Multiple-criteria decision aiding (MCDA) proposes models and methods for different types of decision problems such as choice, ranking or sorting. Such models involve various parameters that must be elicited. In the elicitation process, all necessary information has to be taken from the decision-maker. For this, several direct and indirect elicitation methods are suggested in the literature. In this work, we concentrate on direct elicitation: we aim at providing a rigorous elicitation procedure for the parameters of the Non-Compensatory Sorting model (NCS), which has been characterized by an axiomatic analysis. A basic version of the NCS model is considered, i.e., a model with two categories and no vetoes. The parameters are the set of sufficient coalitions of criteria and the profile separating the two categories. Regarding the sufficient coalitions, we propose several questioning algorithms that engraft questions on classical graph exploration algorithms such as depth-first search. Such algorithms are implemented and evaluated by simulation. For eliciting the limit profile, we design a questioning algorithm using dichotomy. Questioning algorithms can be optimal in the worst case or on average. An optimal algorithm in the worst case has been proposed by Hansel in the framework of monotone Boolean functions. We implement and evaluate this algorithm as a variant of the depth-first search approaches. Regarding optimal questioning strategies on average, we evaluate decision trees for up to four criteria. On the other hand, we assess how more general is NCS as compared to MR-Sort, a particular case of the NCS model in which sufficient coalitions can be represented by additive weights.
Berglund and van der Merwe formalized byte-pair encoding and bounded how far an appended string can change a tokenization, leaving the converse open: how far can a string prepended on the left change it, which dictionaries make that ripple infinite, and can it be computed efficiently when finite? We answer all three for every dictionary. The origin radius, how far a prepend can change covering spans, is decided with radius certificates by a product of two runs of the automaton that guesses the final tokens, in time polynomial in the dictionary written out, and it is infinite exactly when one periodic word has two canonical tilings with no common boundary. Under their ordering condition it is infinite exactly when some rule (t, t) finds its two copies adjacent at its own stage, and such a rule makes any dictionary infinite. A finite radius makes every window of a width read off the radii have a cut, and the radius is finite exactly when such cuts are dense. The token machine completes for every dictionary, and the tokenization is subsequential, extending what Berglund, Martens and van der Merwe proved for proper ones under one pass per rule. A streaming tokenizer emits after every symbol exactly the tokens that no continuation changes, in polynomial time, answering a question of Dalirrooyfard and Ye. Boundedness of the label measure, the edit distance of the token sequences, is decided for every dictionary in polynomial time as well, and under the ordering condition the largest distance one edit causes over dictionaries of m rules lies asymptotically between m^2/10 and m^2/2. GPT-2's merge table carries 244 effective self-pairs, so its origin radius is infinite from the forty-second merge on, and that of cl100k, o200k and Llama 3 from their first. Every measured number is its program's output, and what is not executed is named.
Languages of India need 1.6 to 14 times as many tokens as English on GPT-4o's tokenizer for the same content, which makes them more expensive and slower to serve. We train tokenizers from scratch on human-written text in about 200 languages. Havon v3 is a byte-level BPE tokenizer with 262,144 tokens and a pre-tokenizer that keeps combining marks with their letters and chunks scripts written without spaces. On FLORES-200 devtest it needs fewer tokens than GPT-4o's o200k_base on all 25 languages of India (median 0.74 of GPT-4o's count) and on 172 of 204 languages, but 7% more on English. Havon v4 adds a second, multi-word training stage (SuperBPE) to v3. At exactly GPT-4o's vocabulary size of 200,019 tokens, v4 needs 6.1% fewer tokens than GPT-4o on English and fewer on 161 of 204 languages and all 25 languages of India (median 0.74); at the same size, v3 needs 9.4% more on English and wins on 133. With 262,144 tokens, v4 needs 9.6% fewer tokens on English and fewer on 182 of 204 languages, and with 524,288 tokens it needs fewer on 203 of 204. The gains are largest for English, Hindi and the Romance languages; scripts without spaces, such as Khmer, Burmese and Tibetan, gain little and lose subword merges. Havon v5 replaces v4's Python-only code sample with permissively licensed code in 15 languages: on code it needs 1.03 to 1.22 times GPT-4o's tokens, against 1.11 to 1.29 for v4, while keeping all 25 languages of India (median 0.74, all significant under a sentence-level bootstrap) and English (0.94). These are compression results only: we have not yet evaluated a model trained with v4 or v5.
