Papers

Machine learning models for remote sensing are trained and deployed on a static set of modalities. However, as we equip newer satellites with novel sensors and retire old ones, practitioners may wish to deploy an existing model on a substitution, superset, or subset of modalities with minimal retraining given data availability or practical computational constraints. We study the setting of updating existing models to changing modalities and identify three main scenarios: Modality Transfer (substitution), Addition (superset), and Peeking (subset). We propose DeluluNet, an architecture with modular components for all three changing modality scenarios. DeluluNet is trained end-to-end, learning a multi-modal model from a unimodal teacher and unlabeled multimodal data via modality hallucination–predicting missing modality representations from those that are present. As a result, DeluluNet can keep predicting even when input modalities change, providing a practical alternative to re-labeling and re-training in a changing world.

Precise knowledge of nuclear structure is essential across fundamental physics, yet probing these structures is notoriously difficult. To address this challenge, ultra-peripheral collisions (UPCs) provide a femtoscopic tomography for imaging the atomic nucleus. UPCs offer a pristine electromagnetic pathway: coherent vector-meson photoproduction generates patterns of diffraction and two-source interference that directly encode the nuclear spatial density. Turning these patterns into quantitative constraints is, however, a challenging inverse problem, complicated by correlated sensitivities to deformation and neutron skin, phase smearing, and experimental backgrounds. Here we introduce an interpretable Multitask deep-learning framework that maps transverse momentum distributions to multiple nuclear-structure indicators simultaneously and identifies the kinematic regions driving each inference. We demonstrate the approach with coherent $J/\psi$ photoproduction in $^{96}_{40}\text{Zr} + ^{96}_{40}\text{Zr}$ collisions, showing that the learned features separate diffraction-dominated and interference-dominated information and provide analysis-ready observables for future high-luminosity data.

We present a superhuman AI agent for Generals.io, a real-time strategy game that combines long-horizon planning, imperfect information, and combinatorially large state and action spaces. Much prior work on large games relies on heavy machinery such as leagues of past opponents, reward shaping, and human data. We instead train a transformer policy with plain policy-gradient self-play on the sparse win/loss reward, and carefully ablate which ingredients carry the performance: an exponential moving average of the parameters is consistently the stronger deployment policy, and training only on the highest-advantage samples halves the cost of a training iteration. With seven days on four GPUs the agent reaches superhuman play: it ranked first in all ten weekly standings of a two-month deployment on the public 1v1 ladder, and it beat the two strongest humans 189-67 and 32-18. We also release a JAX-native simulator that reaches over 40 million frames per second on one GPU.

We study the limits of CTC-internal scoring for N-best hypothesis selection and locate the information bottleneck separating acoustic confidence from linguistic plausibility. Eleven CTC-internal and acoustic-feature scoring strategies produce no statistically significant WER improvement over greedy decoding on LibriSpeech dev-other at G=16 (all p > 0.05). The exhaustion is systematic: CTC's Spearman $\rho$ between hypothesis score and per-utterance WER degrades from -0.574 at G=4 to -0.270 at G=128, a 53% loss driven by blank-path proliferation. This establishes that the discriminative capacity of CTC-internal representations is saturated: no recombination of acoustic signals can close the oracle gap. Confirming that the bottleneck is linguistic, not acoustic, external linguistic information introduced via MBR decoding breaks through it. MBR-CER decoding with a RoBERTa pseudo-log-likelihood (PLL) posterior ($\tau$=10, G=128) achieves 5.42% WER on held-out LibriSpeech test-other (greedy 5.96%, $\Delta$=-0.535 pp, p<0.0001, 9.0% relative). RoBERTa PLL $\rho$ degrades only 21% over the same range, retaining discriminating power where CTC loses it. Applied without retuning across two Zipformer architectures, three domains (LibriSpeech, TED-LIUM 3, VoxPopuli), and four MUSAN noise levels, the recipe gives significant gains in 11 of 13 conditions. On the training side, standard MWER training via the CTC forward-backward algorithm implements Rao-Blackwellized REINFORCE at the output projection (variance about 3x below Viterbi). Yet sequence-level fine-tuning fails at near-converged checkpoints: all four MWER configurations on CR-CTC collapse (+6.18 to +8.90 pp WER), as a training oracle gap of 0.007 pp provides no usable reward signal.

