arXiv subjects

Zekun Cai

Publications and source records attributed to Zekun Cai.

At least 19 recordsLinked to original sources

arXiv2026

Rethinking Optimization Methods for Hierarchical Feature Selection

Rare-event forecasting often depends on interpretable conjunctions: a single keyword, location, or temporal cue may be weak, while their joint occurrence can identify a meaningful precursor. Hierarchical-Task Numerical conjunctive feature Learning (HTNL) selects Numerical Conjunctive Features (NCFs) through a structured sparsity objective. However, the resulting optimization problem is challenging, making optimization design critical to its practical performance. We give a unified diagnosis of four HTNL optimization families: original majorization-minimization (MM), variational quadratic form (VQF), alternating direction method of multipliers (ADMM), and weighted group lasso (WGL). Our analysis distinguishes three key factors that jointly determine HTNL optimization behavior. First, VQF and MM are algebraically equivalent under matched variational updates and optimize a squared-penalty formulation. Second, ADMM optimizes the unsquared objective through consensus splitting; we replace its truncated FISTA proximal solver with a scalar bisection update (ADMM-Bisect) for more accurate treatment of near-threshold blocks. Third, WGL targets the same unsquared objective, while all methods face rapidly decaying active-set admission scores. Across two synthetic datasets and 11 civil-unrest and flu-surveillance tasks, VQF achieves the highest mean AUC with the most compact non-collapsing supports. ADMM-Bisect reduces measured runtime, while calibrated WGL avoids the empty-support failures of its original scale. Synthetic results further show how proximal accuracy, pruning, and search budget affect candidate recovery, especially for longer conjunctions. Code will be made publicly available upon publication.

cs.AI↗

arXiv2026

World Model Science: Self-Organized Criticality, Weak Chaos, and Metastable Belief Dynamics in Long-Horizon LLM Agents

Long-horizon LLM agents must maintain and update a world model over extended sequences of observations, actions, decisions and tool calls. We study the dynamics of long horizon agents from three perspectives: self-organized criticality, weak chaos, and metastable belief dynamics. Through extensive experiments on controlled puzzles, tool-use tasks, embodied environments, multi-hop retrieval, general-assistant reasoning, and Game of Life, we study how world-model errors develop over long interactions. We find that externally increased task pressure can sharply raise the probability of collapse, while naturally occurring pressure is only a modest predictor of later failure. Errors are temporally correlated rather than independent, and their propagation changes with dependency depth while remaining bounded in finite tasks. We also find that locally valid actions can continue after the agent's overall task state has become incorrect, that error structure depends on task topology and horizon, and that belief states in embodied tasks are shallow rather than strongly trapped in incorrect states. Larger horizons allow larger error cascades, but their scale remains constrained by the task and the model's capability. Overall, we observe power-law behavior, bounded divergence consistent with weak chaos, and metastable belief dynamics across long-horizon agent interactions. We call this perspective world model science: the empirical study of self organized criticality in long-horizon agent world models.

cs.AI↗

arXiv2026

Heavy-Tailed Memory Traces in Long-Horizon Language Agents

Long-horizon language agents are becoming increasingly important for tasks that require persistent interaction, planning, and reasoning over extended trajectories. External memory plays a central role in these agents by preserving past states, transitions, and experiences beyond the limited context window. Existing memory systems, however, are typically evaluated mainly by task success and token cost, with much less attention to how memory usage is distributed across stored information. We show that repeated retrieval can produce heavy-tailed memory usage, where a small set of states is accessed frequently while many others are rarely retrieved and more prone to prediction errors. We study this phenomenon through a systematic tail audit across retrieval policies and graph structures. We find that memory concentration is reproducible but strongly policy-dependent, ranging from log-normal-compatible patterns under random walks to truncated-power-law-compatible heavy tails under semantic LLM policies. Motivated by these findings, we propose Core-Tail Memory Controller (CTMC), a rank-based memory controller that allocates prompt budget with a single exponent $τ$ while preserving summarized information from the tail. Across Synthetic Graph World and ALFWorld, CTMC substantially reduces memory usage while preserving performance, and on LongMemEval it reduces prompt tokens by 24.48% while maintaining aggregate accuracy. These results show that heavy-tailed memory usage is both a measurable property of long-horizon agent memory and a practical signal for improving memory efficiency.