Commercial language model APIs bill per token, so a tokenizer that needs more tokens for the same content in one language makes that language more expensive, slower, and able to fit less text in a fixed context window. We measure this disparity for nine public tokenizers, including those of GPT-4, GPT-4o, Qwen2.5, DeepSeek-V3, Mistral-Nemo and four Sarvam models, on all 25 languages of India available in FLORES-200 (2,009 professionally translated sentences aligned with English) and on 13,732 human-translated Hinglish-English social media pairs from PHINC. On GPT-4o's tokenizer the same content needs between 1.57 times (Hindi) and 5.01 times (Odia) the tokens of English, and 13.7 times for Santali in Ol Chiki script; GPT-4's tokenizer needs up to 12.8 times. Sarvam-30B/105B is the only third-party tokenizer that stays below 2.6 times on every language (median 1.53). In an illustrative scenario using official list prices accessed on 5 November 2026 and stated assumptions, a support bot handling 100,000 chats a month on GPT-4o would cost about INR 56,640 in English and INR 102,829 in Marathi; these are estimates, not measured production costs. Against an openly licensed Hindi and Marathi morphological segmentation set, no tokenizer places more than 35% of its word-internal boundaries on morpheme boundaries (boundary F1 at most 0.48). In a companion paper we present a tokenizer that reduces this tax. Code, pinned artifacts and raw token counts are available at https://github.com/HavonAI/indic-token-tax.
Indoor localization commonly combines WiFi received signal strength indicator (RSSI) fingerprints with inertial measurements. However, these sensor streams are sampled asynchronously at different rates; therefore, observations rarely align in time and may be missing or stale. Most existing methods resample the data onto a shared clock and process each modality using a separate branch. We instead present a continuous-time transformer that represents observations as an unordered set of timestamped tokens. Each token encodes its measurement, modality, and elapsed time since acquisition. A single attention block jointly fuses information across modalities and time, avoiding both resampling and modality-specific branches while remaining invariant to token order. During training, we randomly mask entire modalities and individual observations, enabling the model to operate under missing or outdated inputs. We validate the modality-specific encoders on their respective benchmarks and evaluate the fused model on both simulated and real datasets, including a crosssession dataset collected on unseen days. The proposed approach improves upon an open WiFi fingerprinting baseline under cross-session drift, remains competitive with learned fusion and dead-reckoning baselines, and degrades gracefully as observations become missing or stale. These results suggest that a single continuoustime transformer provides a practical framework for indoor localization from asynchronous sensors.
A document preserves the visible result of professional work but often discards the research, calculations, assumptions, evidence, and authorization boundaries behind it. We study a different unit of delivery: a Portable Agent Format (PAF) package that carries a work product together with compiled context, evidence, provenance, and rules for subsequent interaction. To test whether this context remains useful beyond the process that produced it, we evaluate four new packages, created after the hosted and offline readers were fixed: an investment proposal, fund research, financial-statement analysis, and an engineering incident review. Two packages were produced by specialized agents and two through a general-purpose assistant's packaging skill. Across 400 frozen questions and 12,000 evaluated responses, a hosted governed reader succeeds on 96% of questions versus 86% with instructions plus raw-file retrieval (Holm-adjusted p < 10^-7). On the 260 answerable questions per condition, success is 94% versus 79%; the gap varies by package. Source-material chat and instructions alone reach 48% and 74% overall. A separate 0.8-billion-parameter offline reader succeeds on 69% overall, but 54% on answerable questions. Its -1 response rate is 2.9%, combining materially wrong and unauthorized answers, versus 56% with the source and instructions alone. Ablations suggest advantages for the integrated selector and package-adapted writer, without isolating representation from training effects. The same run exposes a failure: offline retrieval resolves none of the 40 historical-record questions in one financial-data package. These findings support treating compiled knowledge as part of the delivered work product, while leaving independent standards interoperability and end-to-end governance conformance unproven.