Configuring the hyperparameters of Mixed-integer programming (MIP) solvers is a high-dimensional, instance-dependent optimization problem where suboptimal settings can degrade solving time by orders of magnitude. Default configurations are often suboptimal, while traditional tuning methods either suffer from the “cold-start” problem and inefficient search or heavily rely on expert experience. This paper introduces GRIMIP (\textbf{\underline{G}}eneral \textbf{\underline{R}}easoning for \textbf{\underline{I}}nstance-specific \textbf{\underline{MIP}} configuration), a novel hybrid intelligence framework that synergistically integrates the semantic reasoning capabilities of Large Language Models (LLMs) with the sample-efficient search of Bayesian Optimization (BO). GRIMIP enables the LLM to function as a complete probabilistic surrogate within the BO loop, significantly improving performance and reducing sampling and evaluation costs. On seven benchmarks including MIPLIB, GRIMIP achieves over 40% reduction in Primal-Dual Integral on hard instances, outperforming SMAC and other LLM-assisted BO methods. By granting LLMs sufficient autonomy, GRIMIP combines the expert-level reasoning of LLMs with the efficient search of BO, achieving state-of-the-art performance.

In this paper we prove some concentration inequalities for two types of error probabilities in the Empirical Risk Principle (ERP) in statistical learning, which provide a lower bound and an upper bound for the minimal risk (in terms of the minimal empirical risk) with non-asymptotic high confidence. The usual boundedness condition of the empirical risk function is relaxed to the Gaussian or exponential integrability condition. The confidence of the lower bound of the minimal risk is shown to be independent of the number of training parameters and the dimension of the input vectors, allowing one to detect the deficiency of a learning machine efficiently; and the confidence of the upper bound of the minimal risk is proved to be high provided that the sample size $n$ is much greater than the box dimension of the parameter set $\Theta$ in the Orlicz metric $d_{\psi_1}$ associated with the risk functions. Our work is based on Talagrand's concentration inequalities (the sharp versions by Bousquet and Klein-Rio), transport-entropy inequalities and the recent progress in the theory of empirical processes and statistical learning.

Sorting a huge stream of waste accurately within a short period can be done with the support of digitalization, particularly Artificial Intelligence, instead of traditional methods. The overlap of Artificial Intelligence and Circular Economy can flourish many services in the environmental technology domain, in particular smart ewaste recycling, resulting in enabling circular smart cities. We analyse the growing need for automated ewaste recycling as an essential requirement to cope with the fast growing ewaste stream and we shed the light on the impact of Artificial Intelligence in supporting the recycling process through smart classification of devices, where the smartphone is our case study. Our study applies transfer learning as a special technique of Artificial Intelligence by finetuning the output layers of AlexNet as a pretrained model and perform the implementation on a small size dataset that contains 12 classes from 6 smartphone brands. We evaluate the performance of our model by tuning the learning rate, choosing the best optimizer, and augmenting the original dataset to avoid overfitting. We found that the optimizer of Stochastic Gradient Descent with Momentum and 3e-4 as a learning rate brings almost 98% model accuracy with generalization. Our study supports automated ewaste recycling in decreasing the error rate of ewaste sorting and investigates the advantages of applying transfer learning as the best scenario to overcome the rising challenges.

Knowledge Editing (KE) has emerged as a frontier for updating specific facts in LLMs without costly retraining, but its reliability and underlying mechanisms remain poorly understood. In this work, we examine KE from an adversarial elicitation perspective, revealing that edited knowledge is often not fully erased and continues to surface, with consistent failures observed across diverse model architectures. To explain this behavior, we conduct a mechanistic analysis of popular KE methods. We show that low-rank updates do not overwrite existing knowledge but instead redistribute it within the model's representation space. Furthermore, we find that these methods act as targeted suppression mechanisms that reduce the likelihood of expressing original facts, rather than removing them from the model. Analysis of the loss landscape reveals that edited knowledge lies in narrow, anisotropic regions that are highly sensitive to perturbations, making them highly vulnerable to indirect prompting and adversarial attacks. By exposing these profound architectural vulnerabilities, our work proves that KE algorithms are inherently bypassable and motivates a fundamental reevaluation of how we deploy post-hoc updates in several LLM applications.

