arXiv:2605.21887v1 Announce Type: cross
Abstract: This paper utilizes the {\it Black Hole Vision} smartphone application to catalyze a pedagogical shift in General Relativity education through the quantitative analysis of simulated black hole imaging. Presented here for the Schwarzschild spacetime, the investigation is designed with a hierarchical modularity suitable for undergraduate students, with an expanded version intended for graduate courses in General Relativity or Relativistic Astrophysics. By transforming the mobile device into an educational relativistic imaging tool, we triangulate the simulated Schwarzschild mass through independent probes and characterize anisotropic coordinate transformations via a Jacobian map. Global numerical consistency is investigated through integrated coordinate length, while the exponential instability of nearly bound orbits is quantified through a measurement of the simulated Lyapunov exponent. Finally, symmetry is constrained through a sub-pixel constraint on eccentricity in the simulated spacetime. By integrating this statistical framework, the paper enables students to explore the distinction between physical signatures and instrumental noise using established metrological protocols.
Science Journals
arXiv:2605.22155v1 Announce Type: new
Abstract: Symbolic methods are generally not considered competitive with strong modern learners on realistic supervised tasks. We evaluate Algebraic Machine Learning (AML), a framework that learns through subdirect decomposition of algebraic structure rather than numerical optimization, against standard baselines on image and tabular classification across varying training-set sizes. We find that AML trained only on training data without using validation or cross-validation outperforms a family of cross-validated baseline methods including CNNs on small to medium image datasets (50--2000 training examples). On tabular datasets in the same size range, XGBoost is overall the best performing method, but AML is nonetheless comparable to methods incorporating task-specific biases such as LightGBM and random forests. AML achieves this competitive performance across two very different types of datasets using a generic algebraic inductive bias, rather than the modality-specific biases built into standard baselines like CNNs for images or XGBoost for tabular data, and requires no cross validation because it has no task-dependent hyperparameters to tune.
arXiv:2605.21960v1 Announce Type: cross
Abstract: Minimising EPR consumption is the dominant objective when routing a quantum circuit on a distributed quantum computer (DQC). We present dSABRE, a SABRE-style router for multi-core processors that, on each iteration of a lookahead-driven loop, first resolves any intra-core front-layer gates by SWAP scoring and only falls back to scoring inter-core teleportation candidates when the intra-core front is empty. Three mechanisms drive the improvement over the state of the art: a five-term gate-centric teleportation score that generalises the local SWAP heuristic to the inter-core setting, whose explicit capacity-penalty term keeps the scorer from teleporting into saturated cores; a proactive congestion-relief pass that redistributes idle qubits out of high-demand cores before deadlock; and a BFS-layer construction of the inter-core extended set that respects DAG dependencies layer by layer rather than mixing wires in topological order. Across 18 MQT-Bench circuits at 25, 36, and 64 logical qubits, dSABRE reduces geometric-mean EPR consumption by 41-44% over TeleSABRE and by 16-68% over the gate-teleportation-based pytket-dqc, using standard Qiskit SabreLayout for the initial layout. A large-circuit QFT sweep at 100-360 qubits confirms scalability. Code and online appendices are available at https://github.com/ebony72/dsabre.
arXiv:2605.22153v1 Announce Type: cross
Abstract: Open optical or plasmonic resonators are placed on and connected through surfaces or via waveguides, forming complex lightguiding nanostructures, e.g. for integrated photonic quantum devices. We derive general boundary conditions for quasinormal modes that account for the structure's specific geometry. We then present a general quantization scheme for multiple, interacting quasinormal-mode cavities coupled to quantum emitters and to a non-bosonic bath of propagating photons on waveguides or a surface. We derive a system-bath Hamiltonian with rigorously defined coupling elements that can be computed using Maxwell solvers, including light-matter coupling between the electromagnetic field and quantum emitters. We define system-bath correlation functions for an effective, bath-mediated, and time-delayed interaction between the quasinormal modes and quantum emitters, which is a main ingredient commonly used to simulate open quantum system dynamics.