Output-level fairness does not constrain whether demographic groups rely on similar predictive contributions in representation space. For a nonlinear encoder with a linear head, we show that outcome-conditioned output score disparity equals the signed sum of feature attribution differences across subgroups by allowing small prediction gaps to arise through cancellation effect. We test whether an attribution disparity penalty complements output-level equalized odds based fairness surrogates. Across 960 DINOv2-LoRA models on HAM10000 and Harvard-GF datasets for gender and race fairness, attribution augmentation increases utility-fairness trade-off performance for hypervolume in six different settings with paired 95% confidence intervals and relative gains of 1.49-10.57%. Under a predefined utility constraint, augmented model families also achieve 9.0-38.7% lower equalized odds difference, lower attribution disparity, and reduced cancellation ratio while preserving predictive utility. These results support attribution fairness as a complementary explanation-aware constraint on output fairness.
Although remote sensing image arbitrary-scale super-resolution (RSI-ASSR) methods support reconstruction at arbitrary or continuous scales, they predominantly condition on the relative scale factor and have not sufficiently investigated the absolute spatial resolution itself as a reconstruction prior, thereby disregarding the inherent spatial context of objects, which frequently leads to spatial-related distortions and compromised performance in downstream remote sensing applications. To address these challenges, we introduce SpatialSR, a SRe-centric arbitrary-scale RSISR method that leverages the spatial resolution (SRe) of remote sensing images (RSI) as an essential prior for super-resolution reconstruction. Distinct from existing SR methods, SpatialSR aims to restore high-resolution remote sensing images with a target SRe (typically set to 0.1m/pixel to align with the metadata of commonly used downstream task datasets). Specifically, we design a Spatial Resolution Estimation Module (SREM) to estimate the SRe of low-resolution (LR) images, thereby dynamically determining the required arbitrary scaling factor. Next, we employ a pretrained SRe-to-image diffusion model to extract SRe-specified texture-style codes, which are further incorporated into a carefully designed Resolution-aware Style Consistency Module (RSCM) to constrain the SRe-aware texture consistency of reconstructed images. Furthermore, to overcome the high computational expense inherent in diffusion-based SR methods, we wrap the iterative diffusion process as a Trajectory Consistency (TC) loss, which enables comparable reconstructed texture quality of the generative model in SpatialSR within a single step. Extensive experiments demonstrate that the proposed SpatialSR achieves state-of-the-art (SOTA) performance on commonly used RSISR and downstream (image classification and segmentation tasks) datasets.
Precise multimodal retinal image rigid registration (MRIRG) is essential for aligning cross-modal vessel structures, offering critical guidance for ophthalmic surgery. Despite its significance, the advancement of MRIRG is hindered by the lack of publicly available multimodal retinal image rigid registration datasets spanning various ophthalmic surgery phases (e.g., preoperative, intraoperative, and postoperative). To address this issue, we introduce the MICRO dataset, a large-scale MRIRG benchmark, comprising 1,365 cross-modal retinal image pairs with standardized annotations across all ophthalmic surgery phases. Moreover, existing deep registration methods achieve competitive MRIRG results via a two-step segmentation-then-detection framework, but segmentation quality heavily constrains their performance. We argue that this dependency stems from an implicit need for keypoint topology constraints, which have not been exploited for MRIRG through a one-step registration framework. Inspired by the above analysis, we propose a novel One-step Frequency-Structure Similarity Registration (OFS2Reg) learning framework for MRIRG that leverages the potential of cross-modal vessel structure similarities to construct keypoint topology constraints under a shared detection network. To be specific, OFS2Reg applies a well-designed frequency-guided structure preservation descriptor to construct coarse keypoint topology constraints in descriptors by highlighting significant vessel structures. Additionally, we develop a new cross-modal vessel structure similarity learning method that refines cross-modal keypoint topology constraints to precisely align pixel-level cross-modal vessel structures by leveraging cross-modal vessel structure similarity priors.