In multimodal transportation systems, shared mobility services (SMSs) are promoted for their potential to enhance flexibility and reduce congestion. However, SMS demand is often concentrated in high-density areas, which can limit the effectiveness and accessibility for various commuter groups. This uneven integration challenges transportation system efficiency, especially in terms of emissions and spatial equity. Addressing these issues requires coordination among multiple stakeholders whose objectives frequently conflict. Whereas authorities aim to ensure sustainable and equitable mobility, SMS providers focus on revenue maximization, and travelers seek to minimize personal travel costs. This paper proposes a multi-agent deep reinforcement learning framework that captures these interactions through dynamic pricing and incentivization strategies for SMSs and public transport. The framework integrates two reinforcement learning (RL) agents: (i) a public authority that allocates spatio-temporal public transport incentives to improve equity, emissions, and efficiency, and (ii) an SMS provider that dynamically adjusts fares to optimize revenue. The agents interact with the transportation system and adapt strategies in response to evolving demand, congestion, and network conditions. Numerical experiments conducted over a three-hour morning peak period show that dynamic incentivization effectively reduces congestion peaks, lowers commuters' costs by around 20% and emissions by approximately 10%, while nearly doubling public transport profit and supporting a more equitable distribution of benefits. When combined with dynamic SMS pricing, the two RL agents demonstrate the ability to balance conflicting objectives between private providers and public authorities. The proposed approach provides a decision-support tool for sustainable and equitable multimodal mobility planning.

High-fidelity simulations of free-surface flows using Lagrangian methods such as the Particle Finite Element Method (PFEM) are computationally demanding due to continuous domain updates and repeated solution of the governing equations. This challenge is further amplified by non-Newtonian rheologies, where material nonlinearities increase computational cost. These limitations motivate the development of efficient surrogate models to approximate PFEM dynamics at reduced cost. While data-driven deep learning approaches are promising, a key challenge is designing models that operate on arbitrary and evolving geometries. We propose a self-attention-based neural surrogate for PFEM simulations of free-surface flows. The architecture leverages attention mechanisms to model node interactions and capture complex spatial dependencies, while preserving the PFEM mesh discretization. This provides a geometric and topological framework for remeshing and node redistribution, maintaining high-quality spatial discretization during rollouts, improving long-term stability, and enabling reconstruction of derived mechanical quantities via standard finite element operators. Two attention formulations are considered: a standard self-attention mechanism and a linear variant that reduces computational cost and improves scalability. The models are evaluated on two- and three-dimensional free-surface flow benchmarks with evolving geometries, varying material parameters, and non-Newtonian fluids. Results show accurate prediction of transient dynamics and final configurations, with significantly improved scalability. The mesh-based formulation also enables direct reconstruction of quantities such as stress fields. Overall, the framework provides an accurate and scalable surrogate strategy for PFEM simulations in engineering-scale applications.

Clinical audio diagnosis in low-resource settings requires models that identify conditions from minimal examples without large annotated corpora. We propose Federated Self-Contextualization (FSC), a multimodal language model framework for in-context clinical audio diagnosis across federated hospital clients. FSC constructs pseudo-label episodes via unsupervised clustering of audio representations, bypassing scarce real diagnostic labels, and enables contextual reasoning from support-query pairs. Our progressive three-stage pipeline first aligns audio embeddings with the language model via caption-based pretraining, then adapts it for episodic in-context inference through federated optimization. At test time, given a small labeled support set, the model diagnoses an unseen query through multimodal reasoning. On held-out respiratory and cardiac conditions, FSC achieves 71.6% accuracy in 2-way 2-shot evaluation, outperforming audio-language baselines by over 9%.

