arXiv:2606.06085v1 Announce Type: new
Abstract: We study the revenue of approximate correlated equilibrium in discrete first price auctions - the set of allowable bids is $\mathcal{B} = \{0, 1/k, \dots, 1 - 1/k, 1\}$ for some $k \in \mathbb{N}$. We show that the revenue of any $\epsilon$-approximate correlated equilibrium is at least $v_2 - \Theta(1/k)- \Theta(\epsilon k^2)$, where $v_2 \geq 0$ is the second-highest valuation. Our results establish the first polynomial convergence rates on the revenue generated by no-swap regret bidders in first-price auctions.
For instance, if bidders admit the optimal swap regret of $\mathcal{O}(\sqrt{k T})$, then the time-averaged revenue is at least $v_2 - \Theta(1/k) - \Theta(\epsilon)$ after $\mathcal{O}(k^5/\epsilon^2)$ rounds.
Science Journals
arXiv:2606.06090v1 Announce Type: new
Abstract: LLM-based agents increasingly tackle long-horizon tasks with interdependent decisions, where each action reshapes future constraints and intermediate errors can cascade. Existing RAG and agent memory systems organize histories by semantic similarity, retrieving content-relevant entries at decision time. We argue that this design mismatches execution-state dependencies: it fragments decision trajectories and mixes valid and erroneous traces, hindering coherent state reconstruction and error isolation. We propose MAGE (Memory as Agent-Guided Exploration), an active execution-state manager that stores interactions in a hierarchical state tree. The agent derives its state from the active root-to-current path, combining subgoal summaries, recent traces, and hints from prior branches. Four coupled operations maintain the tree: Grow records new traces, Compress summarizes completed subgoals, Maintain validates summaries, and Revise restores a target boundary and resumes on a new branch. This design bounds context growth while preserving state integrity and isolating flawed segments from the active path. Experiments on MemoryArena show that MAGE improves the average task success rate by 7.8--20.4 pp over baselines, while reducing token consumption by 55.1%.
arXiv:2606.06191v1 Announce Type: new
Abstract: Deep reinforcement learning (DRL) applied to thermal convection control consistently produces \textit{degenerate actuation}: wall-temperature policies whose outputs are saturated, pseudo-random, or spatially incoherent. Two compounding deficiencies are responsible: multilayer-perceptron policies that discard spatial flow structure, and memoryless policies that cannot distinguish self-induced flow changes from background evolution. Together they prevent the discovery of physically meaningful control laws even when cell coalescence (the merging of convection rolls into fewer, larger structures), which would reduce $\mathrm{Nu}$, is accessible to boundary actuation. The present framework addresses both causes through four targeted design choices: convolutional policy networks, Gated Recurrent Unit (GRU) memory, off-policy training (TD3/MADDPG), and action-smoothness constraints. A systematic $2\times2$ factorial design isolates the contribution of each component. On Rayleigh--B\'{e}nard convection at $\mathrm{Ra}=10{,}000$, all four configurations achieve cell coalescence and reduce $\mathrm{Nu}$ to as low as $1.83$ ($26\%$ below the uncontrolled baseline) in 350 episodes, without the full-field data augmentation required by prior work. Crucially, coalescence is achieved even by the single-agent configuration, demonstrating that the multi-agent formulation is not a prerequisite once the policy architecture is sufficiently expressive. Applied to double-diffusive convection in the salt-finger regime, the framework spontaneously discovers a travelling-wave actuation whose phase speed adapts to the evolving mixing state of the flow, enhancing heat transfer by $19.1\%$ and reducing salinity variance by $21.0\%$.
arXiv:2606.06091v1 Announce Type: new
Abstract: Hyperproperties express, e.g., information-flow properties of systems, which involves the simultaneous reasoning about multiple execution traces of a system. Consequently, HyperLTL, the most important specification logic for hyperproperties, extends LTL with quantification over traces. However, HyperLTL can only express synchronous hyperproperties.
