LLM-driven time series forecasting has become increasingly promising by leveraging contextual features to generate future values directly or refine forecast priors. However, incorporating contextual features alone does not ensure that models can identify which features matter for forecasting and understand what they suggest about future dynamics. We argue that the fundamental challenge is how to extract predictive evidence from contextual features and when and how to leverage it to guide forecasting decisions. In this paper, we propose CastReasoner, an evidence-grounded reasoning model for context-aware time series forecasting. Specifically, CastReasoner uses a structured chain of thought (CoT) to extract predictive evidence from contextual features, infer the future dynamics suggested by this evidence, and assess these dynamics against the forecast prior to guide adaptive forecasting decisions. To enable this capability, we design a reasoning-centric two-stage post-training framework including supervised fine-tuning (SFT) and reinforcement learning (RL). First, we synthesize reasoning trajectories through controllable multi-teacher guidance and select high-quality ones using a novel student-aware difficulty metric. Second, we design fine-grained multi-objective rewards to further improve reasoning and forecasting. To address heterogeneous learning needs across domains in both stages, we introduce a convergence-guided data balancing strategy. Through this two-stage training, CastReasoner improves both reasoning quality and predictive accuracy. Extensive experiments across multi-domain forecasting benchmarks demonstrate that CastReasoner achieves superior forecasting performance.
Cryptocurrency markets provide a natural testbed for autonomous AI agents because they are digital-native, high-frequency, and machine-speed environments, yet existing methods often struggle with non-stationary market regimes, opaque decisions, and fixed aggregation rules. In this paper, we propose Agent4Crypto, a dual-reflective and market-feedback-aligned multi-agent system. The framework coordinates specialized agents for data collection, analysis, risk assessment, reflection, and trading, synthesizing heterogeneous signals from prices, indicators, news, and charts in an interpretable decision pipeline. Our central design is a trader-centric decision mechanism: three style-specific trader agents, Aggressive, Neutral, and Conservative, generate candidate actions and natural-language rationales from the same market context, while their influence is learned rather than manually fixed. A PPO-based gating policy dynamically allocates trader weights through a dual-reflective feedback process. One reflection stream evaluates post-trade decision quality from realized portfolio outcomes and distills reusable lessons into a Financial ReasoningBank for retrieval in subsequent trader prompts, while the other estimates trader-level counterfactual contribution, indicating which style has recently aligned better with market movements. The PPO state combines Counterfactual Contribution Memory with market features, previous trader weights, and current trader actions, so trader influence adapts across bull, sideways, and bear regimes. Experiments on BTC, ETH, and XRP show that Agent4Crypto improves total return and Sharpe ratio over traditional strategies, machine learning and reinforcement learning baselines, and existing LLM-based trading systems, demonstrating the value of interpretable trader competition and market-feedback-driven adaptation in volatile cryptocurrency markets.
Time-to-event prediction is fundamentally challenged by partial observability and aleatoric uncertainty. In clinical medicine, for instance, the timing of disease progression or acute readmission is contingent on highly dynamic and largely latent health trajectories. Rather than forcing a deterministic point forecast, distributional regression offers a principled means to address this complexity by predicting a probability distribution over time. However, explaining these distributional predictions presents a key obstacle: raw distribution parameters output by the model are abstract mathematical quantities that do not align well with domain concepts. To gain interpretability, we propose a novel paradigm where output parameters are designed to be additive in log-odds---a widely established measure in clinical risk modeling. Our approach, called ODDS-DR, moves beyond technical convenience to deliver clear domain alignment as well as the layer-wise computational structure required for robust, time-dependent feature attribution. Validated across two real-world clinical cohorts, our framework matches state-of-the-art predictive performance while resolving distributional outputs into interpretable risk markers across flexible prognostic horizons.