cs.AI↗

arXiv2026

When Do Causal World Models Help Modular LLM Agents

LLM agents increasingly operate in modular systems, where planning requires coordinating multiple functional components whose actions can causally change downstream states and behaviors. This makes intervention-aware prediction essential: an agent must anticipate not only what is likely to happen next, but what will happen as a consequence of its own actions. We find that purely observational world models can incur irreducible intervention-time error in such systems, even when they accurately model observational trajectories. To address this limitation, we propose CausalCompose, a causal world-model framework that equips modular LLM agents with causal structure for intervention-aware prediction and planning. We theoretically show that causal composition can overcome this intervention gap with sufficient evidence and controlled local error. Experiments across diverse modular agent environments validate these results: incorporating causal structure consistently improves intervention-sensitive reasoning and planning, with particularly clear gains in structured tool-use systems where causal dependencies directly govern action execution. Our results show when causal world models help modular LLM agents: when causal dependencies are identifiable and actionable for planning.

cs.AI↗

arXiv2026

Agentic vs. Parametric World Models: Hybrid Planning for Reliable Language Agents

Language agents, such as ReAct, Reflexion, and RAP, implicitly predict how the world will change as they plan and act, using imagined state transitions to guide their decisions. These predictions are encoded in the agents' backbone LLMs as internal beliefs about the environment, effectively forming an implicit world model for multi-step planning. However, hallucinated state information can become embedded in this world model, leading to persistent inaccuracies that influence subsequent decisions. Traditional parametric world models, such as Dreamer and MuZero, learn environment dynamics through explicit transition models, but their limited model capacity makes them difficult to apply directly to modern language-agent planning. We study this tradeoff in graph-structured planning environments and seek to combine the strengths of both paradigms. We propose Hybrid World-Model Planning (HYBRID-WM), which preserves the language agent's semantic planning capabilities while leveraging a lightweight parametric world model for more accurate transition predictions. Specifically, the parametric model constructs a compact planning skeleton to guide action generation while maintaining consistency between the agent's imagined transitions and the model's predictions. Experiments across four graph-structured planning benchmarks demonstrate that HYBRID-WM consistently improves task success and reduces hallucinated-state propagation with modest computational overhead. These findings highlight the effectiveness of combining semantic planning with lightweight parametric grounding for reliable long-horizon language-agent planning.

cs.AI↗

arXiv2026

Ask the World Before Acting: Environment Probing for Calibrating Agent World Models

Long-horizon language agents enable autonomous decision-making in complex environments by planning and executing sequences of interdependent actions. To act reliably, these agents must maintain an accurate world model that tracks the environment's evolving state and supports subsequent decisions. However, as interactions accumulate, stored beliefs can become outdated or incorrect, causing agents to plan and act on assumptions that no longer reflect the environment. Such errors are difficult to resolve through additional reasoning or self-reflection alone, because the information needed for correction must be obtained from the environment. Actively probing the environment therefore offers a direct way to verify and correct these beliefs. To this end, we introduce ENVPROBE, a framework that enables agents to actively explore their environments and correct inaccurate beliefs during long-horizon execution. Specifically, ENVPROBE prioritizes belief fields for verification based on task relevance and potential errors. It then selectively probes the environment to retrieve the actual value of a chosen field and incorporates the observation into the agent's world model. With the updated beliefs, the agent revises its plan and continues execution based on the newly acquired environmental information. Extensive experiments across three controlled environments demonstrate that ENVPROBE substantially improves terminal world-state accuracy compared with agents operating without environment probing, leading to more reliable long-horizon decision-making. These results demonstrate the importance of actively verifying and correcting beliefs through environmental interactions for maintaining accurate world models in long-horizon language agents.

cs.AI↗

arXiv2026

Repair the Core, Not the Symptom: Planning-Graph Repair for Long-Horizon Agents

Long-horizon language agents make interdependent decisions across multiple rounds, forming growing planning graphs in which later decisions depend on earlier states, observations, and tool outputs. Errors in these graphs can therefore propagate through subsequent decisions and amplify over long rollouts. Correcting the most visible failure may therefore fix a symptom while leaving the core that drives downstream errors unchanged. Replaying the entire planning graph provides the opposite extreme, but becomes increasingly expensive as the graph grows and forces an LLM to reason over substantial irrelevant context. We study planning-graph repair: given a failed rollout, the goal is to identify and correct the compact core responsible for continued error propagation while preserving the remainder of the graph. We first evaluate a range of practical engineering repair strategies, including single-error and TopK repair, temporal windows, neighborhood expansion, cascade-aware repair, and full-graph LLM correction. We then propose ReCore (Repair the Core), which models the coupled propagation of node and edge errors to identify the core of a failed planning graph. Rather than repairing the largest observed errors or expanding a fixed local neighborhood, ReCore selects a connected core whose correction maximally reduces residual error amplification, and exposes only this core to the LLM for diagnosis and repair. We also theoretically establish a bound on long-horizon planning regret. Extensive experiments on synthetic Agent Calling-Tree graphs and the Who\&When attribution benchmark show that ReCore improves long-horizon stabilization and root-cause localization by targeting high-amplification repair regions, matching full-context recovery with up to $7.4\times$ fewer tokens for every successful repair.