Apnoea of prematurity is characterised by recurrent episodes of cessation of breathing and remains difficult to detect reliably using routinely monitored physiological signals in the Neonatal Intensive Care Unit (NICU). Existing bedside monitors rely primarily on respiratory rate and oxygen saturation thresholds, often generating high false-positive alarm rates and missing short or irregular events. Improving automated detection using routinely acquired clinical signals could enhance identification of clinically meaningful events without additional sensing hardware. We evaluated deep learning-based detection of apnoea-related Cessation Of BrEathing (COBE) events using impedance pneumography (IP), electrocardiography (ECG), and photoplethysmography (PPG) signals from approximately 430 hours of NICU recordings collected from 24 pre-term infants. Three independent reviewers annotated COBE events, producing a dataset of 346 COBE and 608 non-COBE events. We compared a shallow convolutional neural network (CNN), residual networks (ResNets), and a ConvNeXt architecture using an independent held-out test set. Across all architectures, detection performance was influenced more strongly by signal modality than by architectural complexity. Unimodal IP-based models achieved balanced accuracies of 86.8-88.0%, outperforming ECG-derived (62.6-69.7%) and PPG-derived (65.1-66.4%) respiratory surrogates. Multimodal fusion yielded modest improvements over IP alone. The best-performing model, a ConvNeXt architecture combining IP and PPG inputs, achieved 88.7% balanced accuracy and an F1 score of 0.75 on the independent test set. These findings demonstrate that deep learning models applied to routinely monitored NICU signals can reliably detect COBE events and highlight the importance of signal modality in data-constrained neonatal monitoring settings.

Embedded global navigation satellite system (GNSS) interference monitoring requires fast and memory-efficient inference to process large volumes of raw in-phase and quadrature (IQ) samples in real time. At the same time, increasingly expressive deep neural networks (DNNs) are needed for robust interference classification and characterization across diverse signal conditions. This creates a fundamental tension between predictive performance and deployability on resource-constrained hardware. In this paper, we investigate efficient network inference for GNSS interference characterization using iterative structured pruning, post-training static quantization, and hardware-aware zero-shot neural architecture search (NAS). Starting from MCUNet as a compact baseline, we analyze how model compression and automated architecture optimization affect model size, computational complexity, and memory usage while maintaining task performance. Experiments on a GNSS interference dataset, covering both classification and generalized characterization, show the benefits of combining compression and hardware-aware design for embedded deployment. Our results provide practical guidance for developing compact machine learning (ML) models for real-time GNSS interference monitoring on embedded platforms (iMXRT1062 MCU, Raspberry Pi Zero 2W, and Raspberry Pi 5).

Deep Learning (DL) has advanced various fields by extracting complex patterns from large datasets. However, the computational demands of DL models pose environmental and resource challenges. Deep Shift Neural Networks (DSNNs) present a solution by leveraging shift operations to reduce computational complexity at inference. Compared to common DNNs, DSNNs are still less well understood and less well optimized. By leveraging AutoML techniques, we provide valuable insights into the potential of DSNNs and how to design them in a better way. We focus on image classification, a core task in computer vision, especially in low-resource environments. Since we consider complementary objectives such as accuracy and energy consumption, we combine state-of-the-art multi-fidelity (MF) hyperparameter optimization (HPO) with multi-objective optimization to find a set of Pareto optimal trade-offs on how to design DSNNs. Our approach led to significantly better configurations of DSNNs regarding loss and emissions compared to default DSNNs. This includes simultaneously increasing performance by about 20% and reducing emissions, in some cases by more than 60%. Investigating the behavior of quantized networks in terms of both emissions and accuracy, our experiments reveal surprising model-specific trade-offs, yielding the greatest energy savings. For example, in contrast to common expectations, quantizing smaller portions of the network with low precision can be optimal with respect to energy consumption while retaining or improving performance. We corroborated these findings across multiple backbone architectures, highlighting important nuances in quantization strategies and offering an automated approach to balancing energy efficiency and model performance.