Reliable tip-over prevention is critical for planetary rovers operating in unstructured environments where communication delays prevent real-time human intervention. Existing learning-based approaches typically optimize a single prediction objective, limiting their ability to capture both gradual instability evolution and abrupt terrain disturbances. This work presents a hierarchical multi-channel tip-over prediction framework based on isolated temporal learning. A multi-horizon Temporal Convolutional Network (TCN) learns progressive instability patterns from inertial measurements, while a Gated Recurrent Unit (GRU)-based forward dynamics model predicts nominal rover motion and detects unexpected dynamic deviations through prediction residuals. These complementary signals are integrated through decision-level fusion and priority-based arbitration to generate graded safety actions, including early warning and emergency braking. The proposed system is deployed as an Orocos Real-Time Toolkit (RTT)-based safety component within the Rock robotics framework, operating at 100 Hz on a planetary rover software stack. Experiments on Inertial Measurement Unit (IMU) samples collected across heterogeneous terrains demonstrate 97.58% event prediction coverage with a 1.82s median lead time, outperforming individual prediction channels. The results validate that isolated temporal models combined through hierarchical decision fusion provide a robust solution for real-time rover safety in uncertain terrain.
Foundation models increasingly combine geometry, semantics, and spatial context for three-dimensional scene understanding, but their suitability for forensic crime scene reconstruction remains unclear. This study presents a structured forensic evaluation framework for egocentric foundation models and demonstrates it using the Egocentric Foundation Model for 3D (EFM3D). Four simulated indoor crime scenes were recorded with Meta Project Aria Gen~1 at Robert Gordon University's East Lodge House forensic training facility. The framework evaluates room-scale reconstruction, semantic scene understanding, computational performance, and forensic suitability. Across the four scenes, EFM3D generated meshes averaging $12{,}668\pm2{,}430$ vertices and $24{,}838\pm4{,}903$ faces, with 39,529 recorded semantic predictions spanning 21--24 categories at a mean confidence of 0.40. Because reference meshes and expert-labelled 3D ground truth were unavailable, these statistics are treated as descriptive outputs rather than measures of geometric or detection accuracy. Final inference runs processed $1{,}388\pm203$ snippets per scene at $0.69\pm0.01$ samples/s and required $50.77\pm6.12$ minutes per recording. A seven-criterion capability profile indicated stronger support for room geometry, furniture localisation, scene context, and spatial relationships than for fine-scale forensic evidence. The results position EFM3D as a complementary tool for contextual documentation and scene review rather than a replacement for high-fidelity forensic capture. The framework provides a reproducible basis for evaluating future egocentric foundation models in specialised forensic settings.
Reinforcement learning (RL) from raw sensory inputs such as images or onboard sensors can be challenging due to high-dimensional observations and sparse rewards. A common strategy is to train a teacher policy with access to privileged state information and then distill it into a student policy that acts from raw inputs alone. However, when the teacher relies on information unavailable to the student, exact imitation may be impossible, creating an irreducible imitation gap. Existing approaches typically mitigate this mismatch through careful reward shaping or additional reinforcement learning on the student policy after the teacher has been trained. Instead, we introduce Representation-Aligned Imitation Learning (RAIL), which addresses the mismatch during teacher training itself. RAIL learns a latent representation shared across teacher and student observations using contrastive learning, and trains the teacher policy directly in this space. Because the teacher policy is constrained to operate on this shared space, it learns behaviors that are reproducible from student observations. This substantially reduces the imitation gap while preserving task performance. Across multiple environments, RAIL outperforms strong baselines without reward modification or post-hoc student fine-tuning, and surpasses direct reinforcement learning from raw observations. The learned representations further enable zero-shot transfer to new tasks through teacher-only training. Project site: https://merajmammadov.com/rail/.
