Science & Mathematics

Decision models, such as the commercial Jev, are language models that return one probability for each option of a question instead of generating text. We present Decision4Rec, an open and modular framework that makes decision models useful for recommendation. A knowledge graph writes the logs of each recommendation scenario as short statements on a defined scale, such as "sold: top 10%". A pool of agents, proposes candidates, and a residual ranker keeps the best 30. A decision model, trained once on real choices of three scenarios, reads them through one interface of typed questions, and a fusion step combines its answers with the score of the ranker. Any model with this interface can be used, and we run the commercial Jev, the released open Kev and Kev models trained by us on recommendation requests. The framework with a trained 4B decision model is 25 to 36% above the best trained baseline in NDCG@10 in three sparse Amazon catalogues, one of them never used to train the decision model, and almost tie with baselines on the dense Steam dataset. Most of the gain comes from the pool, and with a trained recommender it comes from what the agents know about its candidates rather than from new candidates. The 27B decision model adds to the pool 5 to 6% above the 0.8B model in three of four domains. A trained 0.8B open model beats the released Jev in eleven of fifteen settings and is never worse. On MovieLens-1M, a dense domain never seen in training, the gain is 5 to 7%. Code, configurations and trained models are available at https://github.com/marcopoli/decision4rec and https://huggingface.co/m-polignano.

Financial agents are expected to answer factual investment queries through open search, but financial multi-hop retrieval differs from general multi-hop QA in a crucial way: each intermediate answer is an exact typed constraint, such as an entity, date, event, or numerical indicator, that determines the next query. This strict answer-conditioned dependency makes small intermediate errors cascade into plausible but wrong search trajectories. We introduce FinHopper, a benchmark for complex financial agents with controlled reasoning depth from 1 to 6 hops, covering stocks, funds, and macroeconomic facts, with gold intermediate answers for process-level evaluation. To construct FinHopper at scale, we design a state-transition-based synthesis framework that composes verified atomic financial QA patterns into dependency-preserving chains, followed by targeted in-synthesis verification, model-ensemble checking, and public-search validation by financial experts. Experiments with 20 LLM-agent configurations show that public search tools are necessary but insufficient: performance drops sharply as hop depth increases, the best model remains far below human experts, and process-aware scoring reveals partial reasoning progress hidden by final-answer-only evaluation. The benchmark is available at https://github.com/justajustin/FinHopper/.

Diffusion language models enable parallel token generation, offering a promising alternative to sequential autoregressive decoding. Existing acceleration methods use predictive confidence to enable parallel token commitment while preserving generation quality. However, confidence-based selection prioritizes reliable token predictions without explicitly considering how token positions shape the remaining decoding schedule. Moreover, such ordering does not always reflect inter-token dependencies, suggesting that confidence need not be the sole criterion for token selection. Motivated by this limitation, we propose Balance-Aware Pivot Decoding (BPD), which favors central pivots within contiguous masked intervals and decodes the selected positions in parallel. Balanced pivots yield logarithmic decoding depth in the interval length. Our practical method accounts for model uncertainty by combining confidence-based prefilling with confidence-thresholded center selection. We evaluate the resulting quality-efficiency tradeoff on LLaDA and Dream across five mathematical reasoning, code generation, and instruction-following benchmarks. Our implementation is available at: https://anonymous.4open.science/r/DLM_random_pivot-EE62/

Dense temporal audio-visual captioning is important for multimodal video understanding and generation. However, prior work is constrained by three main limitations: (1) the lack of a dataset combining high temporal resolution, longer videos, and bidirectional alignment between time and events; (2) the reliance of most existing RL methods on checklists; and (3) the lack of a benchmark that bidirectionally evaluates event semantic correctness and temporal accuracy. To address these challenges, we propose: (1) AVTime-50K, a high-quality 50K-scale dataset with temporally precise, semantically accurate captions and longer videos; (2) AVTime, a state-of-the-art model optimized with a novel Bidirectional Temporal-Semantic Alignment (BiTSA) reward that imposes explicit bidirectional constraints between time and events; and (3) AVTime-Bench, a specialized benchmark that introduces a novel bidirectional evaluation of temporal accuracy and semantic correctness. Our code, models, and datasets are available at https://huggingface.co/collections/Apryle/avtime.