cs.AI↗

arXiv2026

Parameter-Efficient Temporal Domain Generalization

Temporal domain generalization (TDG) learns from a sequence of observed domains to predict models for unseen future domains. Existing parameter-dynamics methods become prohibitively expensive at modern backbone scales because their temporal encoders and generators operate on the entire adapted parameter state. Replacing full-model updates with LoRA does not remove this failure mode: the dynamics model must still encode and reconstruct every adapter parameter, making the auxiliary model itself prohibitively large for modern backbones. We propose Low-rank Time-varying Core (LoTiC), a parameter-efficient TDG method that decomposes temporal updates into shared low-rank bases and a compact core. The bases define a stable update subspace, while only the core evolves. LoTiC therefore learns temporal dynamics in an r*r space instead of tracking the full adapter. We justify this factorization through the geometry of parameter increments and support continuous, recurrent, and time-conditioned core models. Experiments span six controlled and real-world benchmarks, three data modalities, classification and generation, and backbones from 66M to 8B parameters. LoTiC ranks first in 14 of 15 settings, including near-perfect future-domain accuracy with an 8B backbone on Rotating 2-Moons. It adds only 0.13--0.51% trainable parameters beyond LoRA and trains 1.8--7.1$\times$ faster than parameter-dynamics baselines. These results establish that temporal parameter dynamics can remain both learnable and efficient as pretrained backbones scale.

cs.LG↗

arXiv2026

Understanding Rollout Error in Graph World Models

World models are increasingly used for planning by predicting future states under candidate actions. Many planning environments, however, are naturally graph-structured, with agents, tools, skills, routes, and dependencies connected through fixed or evolving relations. Understanding how prediction errors accumulate in such graph-structured rollouts is therefore critical for reliable long-horizon planning.In this work, we systematically study how rollout errors propagate in Graph World Models (GWMs). We analyze both fixed- and dynamic-edge regimes and derive topology-aware error bounds that characterize the roles of graph structure, learned dynamics, and evolving edges. We then validate these theoretical findings across synthetic graph topologies and heterogeneous agent-graph testbeds, showing that rollout error and planning regret increase with horizon and that dynamic-edge training is important when graph structure evolves. Together, these results characterize when GWMs remain reliable under long-horizon planning and when graph structure amplifies rollout errors. Motivated by these findings, we propose Error-Aware GWM, which combines spectral regularization, rollout consistency, and critical-node weighting, and improves long-horizon stability without sacrificing one-step accuracy. Experiments show that Error-Aware GWM consistently improves long-horizon rollout stability without sacrificing prediction accuracy.

cs.AI↗

arXiv2026

When AI Reviews Its Own Code: Recursive Self-Training Collapse in Code LLMs

Recursive self-training can degrade neural generative models when generated data is reused without fresh human data or external quality control. We study this risk in code LLMs, where AI-generated code can enter real repositories, later become training data, and create a repository-scale self-training loop. While software development traditionally interrupts this loop through pull-request review, tests, compilation, and human approval, AI coding tools now produce code faster than humans can review it, and code review itself is increasingly automated by AI systems. We therefore compare three recursive fine-tuning regimes: no review, human-style exogenous-gate review using model-independent filters such as compilation and static quality checks, and AI-self-gate review using the code LLM's own signals such as perplexity and binary self-scoring. Across multiple code LLMs and benchmarks, every tested recursive regime falls below its own starting point; exogenous filters preserve executable validity and sometimes reduce decline, while AI-self-gate filters can look strong early but later lose their filtering effect. In the clearest case, the binary self-gate enters a rubber-stamp regime where acceptance scores rise while benchmark correctness falls. We explain this behavior by formulating review as gated distributional reweighting, proving that AI self-gating degenerates to ungated self-training under a self-confirming acceptance condition, and giving a spectral analysis of representation-level covariance concentration under recursive retraining. These results suggest that stable recursive code LLM training requires exogenous verification rather than model-coupled self-review.