arXiv:2605.22433v1 Announce Type: cross
Abstract: As trapped-ion quantum computing scales to larger qubit registers and more complex control protocols, classical control systems face a fundamental tradeoff: sub-microsecond board-level feedback requires tight hardware coupling, whereas maintainability and extensibility require clean, modular software abstractions. This paper presents QuCtrl-BELL (Bell), a compiler-driven software stack for trapped-ion quantum control. The design resolves this tradeoff by decoupling control flow -- including loops, branches, and synchronization -- from hardware state data. A Python-embedded domain-specific language (DSL) is lowered through a six-stage transpilation pipeline covering control flow graph (CFG) construction, static single-assignment (SSA) conversion, liveness analysis, and graph-coloring register allocation. The compiler generates deterministic distributed board-level programs and compact step-table data. A cross-board synchronization protocol supports feedback loops with latency below 700~ns without host intervention. Bell is deployed and evaluated on the QuCtrl-BELL platform (RISC-V + PXIe), demonstrating that a compiler-based infrastructure can provide programmability, deterministic timing, and modularity for scalable trapped-ion quantum control.
arXiv:2306.05905v2 Announce Type: replace
Abstract: A convenient approach to optimally solving combinatorial optimization tasks is the Branch-and-Bound method. Its branching heuristic can be learned to solve a large set of similar tasks. The promising results here are achieved by the recently appeared on-policy reinforcement learning method based on the tree Markov Decision Process. To overcome its main disadvantages, namely, very large training time and unstable training, we propose TreeDQN (Tree Deep Q-Network), a sample-efficient off-policy RL method trained by optimizing the geometric mean of expected return. To theoretically support the training procedure for our method, we prove the contraction property of the Bellman operator for the tree MDP. As a result, our method requires up to 10 times less training data and performs faster than known on-policy methods on synthetic tasks. Moreover, TreeDQN significantly outperforms the state-of-the-art techniques on a challenging practical task from the ML4CO competition.
arXiv:2402.06795v2 Announce Type: replace
Abstract: People naturally sketch strokes over graphical scenes to convey scene changes. We propose automatically interpreting these strokes to execute scene changes with squidgets (sketch-widgets), a novel sketch-based UI framework for direct scene manipulation. Squidgets are motivated by the observation that curves resulting from visually abstracting scene elements provide natural handles for the direct manipulation of scene parameters. Additional curves can be defined by users to author custom handles associated with scene attributes. Users manipulate a scene by simply drawing strokes, partially matched against scene curves to select a squidget and interactively control associated parameters. We present an implementation of squidgets within the 3D animation system Maya, showing 2D/3D stroke input to manipulate 2D/3D scenes. We report on a controlled experiment evaluating squidgets on 2D object translation and deformation tasks, and a broader informal study on squidget creation and manipulation.
arXiv:2605.22355v1 Announce Type: new
Abstract: Public transit route planning traditionally depends on structured map infrastructure and complex routing engines, and no existing dataset supports training models to bypass this dependency. We present TransitLM, a large-scale dataset of over 13 million transit route planning records from four Chinese cities covering 120,845 stations and 13,666 lines, released as a continual pre-training corpus and benchmark data for three evaluation tasks with complementary metrics. Experiments show that an LLM trained on TransitLM produces structurally valid routes at high accuracy and implicitly grounds arbitrary GPS coordinates to appropriate stations without any explicit mapping. These results demonstrate that transit route planning can be learned entirely from data, enabling end-to-end, map-free route generation directly from origin-destination information. The dataset and benchmark are available at https://huggingface.co/datasets/GD-ML/TransitLM, with evaluation code at https://github.com/HotTricker/TransitLM.
arXiv:2605.22356v1 Announce Type: new
Abstract: Large language models are increasingly used as computational tools for modeling human-like behavior. We introduce a behavioral induction framework that modifies model policies through fine-tuning on structured decision-making tasks: using synthetic datasets inspired by maladaptive behavioral patterns, including depression and paranoia, we train transformer-based language models to consistently select specific classes of actions across diverse contexts. We then test whether this behavioral optimization produces systematic changes in generative distributions.
Across two architectures, fine-tuned models show stable, context-general shifts in next-token probability distributions, including increased probability assigned to negative and threat-related interpretations in open-ended language tasks. These effects generalize beyond training contexts and are detectable in qualitative completions, psychometric-style evaluations, and quantitative distributional metrics such as Jensen-Shannon divergence.
Induced behavioral profiles also show partial specificity. Models optimized for different behavioral patterns exhibit dissociable response tendencies across evaluation probes, suggesting that structured behavioral training produces differentiated policy-level biases rather than generic distributional skew.