Recently, several logics for asynchronous hyperproperties have been proposed. Here, we focus on AHLTL, asynchronous HyperLTL, which extends HyperLTL with quantification over trajectories that control the relative speed at which time progresses on the quantified traces. Model-checking AHLTL is known to be undecidable while satisfiability is known to be $\Sigma_1^1$-hard, but the precise complexity of both problems is open.
Here, we close these gaps and show that model-checking is equivalent to truth in second-order arithmetic while satisfiability is $\Sigma_1^1$-complete if the trajectory is existentially quantified and $\Sigma_1^1$-hard and in $\Sigma_2^1$ if the trajectory is universally quantified.
arXiv:2606.06093v1 Announce Type: new
Abstract: Laplace transforms and their numerical inverses arise throughout applied mathematics, physics, finance, and probability theory. Numerical inversion, however, quickly becomes intractable in high dimensions because the number of quadrature evaluations grows exponentially with dimension. We develop a tensor train (TT) formulation of the multidimensional inverse Laplace transform. The method constructs a TT approximation of the transformed function on the complex quadrature grid and then performs the inversion through a sequence of tensor contractions. Under suitable low-rank assumptions, this reduces the computational cost from exponential to polynomial in the dimension, provided that the relevant bond dimensions remain bounded. The method has only a small number of tunable parameters and admits error estimations. We demonstrate its performance in numerical experiments, benchmarked against Monte Carlo estimates and exact references, for multivariate normal-inverse Gaussian, Wishart, and correlated Gamma-type distributions.
arXiv:2606.06096v1 Announce Type: new
Abstract: Policy-gradient methods usually optimize expected return, but many real world applications care about distributional properties of returns: tail risk, outlier robustness, or best-of-K discovery. We introduce OrderGrad, a family of likelihood-ratio and reparameterization gradient estimators for order-statistic objectives. OrderGrad optimizes finite-sample L-statistics, i.e., weighted averages of sorted rewards or costs, recovering objectives such as VaR, CVaR, trimmed means, medians, and top-m/best-of-K criteria by changing only the rank weights. For any fixed sample size and rank-weight vector, OrderGrad provides an unbiased gradient estimator for the corresponding order-statistic objective. The method is implemented as a simple reward transformation that can then be used in an otherwise standard policy-gradient or reparameterized update. We study the resulting estimator's variance behavior and evaluate it on tasks where mean optimization is mismatched to the deployment objective, including LLM math post-training and other tasks. OrderGrad provides a unified, plug-and-play route to risk-averse, robust, and exploratory learning.
Code: https://github.com/paavo5/ordergrad
arXiv:2606.06098v1 Announce Type: new
Abstract: Foundational Large Language Models (LLMs) demonstrate proficiency on a wide range of general tasks, and achieve remarkable results on various specialized tasks via domain-expert LLMs. With the ever-growing list of available LLMs, inference routers are being proposed to select the most appropriate LLM for each prompt. However, existing routing methods either optimize cost across weak-to-strong generalist LLMs or require substantial training to support domain-expertise routing. In this paper, we propose IR3DE, a Ridge Regression-based Router for Domain Experts that provides cheap and fast routing decisions for each prompt. We evaluate IR3DE in two Causal Language Modeling (CLM) settings where the tasks are next-token prediction for all domains, and one reasoning setting where each domain has its own distinct reasoning task. Despite being a linear router, IR3DE achieves performance comparable to the other baselines in both CLM settings, and surpassing them in the reasoning setting, with a normalized performance of 98.4%. Moreover, IR3DE enables the addition or removal of new domain experts without requiring the router to be retrained from scratch, allowing a dynamic set of LLMs to be served with minimal disruption to the router itself. Our code is available at: github.com/gensyn-ai/IR3DE.