Artificial intelligence (AI) systems now produce research results in bulk: in October 2026 a single laboratory released more than 700 AI-generated mathematics manuscripts, only part of them with machine-checked proofs. Such volumes will be checked mostly by AI agents that call specialised software. A passing check certifies only the artefact that was checked. For proofs translated into the proof assistant Lean, it has been shown that this artefact may state something other than the author's text, and that deciding whether a translation is faithful is undecidable in general. The human researcher therefore stays in the loop, and human and AI contributions must be recorded. We propose a domain-independent, human-in-the-loop framework for checking AI-produced results. Validated, agent-usable tools with different grounds of trust check the results; a fidelity audit asks whether each check addressed the claim as stated; the human decides at fixed points; and a dialog report attributes the work to the human and to the AI. We describe two automatic implementations of the tool layer, for mathematics (Wolfram Language, MATLAB, Lean) and statistics (Python and R), and run the mathematical one on a two-claim example. An ambiguous claim was split into two readings, one refuted and one proved.
Static application security testing (SAST) tools surface vulnerabilities early and their verdicts are rule-based and auditable. Yet developers routinely dismiss findings they cannot interpret. Large language models can supply the missing explanation, but designs that also hand the model the detection give up the auditability that makes static analysis defensible. We keep detection with the analyser and give the model only the explaining. A preliminary user study with 20 participants and a workshop with 19 practitioners yielded five requirements. Guided by these we redesigned SAFE, an open-source IntelliJ IDEA plugin that explains findings from any analyser reporting in the Static Analysis Results Interchange Format. It tiers explanations by demand, ties them to the reported data flow, and replaces the single-prompt backend with an agent that must consult the repository and the rule that fired before answering. A within-subjects crossover experiment with 24 participants compared the redesign against its predecessor. Perceived usability rose from 77.3 to 85.0 on the System Usability Scale. Actionability gained most among the explanation-quality items, the aspect both earlier studies flagged as weakest, and the odds of a correct remediation were 2.6 times higher. Triage accuracy and time per finding were indistinguishable, which we read as uninformative rather than as equivalence. Trust was the one measure that did not move, and the closing interviews indicate that grounding made findings easier to understand rather than easier to believe. Participants also drew a boundary no requirement anticipated, wanting the tool to find and explain while they judge and apply.
Vision-language-action (VLA) policies have demonstrated strong generalization in robotic manipulation, yet processing dense visual tokens often incurs high inference latency. Visual token pruning offers a practical way to reduce this cost, but identifying a compact set of tokens that preserves reliable action generation remains challenging. Existing methods largely rely on static or immediate heuristic cues to estimate token importance. We introduce APS-VLA, a framework that learns visual token selection from robot demonstrations while keeping the pretrained VLA backbone frozen. A lightweight Decision Influence Estimator scores tokens conditioned on current visual observations, language instructions, and robot proprioception. Its training combines two complementary signals: action supervision matches both the full-input policy predictions and demonstrated actions, while an auxiliary visual transition objective uses demonstrated actions to predict subsequent visual features from the selected representation. At deployment, the estimator performs a single hard selection and packing step before the policy's native action-generation procedure; the auxiliary predictor is removed. Experiments with OpenVLA-OFT on LIBERO and with $\pi_{0.5}$ and XR-1 on RoboCasa365 demonstrate inference acceleration with modest changes in task success. Across five real-world tasks on two dual-arm platforms, APS-VLA reduces mean policy inference latency from 137.25 to 93.25 ms, achieving a $1.47\times$ speedup with an average success rate of 61%, compared with 60% for the unpruned baseline. Together, these results support APS-VLA as a modular, plug-in framework for efficient robotic control across VLA architectures, preserving the underlying policy parameters and native action-generation procedure.
AAA games offer a rich testbed for evaluating agent perception, reasoning, and control. However, evaluating whether agents can complete tasks in AAA games is challenging and often human-annotated because task outcomes are often difficult to verify automatically, let alone applying it for Reinforcement Learning (RL). We introduce GTAGym, an environment and framework for evaluation and RL on GTA V tasks. Its evaluation benchmark, GTAVBench, derived from GTA V main-story missions, contains 103 online tasks reconstructed across nine task families, alongside 47 image-based shooting and sniping tasks using GTA V imagery. A web server provides access to the online environment and evaluation interface, supporting parallel testing through screenshots and keyboard-mouse actions with automatic engine-state outcome checks. We evaluate six frontier LLMs on GTAVBench with textual task instructions and on its GTAVBench-AGI branch, where instructions appear only in the first screenshot and a harness automatically compresses interaction history. The best configuration achieves 70.9% overall accuracy in GTAVBench-AGI, yet aerial navigation, evasion and interception remain difficult. GTAGym also provides a scalable generator that samples task parameters within user-specified ranges and option sets under fixed templates, creating batches of 3D and image-based shooting environments for RL with automatic reward and termination signals. LoRA RL of Qwen3.8-27B on generated on-foot navigation tasks improves the in-domain performance and transfers to out-of-domain task categories, and native GTA V tasks from 15% to 23%.
