cs.CE↗

arXiv2026

A Compact Selective State-Space Model for Cross-Sectional Stock Return Ranking from Raw Intraday Bars

We present STRATA (Staggered-Timescale Residual Architecture), a 244,633-parameter sequence model that maps five trading days of raw five-minute bar and order-book data directly to a next-day cross-sectional return ranking, with no hand-crafted features. The raw-input setting has a structural obstacle: price series are non-stationary and differ across stocks by orders of magnitude, so a model easily latches onto price level rather than dynamics. STRATA addresses it with a stem of five branches--four learnable causal depthwise convolutions whose effective kernels are initialised to sum to zero, plus one cross-field linear contrast--followed by four selective state-space blocks whose decay biases are staggered across the stack and a four-path readout. Because a score that merely tilts toward common style factors scores well on raw rank correlations, every model's scores are residualised against eight price-volume style factors before any metric is computed. Trained on four years of data covering roughly one thousand mid-capitalisation Chinese A-shares and evaluated once on a held-out year, STRATA reaches a style-residualised rank information coefficient of 0.0728 (information ratio 1.128, signal long-short Sharpe 12.85), ahead of six parameter-matched sequence baselines on all four reported metrics; on rank IC the day-level paired gap against every baseline is significant at p < 0.001, and among the arms competitive on predictive power STRATA's scores are the least explained by the controls. The close-to-close target opens before the score exists: measured instead from the first executable price, the decile spread is indistinguishable from zero, while the ordering of the seven architectures is unchanged and STRATA's margin widens.

cs.CE↗

arXiv2026

Mitigating False Credit Propagation: Probabilistic Graphical Reward Aggregation for Rubric-Based Reinforcement Learning

Rubric-based rewards are increasingly used for open-ended language model post-training, but criterion-level scores are often aggregated as independent utilities. This flat scalarization ignores rubric-specified prerequisite and activation relations among criteria, allowing reward or penalty to be counted even when the condition that licenses it is absent. We call this structural reward-aggregation failure \textbf{False Credit Propagation} (FCP). To address this limitation, we propose \ourname (\textbf{G}raphical \textbf{E}vent \textbf{A}ggregation for \textbf{R}ubric rewards), a probabilistic graphical framework for dependency-aware rubric aggregation. \ourname models each criterion outcome as a latent Bernoulli event in a typed rubric graph, propagates soft suppression from unsupported parent events to their children, and aggregates the resulting event probabilities into a normalized expected signed utility. This yields a linear-time reward computation that can be plugged into standard rubric-based RL pipelines without changing the outer optimization algorithm. Experiments on HealthBench, WritingBench, and PLawBench with two policy backbones show that \ourname consistently improves over flat aggregation and deterministic gating, achieving relative gains of up to 15.5\% over flat aggregation. FCP diagnostics further show that \ourname reduces leakage by 96.5\% relative to flat aggregation while preserving more licensed downstream utility than deterministic gating. Our code is publicly available at https://github.com/LvCan926/GEAR.

cs.LG↗

arXiv2026

All-Mem: Agentic Lifelong Memory via Dynamic Topology Evolution

Lifelong interactive agents are expected to assist users over months or years, which requires continually writing long term memories while retrieving the right evidence for each new query under fixed context and latency budgets. Existing memory systems often degrade as histories grow, yielding redundant, outdated, or noisy retrieved contexts. We present \textbf{All-Mem}, an online/offline lifelong memory framework that maintains a topology structured memory bank via explicit, non destructive consolidation, avoiding the irreversible information loss typical of summarization based compression. In online operation, it anchors retrieval on a bounded visible surface to keep coarse search cost bounded. Periodically offline, an LLM diagnoser proposes confidence scored topology edits executed with gating using three operators: Split, Merge, and Update, while preserving immutable evidence for traceability. At query time, typed links enable hop bounded, budgeted expansion from active anchors to archived evidence when needed. Experiments on \textbf{LoCoMo} and \textbf{LongMemEval-s} show improved retrieval and QA over representative baselines. The code is available at https://github.com/LvCan926/All-Mem.

cs.IR↗

arXiv2026

HarnessForge: Joint Harness and Policy Evolution for Adaptive Agent Systems

LLM agents are increasingly expected to operate across heterogeneous task regimes that require distinct execution paradigms. This challenges fixed agent systems and motivates system-level meta-adaptation beyond isolated component updates. While existing works have adapted external harness or trained underlying reasoning policies, full-system adaptation remains insufficiently characterized. The adaptation space between structure and execution is rarely made explicit, and the compatibility between the external harness and the internal reasoner is not optimized jointly. We propose HarnessForge, a meta-adaptive framework for evolving LLM agent systems. HarnessForge formulates an agent system as a harness--policy pair, defining a stable adaptation space that separates harness-level execution structure from policy-level reasoning behavior. It then performs harness--policy co-evolution through fault-guided harness tailoring and harness-conditioned policy alignment. Experiments across five benchmarks from diverse domains show that HarnessForge consistently improves both Qwen3-4B and Qwen3-8B backbones, outperforming harness-only and policy-only baselines with gains of up to 12.0\% over the strongest baseline and achieving favorable rollout-efficiency tradeoffs, demonstrating that harness--policy co-evolution is effective, and that executable compatibility between the harness and reasoning policy is essential for agent-system adaptation. The code is available at https://github.com/mingju-c/HarnessForge.

cs.CL↗