We present a general method for discovering symbolic local learning rules that can replace backpropagation, applicable, in principle, to any deep learning architecture and task. The method automatically discovers local loss functions specific to each layer by (1) constructing a dictionary of monomials of local tensor variables, inspired by Penrose diagram notation, (2) regressing over coefficients weighting these monomials to build layerwise polynomial losses whose gradients together minimize a target global loss. We test the method on classification, reconstruction, and self-supervised learning tasks with small multilayer networks and minimal datasets. Across all tasks, the discovered rules induce non-trivial interactions between layers, which lead to significant reductions in the corresponding global losses. Furthermore, the method selects sparse combinations of monomials, finding local rules that are more interpretable than those generated by MLP-based alternatives. The potential implications of our method range from energy-efficient deep learning to the discovery of biologically plausible learning mechanisms.
Evaluating generative models requires measuring not only sample fidelity but also how well a generated set covers the real data distribution. Recent work has therefore introduced recall-oriented metrics for distribution-level evaluation. However, existing recall metrics often reduce distributional coverage to geometry-based coverage. This simplifying assumption weakens their diagnostic value and interpretability in two ways: these metrics provide only a single aggregate score that cannot distinguish essential diversity failures such as mode dropping and mode collapse, and geometry-based coverage can provide unreliable estimates by assigning high scores to poorly covered generative distributions. We therefore introduce CADMA, a capacity-aware framework for recall-oriented evaluation and diagnosis. CADMA evaluates distributional coverage through local capacity matching between real and generated samples, separating geometry-based coverage from effective coverage. This yields multi-dimensional coverage diagnostics that quantify mode dropping and mode collapse while reducing the overly optimistic estimates produced by geometry-based metrics. Through extensive experiments on toy data, real-world datasets, and generative models, we show that CADMA provides more reliable coverage estimates and more fine-grained, interpretable diagnostic signals of diversity failures than existing metrics. Our code is available at https://github.com/Open-Debin/CADMA_NeurIPS2026.
Continual Test-Time Adaptation (CTTA) methods update models online using a stream of unlabeled data to overcome distribution shifts encountered at test time. However, CTTA is often unstable, as it can reinforce biases in early predictions and increasingly predict only a few classes, resulting in drift or even collapse. In this paper, we identify narrowing pseudo-label class usage as an indicator of CTTA instability and quantify this behavior. Most existing methods have attempted to address this issue by either selecting reliable adaptation samples or re-weighting the loss to encourage diverse pseudo-labels, but they do not explicitly enforce a class marginal over a temporal window. Instead, we propose Sinkhorn Class-Marginal Regularization (SCALAR), a simple plug-and-play module that controls the pseudo-label distribution over a short temporal window. SCALAR works by maintaining recent predictions in a buffer, balancing their class distribution via the Sinkhorn algorithm to match a specified target distribution, and using the balanced soft targets for adaptation. Without changing the model architecture, SCALAR maintains the per-step runtime and increases the peak allocated GPU memory by only 10 MB on ImageNet-C. Our method is particularly effective on skewed, non-stationary data with severe distribution shifts, where SCALAR significantly alleviates drift and collapse when combined with existing stabilization methods based on entropy or diversity.
Mixture-of-Experts language models are cheap to run per token but expensive to hold in GPU memory, which drives the cost of serving them in production. One-shot expert compression shrinks them after a single calibration pass. Its two current state-of-the-art methods are REAP for expert pruning, which deletes the least salient experts, and REAM for expert merging, which merges them into the most salient ones. Both have been evaluated only on academic benchmarks. We compare the two on five multilingual production tasks of a web-hosting company, from classification to the generation of marketing HTML email templates, with two 30B-parameter models (Qwen3-30B-A3B-Instruct-2507 and the hybrid Mamba-attention Nemotron-3.5-Lightning) at 25% and 50% of experts removed and the same calibration data. Pruning wins: on the same rows, REAP is significantly better in 16 of 20 comparisons and REAM in 2. At 25%, REAP keeps 93.6% of Qwen3-30B's generative quality and REAM 62.5%, although on benchmarks the two are within a point for the same model. The damage concentrates in the output contract rather than the content: 89.8-95.2% of each significant loss on email templates comes from hard-gate failures such as invalid JSON, broken mobile layouts and runaway generation, and most failed non-English domain descriptions copy the prompt's English opening phrase. Cross-perplexity and logit divergence, two diagnostics from the REAP paper, also track production quality, as does the rate of repetition loops; n-gram diversity does not. We close with practical guidance for compressing MoE models for production.