Attributed graph clustering partitions nodes by jointly exploiting node attributes and graph topology. It remains challenging due to attribute heterogeneity and representation degradation during graph learning. Real-world datasets often contain heterogeneous attributes, i.e., numerical and categorical attributes, complicating unified representation learning. This challenge becomes more complex in attributed graphs, where constructing a clustering-friendly graph structure from attributes and topology remains difficult. Under deep graph architectures, repeated graph propagation causes node embeddings to become overly similar, leading to the over-smoothing (OS) effect. Meanwhile, graph representation learning amplifies topological influence, making discriminative attribute information harder to exploit for clustering, an effect we refer to as over-dominating (OD). To bridge these gaps, an end-to-end framework, Any-type attributed Graph REpresentation lEarning (AGREE), is proposed. It unifies attributed graphs and any-type attributed data through multi-level alignment and similarity-based graph construction. Quaternion-based graph convolution strengthens attribute interaction to alleviate OD, while shallow graph architectures help relieve OS. The learned embeddings are jointly optimized for graph reconstruction and clustering, without requiring a predefined number of clusters during training. Experiments on diverse benchmarks show that AGREE achieves strong overall performance in accuracy, robustness, and adaptability.

We study mirror flows generated by a convex quadratic loss and a general convex lower semicontinuous mirror potential. We show that, when initialized near the boundary of the domain of the mirror potential, their rescaled trajectories converge to a limiting mirror flow whose potential is the indicator function of the domain. In this limit, the primal variable minimizes the loss over a time-dependent hypothesis set: the subdifferential of the support function of the domain, evaluated at the dual variable. This characterization provides a general mechanism for incremental learning in mirror flows.

Large Language Model (LLM) agents increasingly rely on memory systems to maintain long-term coherence. Recent work shows that agent memories degrade during continuous consolidation. However, existing research assumes memories are derived from unbiased experiences. In this work, we identify and formalize a novel phenomenon: Memory Contagion – the cross-temporal propagation of evaluator bias through agent memory. We show that when agents are trained or guided by biased evaluators, their experiences become biased; when these trajectories are stored and consolidated into memory, the bias propagates to future agents retrieving from the same memory store, even when consolidation is perfect (oracle). Across two bias types (length preference, authority bias) and four experimental phases, we demonstrate: (1) Memory Contagion occurs for length bias even with perfect consolidation on older models (Gamma_A = 13.18, DeepSeek V4-Chat), while newer models (V4-Pro, Claude) are immune, proving both that biased input is a sufficient cause and that contagion is model-generation-dependent; (2) authority bias fails to propagate in all 15 controlled multi-seed experiments (Gamma_A = 0.00), revealing that not all evaluator biases can cross temporal boundaries through current memory architectures; (3) No observed safe threshold: length bias propagation is detected at contamination rates as low as p=0.2. Our findings expose a critical but contingent vulnerability in current agent memory designs and provide formal tools for measuring cross-temporal bias propagation.

Black-box optimization (BBO) deals with problems where objective functions lack explicit analytical forms and are expensive to evaluate. Factorization machine with quadratic-optimization annealing (FMQA) constructs a surrogate model using a factorization machine (FM) and optimizes it with an Ising machine. Conventional FMQA applies a single integer-binary encoding throughout the optimization process, although the encoding best suited to surrogate learning may differ from the one best suited to Ising-machine solution search. We propose a stage-dependent FMQA framework and derive conversion formulas between one-hot and domain-wall QUBO matrices that preserve the surrogate objective over feasible integer states up to an additive constant. We evaluate the OhDw variant, which employs one-hot encoding for learning and domain-wall encoding for search, on the Rastrigin function with input dimensions N = 2 and 5 and discretization levels q = 61 and 301. Across all conditions, the dominant factor governing optimization performance is the encoding used in the learning stage, with one-hot encoding consistently yielding lower residual errors than domain-wall or binary encoding. The additional benefit of switching to domain-wall encoding for solution search is condition-dependent. For N = 5 and q = 301, OhDw achieves a lower residual error and solutions closer to the global optimum than one-hot-only FMQA, whereas for N = 5 and q = 61 the latter achieves a lower residual error. These results indicate that one-hot encoding in the learning stage is the primary performance driver and that stage-dependent encoding can provide further improvement under finer discretization.

The recently introduced EML (Exp-Minus-Log) function acts as continuous analogue of NAND gates, providing a compositional building block capable of representing elementary functions. In this work, we study the expressive power of tree-structured compositions of EML functions. We show that such trees enjoy a universal approximation property for functions in $W^{k, \infty}$ for $k \in \mathbb N$, drawing on classical neural network approximation arguments while exploiting the ability to explicitly construct EML trees that mimic polynomial representations. We further propose a learning algorithm for EML-type trees equipped with fitting parameters, and demonstrate its feasibility in practical optimization problems. Our results establish EML trees as a theoretically grounded framework for function approximation.