arXiv:2606.06099v1 Announce Type: new
Abstract: Whether Large Language Models (LLMs) exhibit covert psychological manipulation in complex human-AI interactions has garnered increasing safety concerns. However, existing AI safety benchmarks remain largely restricted to explicit rule compliance and static prompts, failing to capture the dynamic and covert nature of manipulative strategies in multi-turn dialogues. We introduce CogManip, a comprehensive benchmark that evaluates 15 manipulation strategy risks across 1,000 multi-turn interaction scenarios, validated by human experts. A systematic evaluation of 13 representative models, including frontier models like GPT-5.4 and DeepSeek-V3.2, reveals significant risk heterogeneities and illuminates the targeted direction for future defense. Further analysis of objective function perturbation reveals that DeepSeek-V3.2's manipulation tactics are highly sensitive to both negative and benign system prompts, demonstrating the critical necessity of prompt-based defense engineering and implicit goal auditing. CogManip offers a robust instrument and perspective for auditing the implicit psychological influence and dynamic strategy selection of modern LLMs.
arXiv:2606.06102v1 Announce Type: new
Abstract: Ultra-short-term solar irradiance prediction is critical for photovoltaic system dispatch and power grid stability. Existing approaches suffer from three key shortcomings: single time-series models cannot capture the spatial dynamics of clouds under complex conditions, standard convolutions inadequately represent multi-scale cloud features, and fixed low-frequency compensation strategies fail to adapt to different prediction steps. To address these issues, this proposes a multi-source data fusion model for ultra-short-term irradiance prediction. The model first employs InceptionNeXt to extract multi-scale, multi-directional spatial features from ground-based cloud images. A step-adaptive low-frequency compensation unit is then introduced to dynamically modulate global low-frequency information based on the prediction step. Eventually, the enhanced image features are combined with meteorological time-series features, and a TempAttnLSTM network captures global temporal dependencies for multi-step prediction. Experiments on the public NREL dataset and practical photovoltaic stations in Shandong illustrate the effectiveness of the proposed method compared with several state-of-the-art approaches.
arXiv:2606.06123v1 Announce Type: new
Abstract: When learning to walk, infants seem to address a coarse version of the problem first - stay upright, reach the caregiver - and refine it only when further practice at that resolution stops paying off. Reinforcement learning offers multiple techniques for building simple versions of complex tasks, but lacks general principles for how to dynamically adjust the granularity of these abstractions during learning. This paper proposes one such principle: refine the abstraction as soon as the learning error within it becomes comparable to the error induced by the abstraction itself. Here, we investigate one way of formalising this principle via a performance certificate that decomposes value error into two terms: a learning error bound captured by a Bellman residual, and an abstraction error bound given by a bisimulation metric. The resulting switching strategy is implemented by soft state-action abstractions built from rate-distortion principles, whose resolution along state and action axes can be continuously adjusted. We validate this construction in a range of tabular settings, showing that near-optimal performance can be achieved under substantial lossy compression of state and action information.
arXiv:2606.06139v1 Announce Type: new
Abstract: We present MotionDisco, a framework that discovers contact-rich, long-horizon humanoid loco-manipulation motions from scratch, without relying on teleoperation or motion retargeting from human demonstrations. This is challenging because the space of possible contact interactions grows combinatorially with the task horizon and the number of objects in the scene. MotionDisco enables rapid discovery of novel motions by coupling a large language model (LLM) guided evolutionary search over sequences of interactions with an efficient sequential kinodynamic trajectory optimizer and pruning strategy, enabling the rapid discovery of novel skills. Through extensive ablation studies, we show that our LLM-guided search discovers successful whole-body trajectories across several challenging long-horizon tasks. Finally, by training reinforcement learning tracking policies on the discovered trajectories, we transfer the motions to a real humanoid robot. This is the first work to discover and deploy long-horizon humanoid loco-manipulation skills entirely through automated evolutionary search. Supplementary videos of the experiments are available at: https://youtu.be/DHiVz34QYlw.