Deep learning for radio frequency (RF) signal analysis relies on large annotated datasets, yet synthetic datasets trade physical realism for controllability, while uncontrolled over-theair measurements offer realism but limited reproducibility. We present CorteXforge, a scenario-based framework for generating realistic and reproducible RF datasets on programmable software defined radio (SDR) testbeds. High-level RF scenarios are automatically translated into synchronized experiments executed on the SLICES-RI/CorteXlab platform, producing fully annotated Signal Metadata Format (SIGMF) datasets that capture hardware impairments and propagation effects. We validate CorteXforge through automatic modulation classification using ModForge, an over-the-air (OTA) dataset generated with the framework. Cross-domain experiments reveal a synthetic-to-real performance gap and show that synthetic pretraining remains beneficial for adaptation to OTA data. CorteXforge provides a practical foundation for reproducible RF dataset generation and future research on domain generalization.

Maximum rank-distance codes were introduced independently by Delsarte (1978) and Gabidulin (1985) for unrestricted matrix spaces, and their study was subsequently extended to the classical form spaces of alternating, symmetric, and Hermitian matrices. For Hermitian $d$-codes, Schmidt (2018) established the upper bound and constructed optimal additive codes for all admissible parameters except when $n$ and $d$ are both even and $4\le d\le n-2$. We settle this final open existence case, more generally, construct additive maximum Hermitian rank-metric codes for every admissible length and minimum distance. Rather than working directly in the space of Hermitian matrices, we construct our objects within central simple algebras with unitary involution arising from quotients of skew polynomial rings. We also introduce scalar-field and Lyapunov operators in these algebras, leading to new invariants for the equivalence of additive Hermitian codes. Explicitly determining these invariants yields strong necessary conditions for Hermitian equivalence and demonstrates that our construction contains a large number of pairwise inequivalent codes.

Deep generative models provide an effective modern solution to alleviating the problem of data scarcity by synthesizing realistic time series. However, despite the considerable progress achieved by existing deep time series generators, two key challenges remain: (1) learning hierarchical representations for heterogeneous temporal patterns; and (2) coordinating the generation process of heterogeneous temporal dynamics. To address these key bottlenecks, in this work, we propose ORBIT, a unified deep generative framework for time series. Specifically, we develop two key components: the Frequency-Hierarchical VQ-VAE (FH-VQ-VAE) and the Frequency-Band Autoregressive Transformer (FBAR). The former learns hierarchical representations by decomposing latent temporal representations into different frequency bands and applying residual quantization with increasing token budgets. The latter establishes a novel low-to-high frequency autoregressive direction, yielding a time series generation process that first constructs global temporal structures and then incrementally incorporates fine-grained temporal details. These compact designs not only disentangle heterogeneous temporal patterns for more effective representation learning but also mitigate potential interference among different temporal dynamics during generation. Extensive experiments on multiple challenging real-world benchmarks demonstrate that ORBIT achieves state-of-the-art overall performance while exhibiting superior inference efficiency compared with advanced baselines.

The security guarantees of existing formally verified compilers for compartmentalized programs in unsafe languages like C stop at a rather high level, assuming a built-in protection mechanism and a block-based memory model. Verifying a concrete, low-level enforcement mechanism for compartment isolation is still a major challenge for current proof techniques. In particular, Abate et al.'s proof technique fails because one of their proof steps, recomposition, essentially requires hiding all information about the code of each compartment, which is not practically achievable at a low level. We propose an extension of their proof technique to a low-level setting where compartments can fully observe each other's code. To do so, we introduce an overlay semantics that allows us to characterize these observations as a form of undefined behaviour, and we use a novel variant of recomposition that we call blame-aware recomposition, which essentially allows recomposing executions as needed in the original high-level proof technique, even if the low-level context observes the code, because this context is ultimately going to be discarded. We apply this new proof to a compartment isolation mechanism that uses a programmable tag-based reference monitor.