AI scientist systems are described as tools, coauthors, or founders, but we evaluate them as if only the final answer matters. This position paper argues that outcome-only evaluation is insufficient, and that task outcome, mechanism fidelity, and epistemic honesty must be measured separately. Our evidence comes from 28 episodes of a coding agent attempting to rediscover a known particle identification observable in a Geant4 simulation, including an 8-episode probe across two additional frontier models. In 4/20 primary-model and 3/8 cross-model episodes, agents reach right-looking results through incorrect reasoning that breaks when conditions change, which we call Correct Answer, Wrong Mechanism (CAWM). Honesty and mechanism fidelity dissociate within a single agent trajectory. When given a partially misleading prior, all five agents reject the false component on evidence, yet one defends its chosen observable with physics inconsistent with its own data. In the simulation-based discovery setting studied here, coding agents prove reliable tools but unreliable scientific co-authors for open-ended claim-making, where co-author trust requires mechanism-fidelity verification they do not reliably self-apply. The failure is detectable, and we propose a lightweight test. A one-step regime-shift check needs only the agent's claim and flags the over-generalized cases. A companion recomputation flags the remaining cases when the correct observable is known. Together, these checks flag every CAWM case in this study.

There has been rapid progress in generative artificial intelligence (AI) models for inorganic crystal design, which can efficiently generate large numbers of candidate compounds after being trained on databases of known crystals. However, it remains unclear whether they genuinely expand the accessible materials search space beyond conventional strategies such as elemental substitution within known structure types. We address this question by developing a workflow to assess whether AI-generated crystals are duplicates of training structures, reproducible by elemental substitution, or unmatched by either criterion. Applying this workflow to representative generative models reveals that 81-92% of chemically valid and metastable generated crystals are either training duplicates or substitution-derived structures. This tendency is particularly strong in high-symmetry crystal systems, even though many possible structural prototypes remain unexplored. Further analysis of the underlying structural fingerprints shows that low-symmetry structures beyond duplication or substitution can be interpreted as interpolation in training-data-rich regions, while high-symmetry duplicates appear to result from memorisation in training-sparse regions. Our findings highlight a limitation in the current generation of models that exhibit a bias towards known structural prototypes in the high symmetry regions, but enable wider exploration of the low-symmetry structural space.

Mean field games efficiently approximate a very large population of strategic agents. While these games can aid the understanding of complex systems, their deployment in real-world settings is challenged by the specification of their parameters: mean field games (MFGs) often involve hidden preferences, constraints, and interactions that can rarely be theoretically derived or directly observed. To address this gap, we present a neural network-based framework for learning parametric, finite-state MFGs from observed population dynamics. To do so, we formulate the parameter calibration as an inverse problem and use implicit differentiation to backpropagate through the games' equilibrium. The resulting approach is fully differentiable and enables us to estimate flexible trajectory-wise parameter paths, including state- and time-dependent specifications without requiring observations of the individual agents' actions or rewards. We provide a proof for the exactness of the gradient computation in a discrete-time formulation. We validate our framework through numerical experiments across four systems of increasing complexity, ranging from synthetic linear-quadratic benchmarks to real-world urban mobility datasets.

Operational event-detection systems are rarely assessed by pointwise accuracy alone. In anomaly detection, changepoint detection, and warning systems, the utility of an alarm depends on its temporal position relative to an event. This produces a score-loss mismatch. Neural networks are commonly trained with classical loss functions, such as cross-entropy, whereas deployment decisions are obtained by thresholding network predictions, merging alarms through post-processing rules, and evaluating them with event-based metrics defined by detection windows and false-alarm costs. This paper studies a temporally localized specialization of weighted score-oriented loss (wSOL) for event prediction. Starting from score-oriented losses based on expected confusion matrices and from the weighted SOL framework of Marchetti et al., we consider temporal weights that discount near-event false positives and reduce false-negative penalties when an event is preceded by an admissible alarm. The resulting objective is differentiable with respect to the network predictions, and therefore can be optimized by back-propagation. It can be instantiated with balanced accuracy, true skill statistic, F1, critical success index, and related confusion-matrix scores. We evaluate the proposed approach by comparing cross-entropy, unweighted score-oriented loss, and wSOL on three benchmark datasets for time-series event prediction and detection. The results show that wSOL can improve performance when the evaluation utility is localized in time and is not already encoded by the pointwise labels.