arXiv:2606.06189v1 Announce Type: new
Abstract: Demand-responsive transit (DRT) is a flexible alternative to traditional, fixed-route mass-transit networks. Although DRT can function well in low-density communities, high operating costs and low reliability are common issues. We propose that these issues can be mitigated by moving from a centralized, manually-scheduled scheme to a distributed system capable of dynamically routing multiple vehicles using a slime-mold-inspired routing algorithm to maximize network effectiveness. We additionally introduce the method of dynamic transfers to further optimize transit network efficiency. All passenger allocation and dynamic transfers are handled via a continual cooperative bidding process by the buses. In this paper, we present simulated results for a swarm-driven transit network in suburban, urban, and semi-rural scenarios, using map networks pulled from OpenStreetMap. We show that our approach increases passenger delivery rates relative to a fixed-network approach by 28%, 49%, and 101%, respectively, and results in over 75% reduction in walking time in all cases.
arXiv:2606.06194v1 Announce Type: new
Abstract: Egocentric human video offers a scalable alternative to robot data for pretraining, yet models pretrained on such video consistently underperform those pretrained on robot data. We attribute this gap to a missing signal, the active perception behavior in egocentric videos, where humans continuously reposition their viewpoint during manipulation, inducing camera motion that standard pipelines treat as noise. To address this, we present ActiveMimic, a pretraining framework that recovers synchronized camera and wrist trajectories from a single body-worn RGB camera, models camera motion as a viewpoint action, and jointly learns active perception and manipulation from in-the-wild egocentric human video before adapting to a target robot. Empirically, real-world experiments across tasks with diverse active perception demands show that ActiveMimic consistently surpasses baselines pretrained on human video and matches state-of-the-art models pretrained on robot data. Further analysis provides evidence that active perception capability originates from egocentric human video pretraining rather than robot-specific fine-tuning, confirming active perception as the key to unlocking egocentric human video for robot pretraining.
arXiv:2606.05596v1 Announce Type: new
Abstract: In this paper, we study multi-objective submodular maximization (MOSM) subject to a cardinality constraint under differential privacy (DP). Specifically, we aim to select a set of at most $k \in \mathbb{Z}_{+}$ elements to maximize the minimum of $d > 1$ monotone submodular functions while satisfying $\varepsilon$-DP. Although extensive studies have been conducted on both differentially private single-objective submodular maximization on sensitive data and non-private MOSM, to the best of our knowledge, there has not yet been any prior work on MOSM with DP. We propose two novel algorithms: the first extends the classic greedy algorithm and the second employs a truncation technique, both of which are integrated with DP mechanisms for privacy protection and achieve approximation guarantees for MOSM. Finally, we conduct numerical experiments on two submodular maximization applications, namely maximum coverage and facility location, in multi-objective settings to validate the efficacy and efficiency of our proposed algorithms.
arXiv:2606.06207v1 Announce Type: new
Abstract: Veterinary pharmacovigilance systems are essential for monitoring adverse drug events (ADEs), yet existing approaches often fail to capture region-specific toxicity patterns shaped by local biological and regulatory contexts. In Japan, these challenges are amplified by species-specific metabolic differences and reporting practices defined by the Ministry of Agriculture, Forestry, and Fisheries (MAFF). Most prior work relies on prediction-oriented models, limiting mechanistic interpretability. This study proposes a regulatory-integrated unsupervised framework for pattern discovery using the National Veterinary Assay Laboratory (NVAL) database. ADEs are encoded into organ system-aligned representations and adjusted for species-specific reporting biases, enabling cross-species comparison. Similarity-based clustering and dimensionality reduction are applied to identify latent toxicity structures. Analysis of 4,120 high-confidence ADE reports (9,080 drug-ADE combinations) identified three significant species clusters (p < 0.01), including hepatic-dominant patterns in companion animals (0.42 $\pm$ 0.06), renal toxicity in ruminants (0.39 $\pm$ 0.07), and dermatological sensitivity in sheep (0.35 $\pm$ 0.07). Drug-level clustering achieved 83% alignment with pharmacological classes, while cosine similarity outperformed alternative metrics (silhouette score: 0.48; cluster precision: 87%). Regulatory validation showed strong agreement with established classifications. These findings demonstrate that regulation-aligned unsupervised analysis can uncover biologically meaningful, region-specific toxicity patterns, providing an interpretable and scalable framework for veterinary drug safety assessment.