Wind farms can provide flexibility by curtailing their output during periods of excess generation in the power grid. This can be done through implicit balancing, where the power setpoints of the farm are adapted to anticipate the imbalance price. By doing so, the wind farm helps reduce the resulting system imbalance while also lowering its own costs. However, implicit balancing is challenging since the imbalance prices are unknown in advance and hard to predict. Furthermore, the control strategy of the wind farm can affect the imbalance price formation. To address both challenges, we develop a control policy with 1-minute setpoints using distributional soft actor-critic. Evaluated on real data of the Belgian grid and an offshore wind farm, we show that our method outperforms a price threshold controller, often used in industry, by 88% and performs on par with soft actor-critic. Furthermore, unlike soft actor-critic, our approach can learn a risk-averse policy by optimizing a risk measure, such as conditional value at risk. The risk-averse policy was found to better account for its own impact on the price, showing a reduction in overshoots, while achieving a similar degree of imbalance cost savings. Last, we find that using 1-minute instead of 15-minute setpoints further reduces imbalance costs for all evaluated control policies.

Programmable logic controllers (PLCs) automate industrial processes through repeated scans, each reading inputs, executing control logic, and updating outputs. Engineers write this logic in Structured Text (ST), a high-level textual language defined in IEC 61131-3. Large language models (LLMs) can now generate ST from natural-language requirements, yet compilation success and checks on selected properties or sampled scans can leave behavioral errors undetected. We introduce CycleBench, a benchmark of 301 ST tasks with behavioral requirements, public interfaces, reference programs, and executable scenarios. Its runtime executes each generated program and its reference under aligned inputs, initialization, and logical time, comparing their outputs after every scan. Across eight LLMs and baseline generation methods, execution pass rate trails compilation pass rate by 11.92 percentage points; observed failures include output-update, arithmetic, and timing errors. On fixed traces of programs directly generated by five LLMs, checking outputs only at every fifth scan misses 21.66% of the 157 failures detected by per-scan comparison; across sampling phases, the missed-failure rate ranges from 12.10% to 28.03%. These misses stem from transient output discrepancies, whose detection depends on the sampling phase. Building on this evaluation, we propose CycleGen, which generates multiple candidate ST programs, repairs their compilation errors with bounded compiler feedback, and selects one using runtime self-tests generated in a separate model context. CycleGen improves the execution pass rate over direct generation by 8.68 percentage points. Component ablations distinguish repair and self-test selection; candidate-count analysis shows diminishing returns, with no execution-pass-rate improvement beyond five candidates in our experiments despite 38.8% higher token consumption at eight.

Drug--target affinity (DTA) prediction plays an important role in computer-aided drug discovery, where reliable computational models can help prioritize compound--protein pairs before experimental validation. A key challenge in DTA prediction is to effectively mine multi-level molecular information from drug structures while accurately modeling the binding-site environment of target proteins. To address these issues, we introduce HierDTA, a hierarchical structure-aware framework for DTA prediction. \textcolor{black}{HierDTA represents each drug as an atom--motif heterogeneous graph and introduces an augmentation module that propagates functional-motif semantics to the corresponding atoms. For proteins, kinase-domain structures are converted into residue interaction graphs. The resulting atom and residue representations are integrated using cross-attention to capture pair-specific relationships, while molecule fingerprints and pretrained ESM-2 embeddings provide complementary global information. Compared with state-of-the-art methods, HierDTA achieves superior performance on the Davis dataset across all metrics and attains the highest \(r_m^2\) of 0.791 on KIBA. It also exhibits strong generalization under cold-start settings, particularly on KIBA, where MSE is reduced by 21.7\% in the unseen-protein scenario. Ablation study and interpretability analysis supported the contributions of the hierarchical augmentation and interaction components.} Case studies further showed that HierDTA provided early enrichment of candidate bioactive compounds. These results indicate that HierDTA is an effective framework for affinity prediction and target-specific virtual screening.

It is commonly believed that a bulk photovoltaic current requires broken spatial inversion symmetry. However, in this work, we demonstrate that dc photocurrents can arise in centrosymmetric solids from the bilinear coupling between the electric and magnetic components of light. This magnetoelectric bulk photovoltaic effect (MBPE) has a quantum-geometric origin in the dipole moments of the Zeeman quantum metric and Zeeman Berry curvature. We identify four distinct MBPE mechanisms, including time-reversal-even and time-reversal-odd injection-like and shift-like photocurrents driven by linearly or circularly polarized light. Our symmetry analysis shows that the MBPE is allowed in broad classes of two-dimensional centrosymmetric nonmagnetic and magnetic materials under normally incident light. As representative examples, we predict sizable MBPEs in centrosymmetric monolayer Pt$_2$HgSe$_3$ and altermagnetic RuF$_4$, with current densities on the order of $1~\mathrm{mA/m}$. Notably, the injection-like photocurrent in RuF$_4$ reverses sign upon switching either the light helicity or the Néel vector. Our results extend the bulk photovoltaic effect to centrosymmetric solids and open new opportunities for electrical readout of altermagnetic order.