cs.SE↗

arXiv2026

World-Model Collapse as a Phase Transition

Water looks unchanged as it warms, then at a critical point it boils. We ask whether long-horizon language agents show an analogous transition in their implicit world models. In some parameter settings, changing state load by a small amount, or adding a single step of horizon, leaves behavior nearly unchanged; near a critical boundary, the same small change causes a sudden world collapse. We study this effect in a deterministic task family with exact per-step gold state. A large grid search over state cardinality, dependency density, horizon, branching, observation mode, and mutation rate reveals a phase diagram: a solved plateau, a narrow transition band, and a collapse floor. Per-step traces show the mechanism: world-state fidelity fails before action validity, so the agent is not merely choosing a bad action; it is acting from a corrupted world. Stronger models translate the critical boundary but do not remove the qualitative transition. These results make world-model collapse a measurable bottleneck for long-horizon agents.

cs.AI↗

arXiv2026

AlgoBench: Benchmarking Algorithmic Adaptation in Code Generation

High pass rates on established programming benchmarks such as HumanEval and LiveCodeBench do not always show whether a model can reason about algorithms. Many fixed benchmarks eventually become part of the public training ecosystem through released problem statements, editorials, and generated solutions, allowing later models to improve partly by exposure rather than by stronger algorithmic ability. We introduce ALGOBENCH, a framework that automatically builds novel algorithmic problems from known competitive-programming problems through structured constraint-shifting transformations. Each accepted ALGOBENCH variant is traceable to a source problem, but must make the original reference algorithm fail. Beyond pass@$k$, we introduce complexity-aware metrics -- including OPTT, OPTS, TRAPRATE, GAPT, and CONSENS -- to test whether a solution is not only functionally correct but also asymptotically suitable for the generated problem. Experiments across multiple LLMs and prompting strategies show that performance drops sharply on ALGOBENCH variants, retrieval can increase reuse of the old algorithm, and many correct-looking solutions fail to meet the required complexity. Error analysis shows that failures are mainly algorithmic rather than implementation-level, suggesting that ALGOBENCH evaluates adaptation beyond functional correctness.

cs.SE↗

arXiv2026

AlgoSkill: Learning to Design Algorithms by Scheduling Human-Like Skills

Designing an algorithm from a natural-language problem statement requires identifying the problem structure, reading constraints, choosing a suitable paradigm, checking correctness, and refining complexity. Existing large language model (LLM) methods often rely on direct generation or generic self-refinement, leaving these steps implicit. We propose AlgoSkill, which models algorithm design as sequential decision-making over a typed library of algorithmic skills, including abstraction, constraint analysis, state design, data-structure selection, proof checking, counterexample construction, and complexity refinement. A learned scheduler proposes skills from the current design state, while a Monte Carlo Tree Search (MCTS) controller explores skill sequences using verification feedback from compilation, testing, stress testing, and complexity analysis. Experiments on competitive programming and combinatorial optimization benchmarks show that AlgoSkill improves over direct LLM generation, chain-of-thought prompting, self-refinement, and MCTS without typed skills. Ablations show that typed skills, verification-based repair, and search-based scheduling each contribute to performance. These results support treating automatic algorithm design as verification-guided skill scheduling rather than one-shot code generation.

cs.AI↗

arXiv2025

Continuous Domain Generalization

Real-world data distributions often shift continuously across multiple latent factors such as time, geography, and socioeconomic contexts. However, existing domain generalization approaches typically treat domains as discrete or as evolving along a single axis (e.g., time). This oversimplification fails to capture the complex, multidimensional nature of real-world variation. This paper introduces the task of Continuous Domain Generalization (CDG), which aims to generalize predictive models to unseen domains defined by arbitrary combinations of continuous variations. We present a principled framework grounded in geometric and algebraic theories, showing that optimal model parameters across domains lie on a low-dimensional manifold. To model this structure, we propose a Neural Lie Transport Operator (NeuralLio), which enables structure-preserving parameter transitions by enforcing geometric continuity and algebraic consistency. To handle noisy or incomplete domain variation descriptors, we introduce a gating mechanism to suppress irrelevant dimensions and a local chart-based strategy for robust generalization. Extensive experiments on synthetic and real-world datasets, including remote sensing, scientific documents, and traffic forecasting, demonstrate that our method significantly outperforms existing baselines in both generalization accuracy and robustness.