arXiv:2606.06195v1 Announce Type: new
Abstract: Wave propagation in a medium differs depending on whether this medium is at rest or moving with respect to an observer. Motion can notably lead to modifications of the wave trajectory, of its polarization, or of its transverse structure. Although these effects are well documented in isotropic dielectrics, they remain largely unexplored and unaccounted for in plasmas, despite the fact that simple models suggest they could in fact be large under certain conditions, as well as recent experimental observations. Here we first review existing models for motion effects on plasma waves, then identify a number of basic challenges that lie in the way of using these models to quantify motion effects in realistic configurations, and finally discuss possible workarounds.
arXiv:2606.06201v1 Announce Type: new
Abstract: Pharmaceutical supply chains (PSCs) struggle with inventory management (IM) due to unpredictable demand patterns and variable lead times associated with restocking. This complexity is further compounded by the finite shelf lives of pharmaceutical products, which necessitate a delicate balance between adequate stock and minimal waste. These intertwined factors create a complex optimization problem that requires sophisticated inventory strategies to ensure both product availability and PSC efficiency. This study aims to develop an optimal inventory replenishment policy for pharmaceutical products that can handle the stochasticity arising from uncertain demand and variable PSC conditions. The objective is to maximize the profitability of the PSC while maintaining a high patient service level. We formulate the problem as a Markov decision process and propose a deep reinforcement learning (DRL) approach, specifically, a hybrid asynchronous advantage actor critic distributed proximal policy optimization (A3C DPPO)algorithm. The A3C DPPO algorithm is tailored to handle the continuous action space inherent in IM. The numerical results demonstrate that the proposed algorithm adaptively updates the inventory replenishment strategy under dynamic scenarios, resulting in lower inventory costs compared to various benchmarks. We also conduct numerical validation using real-world pharmaceutical inventory data to confirm the practical feasibility of the proposed algorithm.
arXiv:2606.06205v1 Announce Type: new
Abstract: Continuous-time event data, in which entities emit instantaneous events over time, arises naturally across many domains such as neuroscience, seismology, and social networks. Non-negative matrix factorization (NMF) is a natural tool to uncover interpretable structure in such data, but it has so far only been applied after binning or smoothing the entity-level counting measures. This preprocessing step comes with the risk of erasing entity-level heterogeneities and fine-grained temporal features. In this paper, we introduce EventNMF, a continuous-time non-negative factorization model that operates directly on event times: each entity's events are modeled as a Poisson process whose intensity factorizes through a non-negative B-spline basis, and a simple estimation procedure recovers interpretable temporal templates shared across entities. The resulting method is mathematically principled, easy to implement, and computationally efficient. We further show that standard binned-count approaches arise as the special case of degree-zero splines, explore bias-variance tradeoffs and compare against existing methods on a synthetic latent factor model, and demonstrate the effectiveness of EventNMF on several real-world applications.
arXiv:2606.06215v1 Announce Type: new
Abstract: Fluid-structure interactions at low Reynolds number can lead to a much richer phenomenology than previously expected. Here, we study the dynamics of a freely suspended, thin elastic disk in a shear flow, where the plane of the disk is initially parallel to the flow plane. Using a combination of experiments and simulations, we demonstrate that beyond a critical flow strength the disk deforms, performing flapping dynamics, in which the disk curves up and down periodically relative to the horizontal shear plane. The bifurcation diagram obtained by simulation reveals several oscillatory solutions, including a wiggling motion that is predicted by a linear stability analysis. The flapping dynamics is shown to be a subcritical instability whose key ingredient is the finite extensibility of the disk. The behavior we observe has implications for emerging investigations on the flow dynamics of sheet-like particles, such as 2D polymers and 2D crystalline materials immersed in viscous fluids.