We interpret these findings as evidence that consistent behavioral optimization in LLMs can generate stable behavioral and distributional patterns consistent with altered latent priors, linking action selection and language generation. More broadly, the results support a view of LLMs as policy-based systems in which behavioral constraints shape emergent representational structure, highlighting their potential as controlled testbeds for studying the relationship between behavior, interpretation, and generative language in computational models of cognition.
arXiv:2605.22368v1 Announce Type: new
Abstract: As large language models (LLMs) are increasingly deployed for software engineering, constructing high-quality benchmarks is crucial for evaluating not just the functional correctness, but also the formal verifiability of generated code. However, existing benchmarks are limited by the quantity and quality of positive and negative test cases, leading to an overestimation of model capabilities in generating specifications and implementations. To address this, we propose VeriScale, a novel framework driven by the adversarial implementations. It consists of two stages: test-suite expansion to construct diverse and challenging test cases, and test-suite reduction to distill them into compact yet discriminative suites. While VeriScale is general, we instantiate it on Verina to construct VerinaPlus, which expands the original test suites by over 83$\times$, and VerinaLite, a lightweight 14$\times$ variant. Our experiments across eight state-of-the-art LLMs demonstrate that VerinaPlus exposes substantial model weaknesses hidden by the original benchmark, evidenced by sharp score drops on both SpecGen and CodeGen tasks, whereas VerinaLite maintains this discriminative power at a fraction of the evaluation cost. The enhanced benchmarks and source code are publicly available at https://github.com/XiaoyangLiu-sjtu/VeriScale.
arXiv:2605.22376v1 Announce Type: new
Abstract: Cross-domain offline reinforcement learning (CDRL) aims to improve policy learning in a target domain by leveraging data collected from a source domain. Existing works typically assess the transferability of source-domain data by measuring its similarity to target-domain transitions, and implicitly perform transition-level selection. Transitions that are considered similar are assigned higher weights or rewards, while dissimilar ones are down-weighted. However, transition-level similarity does not necessarily imply consistency in long-term returns. Even visually or dynamically similar transitions may lead to significantly different outcomes in the target domain, which can mislead policy learning and degrade performance. To address this issue, we revisit the fundamental objective of policy learning. Since policy optimization ultimately relies on Bellman targets to evaluate the quality of decisions, we propose to assess the transferability of source-domain transitions based on their alignment with target-domain Bellman targets, rather than superficial transition similarity. Based on this insight, we propose a method termed Target-Aligned Bellman Backup (TABB), which selectively leverages source-domain data by measuring their contribution to accurate Bellman target estimation in the target domain. We evaluate TABB across a broad range of cross-domain offline RL settings with highly limited target-domain data. Experimental results show that TABB consistently achieves strong performance.
arXiv:2605.22377v1 Announce Type: new
Abstract: Transformer-based language models such as BERT having 110M+ parameters have revolutionized natural language understanding, yet their internal mechanisms remain largely opaque to researchers and practitioners. Traditional attention-based interpretability methods often emphasize structurally important but semantically weak tokens such as punctuation marks rather than meaningful semantic relationships. This work introduces a lightweight and model-agnostic framework for quantifying token-level representational importance using hidden-state activation strengths at Layer 8 of BERT. The proposed Activation Flow Network (AFN) framework computes Token Activation Strength using the L2 norm of Layer-8 hidden representations, enabling direct ranking of semantically salient tokens. The study further introduces a threshold-based activation bucket formulation that partitions tokens into HIGH-activation and LOW-activation groups using an empirical upper-quartile activation boundary. Experimental observations demonstrate that semantically meaningful content words consistently occupy the HIGH-activation bucket and dominate representational activation shifts, while structurally supportive tokens contribute comparatively less. The results suggest that Layer 8 acts as a critical semantic consolidation zone balancing structural and semantic information processing. By revealing how activation magnitudes concentrate around semantically informative tokens, this work provides an interpretable and computationally efficient alternative to attentioncentric analysis, contributing toward transforming BERT from a "black box" into a more transparent "glass box" model for natural language understanding.
arXiv:2605.22379v1 Announce Type: new
Abstract: With the advancement of science and technology, the importance of emotion research has become increasingly evident. Electroencephalography (EEG)-based emotion recognition has emerged as an active research area in recent years, owing to its objectivity and high temporal resolution. However, most existing methods focus on optimizing encoder structures to enhance feature extraction capabilities, while paying relatively little attention to similarity calculation strategies, particularly overlooking the potential temporal misalignment of responses among different subjects. To address these shortcomings, this paper draws inspiration from the late interaction mechanism of ColBERT in natural language processing (NLP) and proposes a Temporal Asynchronous Alignment-based Contrastive Learning (TA2CL) framework. This method transforms the traditional global "hard alignment" similarity calculation approach into a fine-grained local matching mechanism, enabling the model to adaptively search for and align "locally highly correlated" segments between two EEG signals, thereby effectively mitigating the effects of inter-subject differences and temporal delays. Experimental results demonstrate that the proposed method achieves strong performance across multiple public datasets. Specifically, on the FACED dataset, it achieves an accuracy of 64.5% for the nine-class classification task and 79.5% for the binary classification task, while on the SEED and SEED-V datasets, it achieves accuracies of 86.4% and 70.1%, respectively, validating the method's effectiveness and generalization capability.