stat.ML↗

arXiv2024

Continuous Temporal Domain Generalization

Temporal Domain Generalization (TDG) addresses the challenge of training predictive models under temporally varying data distributions. Traditional TDG approaches typically focus on domain data collected at fixed, discrete time intervals, which limits their capability to capture the inherent dynamics within continuous-evolving and irregularly-observed temporal domains. To overcome this, this work formalizes the concept of Continuous Temporal Domain Generalization (CTDG), where domain data are derived from continuous times and are collected at arbitrary times. CTDG tackles critical challenges including: 1) Characterizing the continuous dynamics of both data and models, 2) Learning complex high-dimensional nonlinear dynamics, and 3) Optimizing and controlling the generalization across continuous temporal domains. To address them, we propose a Koopman operator-driven continuous temporal domain generalization (Koodos) framework. We formulate the problem within a continuous dynamic system and leverage the Koopman theory to learn the underlying dynamics; the framework is further enhanced with a comprehensive optimization strategy equipped with analysis and control driven by prior knowledge of the dynamics patterns. Extensive experiments demonstrate the effectiveness and efficiency of our approach. The code can be found at: https://github.com/Zekun-Cai/Koodos.

cs.LG↗

arXiv2023

MemDA: Forecasting Urban Time Series with Memory-based Drift Adaptation

Urban time series data forecasting featuring significant contributions to sustainable development is widely studied as an essential task of the smart city. However, with the dramatic and rapid changes in the world environment, the assumption that data obey Independent Identically Distribution is undermined by the subsequent changes in data distribution, known as concept drift, leading to weak replicability and transferability of the model over unseen data. To address the issue, previous approaches typically retrain the model, forcing it to fit the most recent observed data. However, retraining is problematic in that it leads to model lag, consumption of resources, and model re-invalidation, causing the drift problem to be not well solved in realistic scenarios. In this study, we propose a new urban time series prediction model for the concept drift problem, which encodes the drift by considering the periodicity in the data and makes on-the-fly adjustments to the model based on the drift using a meta-dynamic network. Experiments on real-world datasets show that our design significantly outperforms state-of-the-art methods and can be well generalized to existing prediction backbones by reducing their sensitivity to distribution changes.

cs.LG↗

arXiv2023

Real-World Video for Zoom Enhancement based on Spatio-Temporal Coupling

In recent years, single-frame image super-resolution (SR) has become more realistic by considering the zooming effect and using real-world short- and long-focus image pairs. In this paper, we further investigate the feasibility of applying realistic multi-frame clips to enhance zoom quality via spatio-temporal information coupling. Specifically, we first built a real-world video benchmark, VideoRAW, by a synchronized co-axis optical system. The dataset contains paired short-focus raw and long-focus sRGB videos of different dynamic scenes. Based on VideoRAW, we then presented a Spatio-Temporal Coupling Loss, termed as STCL. The proposed STCL is intended for better utilization of information from paired and adjacent frames to align and fuse features both temporally and spatially at the feature level. The outperformed experimental results obtained in different zoom scenarios demonstrate the superiority of integrating real-world video dataset and STCL into existing SR models for zoom quality enhancement, and reveal that the proposed method can serve as an advanced and viable tool for video zoom.

cs.CV↗

arXiv2022

ST-ExpertNet: A Deep Expert Framework for Traffic Prediction

Recently, forecasting the crowd flows has become an important research topic, and plentiful technologies have achieved good performances. As we all know, the flow at a citywide level is in a mixed state with several basic patterns (e.g., commuting, working, and commercial) caused by the city area functional distributions (e.g., developed commercial areas, educational areas and parks). However, existing technologies have been criticized for their lack of considering the differences in the flow patterns among regions since they want to build only one comprehensive model to learn the mixed flow tensors. Recognizing this limitation, we present a new perspective on flow prediction and propose an explainable framework named ST-ExpertNet, which can adopt every spatial-temporal model and train a set of functional experts devoted to specific flow patterns. Technically, we train a bunch of experts based on the Mixture of Experts (MoE), which guides each expert to specialize in different kinds of flow patterns in sample spaces by using the gating network. We define several criteria, including comprehensiveness, sparsity, and preciseness, to construct the experts for better interpretability and performances. We conduct experiments on a wide range of real-world taxi and bike datasets in Beijing and NYC. The visualizations of the expert's intermediate results demonstrate that our ST-ExpertNet successfully disentangles the city's mixed flow tensors along with the city layout, e.g., the urban ring road structure. Different network architectures, such as ST-ResNet, ConvLSTM, and CNN, have been adopted into our ST-ExpertNet framework for experiments and the results demonstrates the superiority of our framework in both interpretability and performances.

cs.LG↗