The Chinese idiom "kou mi fu jian" ("honeyed words, hidden daggers") names speech that is polite in form but hostile in function. We present ZhenhuanBench, a 1,592-item benchmark built from expert commentary on Empresses in the Palace, covering speaker intent, emotion-language mismatch, and per-listener uptake. Text-only evaluation covers 34 LLMs. GPT-6 Astra leads the leaderboard at 73.48% overall, while its listener-reaction score is 58.37%. The annotations also supply supervised traces and string-verifiable training rewards. With 338 gold traces, an 8B model outperforms teacher distillation on identical items. Subsequent RL saturates, and chance correction reduces the estimated recoverable performance gap by a factor of 7.5. Our CPR-GRPO loop (Compress-Probe-Replenish) uses this estimate to weight gold supervision, reaching 67.0 on public multiple-choice tasks and 74.5 under the training-matched hidden protocol. Without target-task training, CPR outperforms Base, SFT, and GRPO across six Chinese and English pragmatic and sarcasm evaluations. On three external datasets, training with CPR yields higher mean Macro-F1 than SFT and GRPO.

External skills provide language-model agents with reusable procedural knowledge, yet how models use such knowledge after it is supplied remains poorly understood. We identify Procedural Exposure Bias (PEB), the tendency of language models to increasingly favor an action as its associated procedural pattern is repeatedly exposed in a skill, even when the interaction state and action relevance remain unchanged. We find that PEB strongly depends on how procedural knowledge is represented: presenting the same content visually substantially attenuates exposure-induced shifts in action preference relative to textual presentation. Further analysis shows that this attenuation primarily arises from visual encoding rather than structured layout. Although repeated procedural signals remain recoverable from Visual representations, they exert a weaker influence on downstream action prediction, distinguishing signal accessibility from its use as action evidence. This effect also extends to executable action selection, where increasing exposure more strongly favors the corresponding action under Text than Visual. Layer-wise analyses show that exposure-dependent action evidence becomes more pronounced toward the model output, with stronger evidence under Text, while activation interventions reveal architecture-specific roles of internal components in transmitting or counteracting the preference shift. Complete-agent evaluations across ALFWorld, ScienceWorld, and WebShop further show that Visual skills achieve higher macro-average task success than Text skills for all four evaluated models. These findings establish skill representation as an important factor governing how procedural knowledge is translated into agent decisions. All code is available at https://github.com/NEUIR/Vis-Skill.

Generating high-quality synthetic datasets at scale remains a fundamental challenge for training specialized AI systems in data-scarce domains. We present a comprehensive methodology that produces 14.3 million quality-filtered synthetic training examples from a structured taxonomy of 632,000 scenario specifications. Our four-stage pipeline ad- dresses critical challenges in synthetic data quality, diversity, and realism through: (1) systematic pre-generation research establishing optimal generation parameters via diminishing returns analysis, (2) conditional auto evolution mechanisms that remediate low-diversity batches during generation, (3) a two-layer semantic deduplication architecture combining lexical and embedding-based methods, and (4) LLM-as-Judge quality evaluation filtering samples based on behavioral correctness and tuple fidelity. Empirical analysis across three domains (Banking, Insurance, Financial Services) demonstrates domain-specific deduplication patterns ranging from 9.9% to 40.6%, with output generation modes exhibiting highest redundancy. Our batch size optimization reveals mode-specific diversity saturation thresholds, enabling principled allocation strategies. The complete pipeline reduces 26.4M raw samples to 14.3M high-confidence examples through merge (20.9% reduction), semantic deduplication (18.7% reduction), and quality-based filtering (15.9% reduction), providing a reproducible framework for domain-specific synthetic data generation with rigorous quality control.