arXiv:2605.22385v1 Announce Type: new
Abstract: Explaining graph neural networks (GNNs) has become more and more important recently. Higher-order interpretation schemes, such as GNN-LRP (layer-wise relevance propagation for GNN), emerged as powerful tools for unraveling how different features interact thereby contributing to explaining GNNs. GNN-LRP gives a relevance attribution of walks between nodes at each layer, and the subgraph attribution is expressed as a sum over exponentially many such walks. In this work, we demonstrate that such exponential complexity can be avoided. In particular, we propose novel algorithms that enable to attribute subgraphs with GNN-LRP in linear-time (w.r.t. the network depth). Our algorithms are derived via message passing techniques that make use of the distributive property, thereby directly computing quantities for higher-order explanations. We further adapt our efficient algorithms to compute a generalization of subgraph attributions that also takes into account the neighboring graph features. Experimental results show the significant acceleration of the proposed algorithms and demonstrate the high usefulness and scalability of our novel generalized subgraph attribution method.
arXiv:2605.22387v1 Announce Type: new
Abstract: Accurate electricity price forecasting (EPF) is essential for market participants to support operational planning and risk management, yet remains challenging due to strong volatility, nonlinear dynamics, and frequent extreme price spikes. These challenges are particularly pronounced in the Australian National Electricity Market (NEM), where high renewable penetration further increases uncertainty. This paper investigates week-ahead electricity price forecasting and proposes a hybrid KAN+XGBoost framework that integrates Kolmogorov-Arnold Networks (KAN) with tree-based learning. The proposed approach combines the global nonlinear representation capability of KAN with the local robustness of XGBoost to capture both long-term dependencies and short-term price fluctuations. Experiments are conducted on real-world NEM data using an expanding window evaluation strategy. The results demonstrate that the proposed model outperforms benchmark methods, including SARIMAX, Long Short-Term Memory (LSTM), standalone KAN, and XGBoost, reducing MAE by approximately 12% compared to XGBoost and by over 50% compared to a naive baseline. The results suggest that hybrid learning strategies provide an effective and robust solution for electricity price forecasting in highly dynamic electricity markets.
arXiv:2605.22389v1 Announce Type: new
Abstract: Effectively training Large Language Models (LLMs) for complex, long-CoT reasoning is often bottlenecked by the need for massive high-quality reasoning data. Existing methods are either computationally expensive or fail to reliably distinguish high- from low-quality reasoning samples. To address this, we propose High-Entropy Sum (HES), a training-free metric that quantifies reasoning quality by summing only the entropy of the top (e.g., 0.5\%) highest-entropy tokens in each reasoning sample. We validate HES across three mainstream training paradigms: Supervised Fine-tuning (SFT), Rejection Fine-tuning (RFT), and Reinforcement Learning (RL), with extensive results demonstrating its consistent effectiveness and significantly reduced computational overhead. In SFT, training on the top 20\% HES-ranked data matches full-dataset performance, while using the lowest-HES data degrades it. In RFT, our HES-based training approach significantly outperforms baseline methods. In RL, HES-selected successful trajectories enable the model to learn strong reasoning patterns, significantly surpassing other compared methods. Our findings establish HES as a robust, training-free metric that enables a unified, effective, and efficient method for developing advanced reasoning in LLMs.