arXiv:2606.06235v1 Announce Type: new
Abstract: Mortality forecasting plays an important role in actuarial and policy decision-making, but its implementation remains technically complex and inaccessible to non-expert users. This project proposes a reliable large language model (LLM)-integrated interface that improves usability while maintaining statistical power. The LLM is designed as a constrained orchestration layer that translates natural-language inputs into structured configurations for a deterministic forecasting pipeline. A three-phase methodology is employed to ensure accuracy, usability, and transparency. First, a baseline pipeline is implemented using the CoMoMo package, reproducing established mortality forecasting results. Second, the pipeline is extended to generate multi-step forecasts using rolling-origin evaluation and mean squared error (MSE). Third, a prototype interface uses a local LLM to handle users' forecasting requests in plain language. The system demonstrates that LLMs can enhance accessibility without compromising reproducibility, transparency, or actuarial validity in high-stakes analytical workflows.
arXiv:2606.06236v1 Announce Type: new
Abstract: Pretrained diffusion models demonstrate impressive potential in solving highly ill-posed 3D computed tomography (CT) inverse problems, while the inference process suffers from significant computational overhead. Furthermore, existing uniform timestep schedules fail to capture the non-uniform evolution of the reverse conditional diffusion stochastic differential equation, thereby introducing substantial truncation errors. To overcome this limitation, we propose Tracing the Oracle (TrO), a plug-and-play framework for improved timestep scheduling. Specifically, we treat densely sampled numerical integration trajectories on a few samples as the reference oracle. The optimized schedule is extracted by leveraging dynamic programming to globally minimize the cumulative error between the few-step approximation and the oracle. This mechanism precisely allocates the limited sampling steps to critical evolution stages that are highly susceptible to truncation errors. Our extensive experiments on the AAPM dataset across multiple 3D CT reconstruction tasks demonstrate that, when combined with the state-of-the-art 3D CT reconstruction method DDS, our optimized timesteps significantly improve reconstruction fidelity and computational efficiency compared to existing heuristic schedules, especially under a strict budget of no more than 10 sampling steps.
arXiv:2606.06247v1 Announce Type: new
Abstract: In this paper, a high-order Fourier Continuation (FC) algorithm is introduced into the spectral smoothed particle hydrodynamics (SPH) scheme to simulate the wall-bounded incompressible flows. This work aims to extend the spectral ISPH scheme towards the high-order simulation of flows with non-periodic wall boundary conditions. Herein, a polynomial-based Fourier continuation technique is applied to the velocity and pressure to make the domain both periodic and Cp smooth. The spatial SPH discretisation is performed subsequently in the frequency space on the FC-extended domain by building upon the convolution theorem using fast Fourier transform (FFT). The incorporation of Neumann boundary conditions is straightforward, and more generally, the FC method enforces periodicity across the domain regardless of the boundary condition type. The convergence order, additional computational cost, and implementation technique of the FC method are also discussed. Combined with a projection-based time integration scheme and a spectral PPE solver, the FC-based spectral ISPH framework is validated against several classical CFD benchmarks. The principal finding of this work is that the incorporation of FC techniques enables the spectral ISPH scheme to simulate wall-bounded flows with high-order convergence, and accurately capturing complex vortex dynamics. This work therefore represents a step towards a fully high-order spectral Lagrangian SPH solver with complex geometries
arXiv:2606.06250v1 Announce Type: new
Abstract: Continuous-time SLAM provides a principled framework for fusing heterogeneous sensors while estimating smooth trajectories, and is particularly well-suited for handling heterogeneous, asynchronous sensor streams with non-uniform readout patterns, such as rolling shutter cameras, LiDAR scanners, radar sweeps, or event-based sensors. In this work, we introduce G-solver, a fully Gaussian and distributed framework that combines Gaussian Belief Propagation (GBP) with Gaussian Process (GP) motion priors for continuous-time trajectory estimation. Our GP model provides a probabilistic representation of the trajectory, enabling consistent interpolation and the use of data-driven hyperparameters, while GBP offers a scalable message-passing formulation well-suited for decentralized settings. The resulting solver naturally extends to multi-camera scenarios without specialized synchronization or engineering effort. We evaluate the approach on synthetic and real data, including rolling shutter and distributed multi-camera optimization, demonstrating accurate and stable estimation with runtimes comparable to existing continuous-time methods. An open-source implementation is released.