On-policy self-distillation lets a language agent act as its own Teacher by conditioning on privileged context such as skills written from its own trajectories. Its guidance, however, is weighted by how much this context shifts token likelihoods rather than by whether it improves behavior, so errors in self-written skills return through later updates. We propose environment-calibrated on-policy self-distillation, which uses environmental outcomes to decide how much to trust the Teacher's current guidance and which self-written skills to reinforce. Command-level advantages, estimated by comparing commands executed at similar observations, scale each Teacher preference and reward reflections and skills whose induced preferences favor higher-advantage commands. Without skills at inference, 3B and 7B agents reach 90.5% and 93.3% macro success on seen ALFWorld tasks and 75.5% and 80.5% success on WebShop. On unseen ALFWorld tasks, they outperform GRPO under the same protocol by 11.0 and 7.6 percentage points, respectively. In an ablation study, supervising only executable commands rather than the Actor's full response raises 3B WebShop success from 52.3% to 75.0%.
A worked solution can make self-generated training targets shorter and cheaper, but accuracy on the practiced operation may hide how well a model can reuse newly learned rules. We study this choice in a 2B-parameter reasoning model learning two invented quantitative rules. The model generates targets from a rule sheet alone or from the sheet plus a program-written worked solution; targets are accepted for final-answer correctness, and the support is removed for fine-tuning and evaluation. Training questions request forward use in one segment; composition requires sequential reuse of the same rule across two segments. After six passes, using a program-checked numerical signature, sheet-only targets lead by 24.3 percentage points (95% item-bootstrap CI 14.6-34.0) on composition in a prospective test. On unseen argument values, observed forward accuracy is identical (87.5% each), but composition accuracy is 58.3% versus 23.6%, a 34.7-point gap (25.7-43.8). An independently sampled second target-bank pair repeats the direction (+14.6 points, 2.1-26.0). Ten passes on the shorter worked-solution targets, approximately matching the domain tokens of six sheet-only passes while changing other training quantities, narrow the gap to 16.3 points (6.3-26.7) in the first bank pair and 6.2 points (-6.3 to 18.8) in the second, where the pre-specified test does not hold. At six passes each, the worked-solution policy uses fewer capture tokens and less update time, yet the trained models use 2.6 times as many generated tokens per composition success under the numerical signature (2.1 times in the second bank pair). Choosing a capture context therefore requires testing how the new rules are reused, alongside the inference cost of those uses, rather than relying on forward accuracy alone.
Systematic literature reviews (SRs) remain slow and labor-intensive, and most automation tools address single stages of the workflow. We present PARIS (PRISMA Agentic Research Intelligence System), an open-source framework in which a conductor agent running in Claude Code orchestrates 15 specialized large language model (LLM) agents, from scoping a research question to exporting a manuscript draft. Agents are specified declaratively, coordinate exclusively through persisted artifacts that the conductor validates before every transition, and delegate retrieval, bibliometric analysis and scoring to deterministic Python utilities. Three human checkpoints gate study inclusion, corpus adequacy and the synthesis outline. We also introduce QualityScore, a pre-synthesis index that aggregates normalized network, bibliometric, semantic and process sub-scores into a 0-100 grade. Two case studies illustrate both: AI chatbots and university students' learning (388 records, 100 included) and AI-assisted SR methods (131 records, 123 included, QualityScore 69.1 recomputed with a proxy network sub-score). They show end-to-end feasibility across domains and the diagnostic use of QualityScore: in the first run, corpus improvements chosen at the corpus-adequacy checkpoint raised the score from 74.6 to 81.9 and the number of pre-2023 studies from 1 to 19. Audits of both runs exposed failure modes: incomplete retrieval, counts that drifted between agents, unrecorded or delegated checkpoint decisions and a grade that depends on how citation data are handled. PARIS fully supports 8 of the 27 PRISMA 2020 items and partially supports 11. Output quality has so far been judged only by an internal reviewer agent; independent validation against human reviews is the next step.