arXiv:2605.22401v1 Announce Type: new
Abstract: Does the relationship between learning rules and brain alignment generalize across species? We extend our prior finding that untrained CNNs match backpropagation at human V1 by testing the same five learning rules against macaque electrophysiology. The rules are backpropagation (BP), feedback alignment (FA), predictive coding (PC), spike-timing-dependent plasticity (STDP), and an untrained random-weights baseline. The macaque data come from two datasets: MajajHong2015 (V4/IT, 3,200 stimulus presentations, 88/168 neurons) and FreemanZiemba2013 (V1/V2, 135 stimuli, 102/103 neurons). Using RSA with identical model weights from our human study, we find: (1) all models achieve higher alignment with macaque early visual cortex (rho = 0.15-0.30 at V1/V2) than with human fMRI (rho = 0.01-0.08), consistent with the higher signal-to-noise ratio of electrophysiology; (2) STDP and PC produce the highest macaque V1/V2 alignment (rho ~ 0.30 and 0.28), consistent with their leading position among trained rules in human V1; (3) at IT, learning rule rankings show no detectable correlation across species (Kendall's tau = 0.00, p = 1.00), though this null result is expected given that n = 5 provides power only at tau = +/-1.0, and is further confounded by stimulus set differences; (4) a pretrained ResNet-50 (ImageNet) achieves rho = 0.25 at macaque IT, substantially above all custom CNN conditions (rho = 0.07-0.14), suggesting IT alignment is limited by model capacity and training data rather than by the learning rule. Noise ceilings, multi-seed variability (5 seeds), and a stimulus-control analysis are reported. These results demonstrate that early visual alignment is robust across species, while higher-area alignment is modulated by model capacity and stimulus domain.
arXiv:2605.22403v1 Announce Type: new
Abstract: Surface electromyography (sEMG) signal-based activity recognition has attracted increasing research attention in recent years. To develop accurate sEMG signal-based activity recognizers, numerous approaches have been proposed. Some studies focus on designing larger and more expressive model architectures to enhance the representational capacity of sEMG signals, while others aim to enrich model priors through large-scale pretraining, thereby improving recognition performance. Recently, large language models (LLMs) have shown remarkable generalization and reasoning capabilities in natural language processing, whose implicit knowledge, learned from extensive linguistic descriptions of actions, opens new possibilities for interpreting sEMG signals and inferring activity intentions. Motivated by this, we propose LLM-sEMG, a novel framework that leverages LLMs as sEMG activity recognizers. Within this framework, we design a language-oriented mapping mechanism that converts continuous sEMG sequences into sEMG language, integrating several strategies to further facilitate the signal-to-language mapping process. Extensive experiments demonstrate that the proposed framework achieves highly accurate sEMG signal-based activity recognition using large language models.
arXiv:2605.22413v1 Announce Type: new
Abstract: Extracting structured information from visual documents (Visual Information Extraction, VIE) is a cornerstone of business automation. While recent Multimodal Large Language Models (MLLMs) have shown promising capabilities, existing benchmarks suffer from critical limitations in scale and realism, lack semantic granularity, and fail to cover diverse document types. To bridge this gap, we introduce ReceiptBench, a large-scale, human-annotated benchmark consisting of 10k diverse receipts, organizing information extraction into four hierarchical sub-tasks: (1) Basic Perception for raw text spotting, (2) Format Normalization for strictly following standardization instructions, (3) Semantic Reasoning for inferring implicit attributes from context, and (4) Structure Parsing for handling nested line items. Furthermore, we propose a two-stage training framework incorporating Metric-Aware Group Relative Policy Optimization (GRPO), which translates rigorous evaluation constraints into reinforcement learning signals to enhance structural consistency. Extensive experiments demonstrate that our method yields state-of-the-art performance, surpassing leading proprietary models on complex reasoning tasks. We release our datasets and code at https://github.com/wwwT0ri/ReceiptBench.
arXiv:2506.23808v2 Announce Type: replace
Abstract: A recent series of works has shown that initialization-free BA can be achieved using pseudo Object Space Error (pOSE) as a surrogate objective. The initial reconstruction-step optimizes an objective where all terms are projectively invariant and it cannot incorporate knowledge of the camera calibration. As a result, the solution is only determined up to a projective transformation of the scene and the process requires more data for successful reconstruction.
In contrast, we present a method that is able to use the known camera calibration thereby producing near metric solutions, that is, reconstructions that are accurate up to a similarity transformation. To achieve this we introduce pairwise relative rotation estimates that carry information about camera calibration. These are only invariant to similarity transformations, thus encouraging solutions that preserve metric features of the real scene. Our method can be seen as integrating rotation averaging into the pOSE framework striving towards initialization-free calibrated SfM.