Finding the shortest path in non-geometric network graphs, where edge weights encode arbitrary metrics such as latency or monetary cost rather than spatial distance, poses a challenge for informed search algorithms. Their efficiency depends on an informative heuristic, typically supplied in spatial domains by geometric distances that have no counterpart on non-geometric graphs. We propose a large language model (LLM)-aided A* algorithm in which an LLM generates intermediate waypoints that guide the A* expansion toward promising graph regions. At the core of the approach are landmark distances, which serve both as an admissible landmark-based (ALT) heuristic for the search and as a compact structural feature that, supplied to the LLM, restores the distance-to-destination signal it would otherwise lack on non-geometric graphs. Our comprehensive experiments on multiple graph topologies with up to 2,000 nodes demonstrate that LLM-generated waypoints reduce the number of expanded nodes by around 50% while incurring only a marginal path cost increase compared to the optimal solution. We further analyze the impact of prompt engineering and show that incorporating compact structural features, namely heuristic estimates, is more effective than advanced prompting techniques. These findings demonstrate the potential of combining LLM- based guidance with classical search algorithms for efficient network optimization.

Implicit Neural Representations (INRs) have been proven successful in encoding continuous signals through coordinate-based networks, yet facing a spectral dilemma: periodic activations capture fine details but act as all-pass filters that memorise noise, while spatially compact activations regularise effectively but suffer from low-frequency bias. Existing attempts to resolve this trade-off introduce computational overhead or tuning frailty. We propose to model each neuron's activation as the steady-state response of a sinusoidally-forced damped harmonic oscillator, whose amplitude naturally governs the network's spectral selectivity during training. By jointly optimising the oscillator parameters alongside the network weights, our method adapts to the target signal's spectral content without explicit regularisation. Initialised in the stopband, the network exhibits a coarse-to-fine learning curriculum that progressively expands its spectral gate, capturing low-frequency structures first and high-frequency details only when justified by the reconstruction objective. Comprehensive experiments show that our approach consistently achieves state-of-the-art or competitive results against established INRs, while requiring no task-specific tuning of any hyperparameters.

Deep learning models, particularly Transformers and Neural Operators, exhibit a well-documented "spectral bias," effectively acting as low-pass filters that smooth out high-frequency information. While benign in fluid dynamics, this bias is catastrophic for Chaotic Dynamical Systems, where the underlying strange attractor is characterized by fractal geometry and infinite spectral density. We introduce the Fractal Neural Operator (FNO), a novel architecture that utilizes a non-resonant prime number basis to approximate continuous dynamical systems. Unlike geometric encodings ($2^k$), which suffer from spectral gaps and resonance, our Harmonic Weierstrass Encoder injects infinite spectral resolution into the latent space. We demonstrate that FNO extends the valid prediction horizon of the Lorenz-63 system to 347 Lyapunov times, exceeding state-of-the-art Reservoir Computing baselines by a factor of 2.3x. These results suggest that "chaos" is not inherently unpredictable to neural networks, but rather requires non-differentiable, fractal embedding manifolds.

Source-free domain adaptation (SFDA) adapts a pretrained model to an unlabeled target domain without accessing source data, enabling the reuse of time-series predictors across sensing systems. Existing methods commonly update the encoder using unlabeled target objectives. Yet changes in users, equipment, and operating conditions also alter the spectral characteristics of input signals, raising a fundamental question: does adaptation require changing the encoder, or can we instead adapt the signal it receives? We propose Source Free Frequency-Prototype Alignment (SF-FPA), which learns a spectral input transformation while freezing the source encoder and classifier parameters. First, a frequency adaptation layer applies amplitude and phase corrections to target signals. Separate branches adjust spectral magnitudes and phase relationships, with circular phase encoding avoiding discontinuities at the angular boundary. Second, confidence-filtered prototype alignment guides target representations toward source class prototypes. Combined with information maximization, this objective propagates gradients through the source predictor to learn spectral corrections that support class discrimination. Extensive experiments on three real-world time series benchmarks demonstrate state-of-the-art average Macro-F1 performance. Component ablations and controlled comparisons with encoder adaptation and a parameter-matched time-domain adapter further support spectral input adaptation as an effective approach to reusing source-trained predictors. The implementation is available at https://github.com/ShichangMeng/SF_FPA.