Quantum kernel methods have been proposed for network intrusion detection, yet their performance limitations are often attributed to noise in current quantum hardware. This study evaluates that assumption through a controlled analysis of Pegasos-QSVC on the NSL-KDD and ToN-IoT datasets. We perform 56 quantum experiments using one exact noiseless statevector backend and three noise-simulated superconducting backends while keeping preprocessing, PCA features, data splits, and classical baselines identical. XGBoost achieves accuracies of 98.94% on NSL-KDD and 99.41% on ToN-IoT, compared with 91.80% and 82.98% for Pegasos-QSVC. The quantum model also exhibits substantially greater seed-dependent variability. Eight runs collapse into degenerate classifiers, including four recall-collapse cases and four trivial-positive classifiers. Notably, two failures occur on the noiseless backend, demonstrating that hardware noise is not a necessary condition for classi-fier collapse. Kernel Gram matrix analysis further reveals low effective rank and severe ill-conditioning even in the absence of noise, indicating intrinsic fragility in the learned quantum representation. These results show that average accuracy alone can conceal operationally unacceptable behavior and highlight the importance of reporting worst-case performance, seed-level variability, class-specific metrics, and explicit degeneracy tests in quantum intrusion detection studies.
Transformers are built with continuous symmetries in the attention mechanism. We derive $\rho$, a scalar invariant that measures each head's capacity to couple to rotary positional encodings. This framework allows us to analyze heads without forward passes, opening a new research direction into data-independent analysis of modern LLMs. Empirically, we find that $\rho$ predicts induction heads, the most-studied circuit type in mechanistic interpretability, with Spearman correlation between $-0.55$ and $-0.74$ across four bias-free RoPE models spanning $135$M to $7$B parameters. As a preliminary proof of concept, we use $\rho$ to apply YaRN to only the top-50% positional heads, which degrades perplexity less than uniform application. Code is available at https://github.com/MPedraBento/rho-invariant.
In patrol missions involving multiple unmanned surface vehicles (USVs), a designated command USV can support fleet management by sequentially rendezvousing with each patrolling USV. These operations require motion-feasible route planning and priority-aware rendezvous sequencing, with the objective of minimizing mission completion time and total tardiness relative to priority-dependent due dates. We propose a framework combining grid-based path planning, surrogate travel-time prediction, and graph-based reinforcement learning. The path planner accounts for ocean currents and approximates turn-rate constraints through multi-stage maneuvers, with path feasibility empirically validated through trajectory-tracking simulations. For rendezvous sequencing, a Feasible Transition Graph (FTG) represents candidate rendezvous events and temporally feasible transitions, enabling a graph neural network to exploit spatiotemporal rendezvous relationships for policy learning. A surrogate predicts travel times from compact current and motion features to avoid repeated path searches during FTG construction and policy training. Evaluation using grid-based path planning shows statistically significant average objective reductions of 10.5%, 3.5%, and 1.5% compared with greedy, MLP-RL, and 2-step lookahead greedy, respectively. The policy achieves these improvements while requiring approximately 1/35 of the latter's computation time. It maintains its advantage over MLP-RL under unseen priority-tier settings, with larger gains as fleet size increases.