arXiv:2606.06103v1 Announce Type: new
Abstract: Medical image segmentation is often framed as a search for stronger architectures, but this can obscure a more fundamental question: what does the dataset require from the model? In medical imaging, this requirement is shaped by foreground occupancy, morphology, boundary ambiguity, topology sensitivity, annotation quality, acquisition variation, and operating point.
This paper introduces the Medical Segmentation Dataset Knowledge Card (MS-DKC), a framework for making these factors explicit. MS-DKC records dataset evidence through image/acquisition, morphology, supervision, context-dependence, and deployment-risk descriptors. These descriptors are mapped to failure modes, design priors, and risk-aligned criteria, making segmentation design more traceable than architecture-first comparison.
We evaluate MS-DKC on DRIVE, ISIC2018, and ACDC, representing distinct regimes. DRIVE contains sparse, thin, branching vessels, favoring detail-preserving models, sensitivity-aware optimization, threshold analysis, and topology-aware metrics. DKC-TNet-v2 achieved Dice 0.8044 and IoU 0.6730 with 35103 parameters, while SA-UNetv2-DKC-AmbRef reached Dice 0.8141, IoU 0.6865, sensitivity 0.8265, specificity 0.9804, and AUC 0.9853. ISIC2018 involves compact but appearance-variable lesions; validation-constrained score-function selection on Att-Next-Topo/ATTNext produced MS-DKC-AttNextTopo-VCSF-NoAug with Dice 0.8872, IoU 0.8214, precision 0.9173, Boundary F1 0.4878, and ASSD 4.13, while plausible additions failed to improve the risk-aligned profile. ACDC provides a multi-class cardiac case, where MS-DKC recommends four-class softmax segmentation, class-balanced Dice/CE supervision, and class-wise surface evaluation. Overall, the results support dataset-conditioned design: different datasets require different priors, operating points, and evidence before a model can be judged appropriate.
arXiv:2606.05513v1 Announce Type: new
Abstract: Epidemic LLM forecasters are usually trained and evaluated as static supervised models, whereas operational pandemic forecasting is a streaming process in which labels arrive after predictions and disease regimes shift over time. We study this mismatch in weekly COVID-19 hospitalization trend forecasting across five variant regimes. We introduce EpiEvolve, a self-evolving agent that wraps an LLM forecaster trained on the warm-start period and keeps its weights fixed during streaming. EpiEvolve adapts by storing forecast outcomes in a hierarchical episodic memory, reflecting on delayed labels, retrieving cases relevant to the current regime, and distilling recurring errors into strategic rules. The resulting context lets the forecaster reuse its own past predictions and outcomes in later weeks while following a chronological protocol that prevents future leakage. On the streaming dataset, EpiEvolve reaches $0.629$ average accuracy, compared with $0.561$ for the static backbone and $0.325$ for the external CDC ensemble, and reduces recovery lag after regime shifts from $5$ to $2$ weeks. Ablations show that reflection, strategic memory, and regime-aware retrieval each contribute to the gains.