Our experimental evaluation shows that we are able to reliably optimize our objective, achieving convergence to the global minimum with high probability from random starting solutions, resulting in accurate near metric reconstructions.
arXiv:2605.22420v1 Announce Type: new
Abstract: Urban scene reconstruction from real-world observations has emerged as a powerful tool for self-driving development and testing. While current neural rendering approaches achieve high-fidelity rendering along the recorded trajectories, their quality degrades significantly under large viewpoint shifts, limiting the applicability for closed-loop simulation. Recent works have shown promising results in using diffusion models to enhance quality at these challenging viewpoints and distill improvements back into 3D representations. However, they often require costly per-scene optimization, and the distilled representations remain fragile and fail to generalize beyond limited synthesized views. To address these limitations, we propose GenRe, a novel diffusion-guided generalizable enhancer for urban scene reconstruction. GenRe takes as input any pretrained 3D Gaussian representation and fixes the deficiencies within a few minutes. By learning to distill generative priors across diverse scenes, GenRe produces robust and high-fidelity representation efficiently that generalizes reliably to challenging unseen viewpoints (e.g., lane change). Experiments show that GenRe outperforms existing methods in both quality and efficiency and benefits various downstream tasks, enabling robust and scalable sensor simulation for autonomous driving.
arXiv:2507.19464v3 Announce Type: replace
Abstract: Time-dependent density functional theory (TDDFT) is a standard approach for calculating optical excitations of molecules and solids, while ensemble DFT (EDFT) is a promising alternative under development. We introduce ensemble TDDFT (ETDDFT), a practical theory that combines the two, generalizing both; we ensemble-generalize the Gross-Kohn equation and the exchange-correlation kernel of TDDFT, and generalize EDFT to time-dependent problems. We relate coordinate scaling to the adiabatic connection. The new theory provides multiple avenues for constructing and using approximations. We illustrate these on the 2-site Hubbard model. We connect our results to the more general case of non-perturbative time-dependence.
arXiv:2510.15881v2 Announce Type: replace
Abstract: This work introduces ParamRF: a Python library for efficient, parametric modelling of radio frequency (RF) circuits. Built on top of the next-generation computational library JAX, as well as the object-oriented wrapper Equinox, the framework provides an easy-to-use, declarative modelling interface, without sacrificing performance. By representing circuits as JAX PyTrees and leveraging just-in-time compilation, models are compiled as pure functions into an optimized, algebraic graph. Since the resultant functions are JAX-native, this allows computation on CPUs, GPUs, or TPUs, providing integration with a wide range of solvers. Further, thanks to JAX's automatic differentiation, gradients with respect to both frequency and circuit parameters can be calculated for any circuit model outputs. This allows for more efficient optimization, as well as exciting new analysis opportunities. We showcase ParamRF's typical use-case of fitting a model to measured data via its built-in fitting engines, which include classical optimizers like L-BFGS and SLSQP, as well as modern Bayesian samplers such as PolyChord and BlackJAX. The result is a flexible framework for frequency-domain circuit modelling, fitting and analysis.
arXiv:2605.22448v1 Announce Type: new
Abstract: Multi-frame story illustration requires long-horizon coherence beyond single-image text-to-image generation, including narrative decomposition and persistent character identity, layout, and affect across frames. We propose Story-to-Executable Descriptions (S2ED), a training-free, model-agnostic, prompt-layer framework that converts a full story into a sequence of explicit, editable executable descriptions for more consistent rendering. S2ED coordinates three agents to segment the narrative, ground canonical character attributes, and enrich spatial and affective cues, enabling interpretable prompt-carried state propagation and local edits to repair drift without retraining the generator. Experiments on Flintstones and Shakoo Maku show that S2ED improves sequence-level consistency and character fidelity over strong prompting, large-model planning, and a reference training-based method, under both automatic metrics and human judgments. We also deploy S2ED in an end-to-end story-to-storybook system for children's illustrated stories, with a supplementary video.
arXiv:2605.22454v1 Announce Type: new
Abstract: Data rehearsal has emerged as a leading approach for mitigating catastrophic forgetting in Continual Reinforcement Learning (CRL). However, existing work remains confined to policy gradient frameworks, regularizing only actors due to the performance degradation incurred by critic regularization. This actor-centric approach overlooks the potential of data rehearsal for value function approximation. Moreover, existing evaluations in CRL rarely consider multi-cyclic environments where task sequences repeat, a critical real-world scenario that exacerbates forgetting and plasticity. We investigate data rehearsal for Deep Q-Networks using Q-value regularization in multi-cyclic settings and propose Qreg+NWLU which introduces two simple modifications: (1) continuous data rehearsal that dynamically collects and updates stored Q-values throughout training, and (2) "No-Wait" regularization that applies immediately rather than after the first task. Together, these modifications yield improvements in learning efficiency, forgetting mitigation, and knowledge transfer over Qreg and conventional CRL methods within value function approximation settings.