A sparse terminal reward tells an agent little about its steps. Much of the missing information is implicit in the environment's dynamics, because each state transition can change the conditions for success long before the reward arrives. A world model trained on real transitions can learn to predict what each action changes, but a predicted observation does not by itself show whether success has become more likely. With a binary terminal reward, PPO's critic makes this judgment by estimating the current policy's probability of success, but only for states the agent reaches. Our method, WM-PPO, elegantly brings the two together. For each action the policy samples, executed or not, the world model predicts the next observation and the critic values the predicted successor. In the policy channel, comparisons among these values give the policy auxiliary advantages; in the value channel, their mean gives the critic an auxiliary target. A shared gate weights both channels according to how reliably comparisons with similar margins predicted real outcomes earlier in training. The policy, critic and world model all learn from the agent's own interaction, with no teacher or privileged information, and only the policy is used at inference. With a comparable number of real environment steps, WM-PPO raises success from PPO's 63.5% to 87.5% on ALFWorld and from 39.3% to 52.9% on ScienceWorld. It matches the strongest PPO refinements we tested on ALFWorld and exceeds the best of them by 6.5 points on ScienceWorld. The trained policy more often completes prerequisite steps before placing an object or committing to an answer.
Verbal memory retrieval carries rich diagnostic information for Alzheimer's disease through the order, hesitation, and categorical organization of recalled words, yet computational modeling of this retrieval process remains largely unexplored. A key challenge is that timing and hierarchical structure are diagnostically entangled: the diagnostic meaning of retrieval timing depends on its hierarchical context. We propose TrajLift, a spectral framework that models verbal recall as timed trajectories on a hierarchical graph. Our central mechanism is latency-modulated heat diffusion on the semantic hierarchy, a unified operator that lifts each recalled word into a representation of both multi-scale hierarchical structure and continuous retrieval timing. Formal analysis shows that the resulting representation is first-order separable into structural and temporal components with spectral selectivity across hierarchical levels. On synthetic benchmarks requiring structure-time interaction, TrajLift stays near-perfect on deep hierarchies where all baselines fail. On real-world clinical data, we show that TrajLift outperforms diverse baselines by jointly modeling timing and semantic structure, with gains that grow as retrieval timing becomes more informative.
Computational advances in natural language processing and machine learning have substantially expanded the use of automated approaches in verbal deception research. However, existing verbal deception datasets remain fragmented across disciplines, as they are often difficult to identify, access, and reuse, limiting cross-study comparisons and the development of robust computational models. We present the Deception Archive, a curated and standardized repository of publicly available verbal deception datasets built upon a large-scale systematic review. Eligible datasets were retrieved, manually inspected, and standardized following a structured data curation workflow. The current repository comprises 41 publicly available datasets containing 387,995 truthful or deceptive statements. The archive represents a heterogeneous repository (e.g., in data size, topic of investigation, ground-truth, etc.) and a living resource for integrating newly available deception datasets over time. By providing centralized access to standardized datasets and accompanying metadata, the Deception Archive facilitates dataset discovery, cross-dataset validation, and reproducible evaluation of computational models.
Federated learning (FL) is a promising research paradigm that enables collaborative model training across distributed clients without data sharing. To handle unlabeled data in local clients, existing unsupervised FL research has primarily focused on tasks like clustering and self-supervised representation learning, leaving the critical problem of unsupervised visual recognition largely unexplored. To bridge this gap, we introduce FedUR, the first framework designed for unsupervised recognition by leveraging the zero-shot capabilities of pre-trained vision-language models like CLIP. Specifically, FedUR addresses data heterogeneity through a pseudo-labeling mechanism with a class-aware data selection strategy. This strategy first estimates a global pseudo-label distribution, then allocates pseudo-labeled data to clients accordingly. To handle the uncertainty of conventional positive pseudo-labels, FedUR further incorporates class weights and a negative pseudo-labeling strategy. Furthermore, FedUR proposes a personalized prompt aggregation protocol where general visual prompts are aggregated on the server, and personalized text prompts encoding local knowledge are retained on clients. Extensive experiments demonstrate that FedUR consistently outperforms baseline methods under both label-level and feature-level distribution shifts. We believe this work paves the way for further advances in federated unsupervised recognition. Code will be available at {github.com/tim-learn/FedUR/}}.