Multi-turn tool-using agents must coordinate long-horizon tool sequences while tracking dialogue state and policy constraints. Existing approaches often separate inference-time orchestration from parameter-level learning, leaving tool selection weakly structured and preference updates vulnerable to train–deployment prompt mismatch. For within-benchmark self-improvement, ToolGraph combines schema-derived topology, transition weights estimated from successful rollouts, and history-aware controls for write prerequisites and repeated-search loops. We then construct 161 preference pairs by locating divergence points via state-based matching and prefix-based alignment, filtered through action-correctness annotations, and train DPO under the same ToolGraph context used at inference. Across 375 tau2-bench tasks, ToolGraph raises the weighted average reward from 0.304 to 0.338 (+11.2% relative), while ToolGraph+DPO reaches 0.355 (+16.8% over the baseline), with the DPO gain concentrated in airline and retail. Fine-grained diagnostics further show that roughly half of telecom trajectories exhaust the step budget before action execution and that chosen reward positivity is the most useful checkpoint signal across our 16 evaluated DPO configurations.

Outer-loop link adaptation (OLLA) is widely deployed in 5G NR to track channel variations, yet its reliance on first-order, single-bit feedback degrades performance significantly under high-mobility and fast-varying channels. This paper presents LOLLA (Learned Outer-Loop Link Adaptation), a deep reinforcement learning framework that replaces the conventional OLLA staircase with a learned, continuous SINR offset conditioned on rich PHY/MAC telemetry inaccessible to OLLA. The offset modulates the SINR-to-MCS lookup table, preserving 3GPP-compliant MCS selection and provably subsuming the conventional OLLA update rule. A Proximal Policy Optimization (PPO) policy trained under a Lagrangian block error rate (BLER) constraint automatically enforces tunable reliability targets from 1% to 15% without manual penalty calibration. The framework is realized as the first closed-loop AI-native control dApp on a GPU-accelerated 5G NR stack, achieving end-to-end control latencies under 500 microseconds. Evaluations under 3GPP TDL channel models demonstrate 15% to 92% throughput gains over OLLA across Doppler frequencies up to 400 Hz, while attaining a Pareto frontier that strictly dominates OLLA across all evaluated reliability targets. The learned policy generalizes to unseen channel models and scales to eight concurrent UEs under shared-resource scheduling. In the uplink formulation, the gNB directly observes decoding outcomes, enabling simulation-to-deployment parity.

On-policy distillation (OPD) improves LLM reasoning by training a student model on its own generated outputs, but standard OPD treats all student-generated outputs (SGOs) equally regardless of their informativeness. We observe a consistent asymmetry in controlled filtering experiments: in both OPD and on-policy self distillation (OPSD), training only on incorrect SGOs outperforms training only on correct ones. Our further analysis suggests that models trained on correct-only SGOs tend to generate shorter reasoning traces and show weaker reflection behavior, while incorrect SGOs better preserve exploratory reasoning near the model's capability boundary. To exploit this signal without requiring full answer-containing rollouts, we introduce ReNIO, which Reweights Negative trajectory Importance for LLM On-policy distillation. By using the student-to-teacher probability ratio, ReNIO identifies pivotal tokens leading to wrong reasoning traces and aggregates their information into a normalized sample weight, inherently assigning larger weights to likely negative trajectories without observing the correctness of final-answer. Since Re-NIO only uses prefix-conditioned token probabilities, it preserves OPD's prefix training advantage over full-rollout reinforcement learning. Across both mathematical reasoning and code generation tasks, ReNIO improves both OPD and OPSD, with representative relative gains of up to 8.90% for Qwen3-1.7B and 10.00% for R1-Distill-Qwen-7B on mathematical reasoning benchmarks. Code repo: https://github.com/BDML-lab/ReNIO.