Can a scientific figure generator know when it does not have enough evidence to draw? Complete-source illustrators map a finished methodology section to a schematic, or a table to plotting code, and are never asked this question because their input is assumed complete. The research record is rarely that complete: an outline, a draft paragraph, a table, or a one-sentence intent, in any combination, and painting measured numbers with an image model when the table is missing is a hallucinated result. TypedInk infers a figure type from {diagram, plot, plate} from a possibly incomplete source and dispatches to the renderer that can honor that type's evidence contract, refusing dispatch when the contract cannot be met. On the six gold plots in TypedFigBench (5 views, n=30), TypedInk's code path inks every gold table token when a table is observed (12/12), refuses the 12 hidden-table views instead of inventing numbers, and marks 6 intent-only views illustrative, with no execution failures. An always-draw coder with no such gate inks 30/30 charts, including all 18 with no table, and recovers zero gold numbers on any of them; two vision-language raters (not humans; human replication is future work) still judge those always-draw charts submittable 17/36 times, against TypedInk's refusal 0/36 times. A diagnostic field ablation of the type router (16 illustrations, 5 views each, n=80) locates where this failure would originate if unchecked: an outline alone yields type accuracy 0.31 and sends every gold plot into an integrity violation, and a chat-model planner narrows but does not close the gap (overall accuracy 0.85, still integrity-failing on 25% of outlines). Diagram-label recall is 0.47 on the full record and 0.19 on an outline. Gold plates are typed at dispatch (oracle integrity-fail 0.00) and rendered as stills.
Coding agents increasingly receive UI screenshots as part of bug reports, yet degraded image quality - caused by low resolution, compression artifacts, or small rendered text - can prevent them from identifying the relevant UI elements and source files. Vision-language models become less reliable as image quality deteriorates, while OCR-assisted approaches may propagate recognition errors from degraded inputs. Both approaches are limited when the visual evidence required for localization is no longer clearly observable in the query screenshot. We propose augmenting coding agents with repository-rendered reference images: crisp screenshots generated from the same codebase that the agent is tasked with modifying. These references provide readable visual evidence of the application's UI while preserving the current, potentially bug-containing implementation state. Because they are generated from the available repository rather than requiring a known bug-free version, they can be produced automatically for reproducible application states without manual image annotation. We evaluate the approach on a controlled benchmark derived from SWE-bench Multimodal by injecting visual UI bugs into a real web application. We vary screenshot degradation and reference-image provision and measure whether the coding agent identifies the correct source file. The results show that reference images improve file localization especially for degraded image quality, increasing the gold file match rate under fourfold downsampling with JPEG compression. We confirmed with further analysis that the proposed approach recognizes text approximately 12 percentage points better than language biased OCR under the same degradation.
As LLMs increasingly mediate everyday interactions, subtle gender biases in open-ended responses pose growing societal risks. Existing probing methods have two key limitations: they rely on fixed-format tasks, such as pronoun resolution or occupation association, that alignment-tuned models can easily handle; and they compare raw group-level rubric scores, ignoring per-dimension variance and obscuring which bias dimensions reliably distinguish gender groups. We introduce GRAP (Gender Rubric-Aware Probing), a framework that uses a lightweight logistic probe to identify the most discriminative bias dimensions from responses scored across 10 rubric dimensions, operating task-agnostically on open-ended instruction following.
AI systems can produce reliable results without developing human understanding. However, human understanding remains valuable for shaping subsequent AI-supported work, forming preferences, governing collective choices and satisfying curiosity, even when human participation is unnecessary for correcting errors. We propose understanding-first AI engineering, which treats purpose-specific human understanding as a first-class design principle guiding requirements, design and operation. The purposes of understanding are organised into four complementary rationales: reflective agency, effective action, collective governance and epistemic value. A reference architecture connects production, verification, use, and learning and exploration through shared records, supporting understanding through explanation and participation. Five candidate architectural patterns guide choices about parallel learning, explanatory models, goal exploration, audience-specific views and distributed expertise. Qualitative scenario analysis of two hypothetical use cases demonstrates how the purpose classification, reference architecture and selected patterns guide investigation, participation and decision processes while retaining operational controls. The cases illustrate design consequences and limitations rather than establish empirical effectiveness. The approach makes human understanding an explicit engineering commitment whose justification need not depend on improving the original AI task.
Electronic design automation (EDA) workflows remain highly fragmented, requiring engineers to coordinate heterogeneous representations, specialised software tools, and iterative verification across multiple design stages. Although large language models have brought more general reasoning and code-generation capabilities to EDA, many existing approaches remain focused on isolated tasks and struggle to plan, verify, and repair designs autonomously within executable engineering environments. Here we introduce ChipOmni, a knowledge-enhanced, context-evolving AI agent framework for EDA. ChipOmni integrates multimodal understanding, dynamic planning, multi-agent collaboration, and verification-driven tool interaction into a unified autonomous system. By continually incorporating feedback from external engineering tools, the framework maintains an evolving task context, integrates domain knowledge, and iteratively refines intermediate designs. Across benchmarks for analog circuit, RTL, testbench, and formal-assertion generation, ChipOmni improves multiple task-specific measures over the evaluated baselines and solves all 30 AnalogCoder-Pro tasks. In a long-horizon case study, ChipOmni advances a natural-language RISC-V specification through three design levels to a micro-SoC GDSII layout that passes physical sign-off, without human intervention after initialization. Together, these results show that ChipOmni can coordinate verification-driven design across heterogeneous tasks and sustain long-horizon engineering workflows, advancing a shift from fragmented, tool-centred processes towards agent-centred, closed-loop systems.
Block drafters such as DFlash draft a block of future tokens in one forward pass, and draft trees such as DDTree verify the prefixes ranked highest by the product of the drafter's per-position distributions. Under tree verification, the expected number of accepted tokens is the probability mass the tree covers. We show that a product misallocates it: when the continuation can follow several futures, it splices them into chimeras, prefixes stitched together from different futures that the target never produces, and gives them as much budget as the futures themselves. Barring ties, no product-ranked tree can hold two futures that disagree in $m$ positions without $2^{m-1}-1$ chimeras, however well the drafter is trained, and in practice chimeras make up 27-33% of a 64-node tree. We propose Multi-Future Drafting (MFD), which restores the coupling with a latent variable instead of the chain rule: $K$ learned mode embeddings at the drafter's input draft $K$ coherent blocks in the same pass, and a mixture DDTree lets the verifier pick the future the target takes, losslessly and with $K+1$ extra vectors. On four Qwen targets from 4B to 27B parameters, at batch size 1, MFD reaches up to 7.8$\times$ speedup, 5-8% faster than the best single-mode tree trained identically and, on Qwen3-4B and Qwen3-8B, 9-16% faster than the released DARTree trained on the same prompts.
Ito et al. (2026) showed that sufficiently wide, independently trained neural networks can exhibit linear mode connectivity (LMC) with softmax temperature calibration, even without permutation. LMC means that a low-loss linear path connects the models' parameters. This behavior is explained by layerwise exponentially weighted connectivity (LEWC), whereby each merged layer outputs an exponentially weighted sum of the original models' corresponding representations, yielding ensemble-like predictions. However, prior work observed LEWC with substantial widening, such as 8192 units per hidden layer for MLPs and $16\times$ width for VGG-11, raising the question of whether such widths are necessary. We therefore search for permutations that encourage reciprocal orthogonality (RO), which suppresses interference between each model's weights and the other model's representations. This allows both models' computations to coexist in the merged network, promoting LEWC at smaller widths. We formulate the search as independent layerwise quadratic assignment problems and approximately solve their continuous relaxations using Frank--Wolfe. Experiments show that our method facilitates LMC with less width expansion. It also merges models trained on different data distributions, where weight matching, which maximizes weight similarity through permutations, struggles. The merged predictions approximate those of a logit ensemble of the original models.