arXiv subjects

Meng Lu

Publications and source records attributed to Meng Lu.

At least 19 recordsLinked to original sources

arXiv2026

Exploiting Variable Implications in Presolve for Mixed Integer Programming

Presolve for mixed integer programming (MIP) problems aims to eliminate redundant information, strengthen the formulation, and extract useful structural information for the subsequent branch-and-cut process. An important type of such structural information is the variable implications (VIs), which describe how a bound on a variable depends on a bound of a binary variable. In this paper, we develop two new presolve techniques that exploit VIs to derive reductions for MIP problems. The first technique, called VI aggregation, aggregates multiple VIs into a single inequality by using implications between a variable and a set of binary variables that form a clique. This aggregation can reduce the number of constraints and tighten the linear programming relaxation. The second technique, called VI-aware linear constraint propagation (LCP), builds on the standard LCP but incorporates VIs associated with the variable being tightened to derive more reductions and can derive tighter variable bounds. We show that although VI information is additionally considered, the tightest lower or upper bound of a variable can still be derived in linear time. Moreover, compared with a state-of-the-art approach in the literature, the proposed VI-aware LCP can derive tighter variable bounds. Computational results on MIPLIB 2017 benchmark instances demonstrate the effectiveness of VI aggregation and VI-aware LCP in improving the performance of the open-source MIP solver HiGHS. In particular, using the two proposed presolve techniques, a reduction of 4% in solving time and 6% in node number on HiGHS can be achieved.

math.OC↗

arXiv2026

Perceive, Interact, Reason: Building Tool-Augmented Visual Agents for Spatial Reasoning

While recent vision-language models (VLMs) demonstrate strong multimodal understanding, they remain limited in spatial reasoning tasks that require active evidence acquisition and multi-step visual interaction. This limitation suggests that relying solely on implicit visual representations from vision encoders is insufficient for recovering fine-grained spatial evidence. We introduce PERception-Interaction-reason Agent (PERIA), a tool-augmented visual agent for spatial reasoning tasks across map reasoning, visual probing, and vision reconstruction. PERIA uses two lightweight tool families: vision perception tools for exposing textual, symbolic, and spatial evidence, and vision interaction tools for manipulating visual context, tracing paths, and verifying spatial relations. To train PERIA, we develop a unified recipe that combines supervised tool-use trajectory synthesis, composite rewards, and Observation-Relaxed Group-in-Group Policy Optimization (OR-GIGPO) for effective multi-tool behavior. Experiments on 13 benchmarks from 8 datasets show that PERIA-8B improves over the Qwen3-8B backbone by 10.0% on in-distribution benchmarks and 4.4% on out-of-distribution benchmarks, while outperforming previous state-of-the-art baselines of similar size by 7.0%-14.8%. It also achieves performance comparable to much larger models such as Qwen3-VL-235B-A22B-Thinking and GPT-5, demonstrating the effectiveness of PERIA in enhancing spatial reasoning capabilities.

cs.CV↗

arXiv2026

Bilevel Autoresearch: Meta-Autoresearching Itself

If autoresearch is itself a form of research, then autoresearch can be applied to research itself. We present Bilevel Autoresearch, a bilevel framework in which an outer autoresearch loop improves an inner autoresearch loop by reading its code and traces, identifying bottlenecks, and generating injectable Python search mechanisms at runtime. The inner loop optimizes task performance; the outer loop optimizes how the inner loop searches. Both loops use the same LLM, so improvements come from the bilevel architecture rather than a stronger meta-level model, although the outer loop consumes additional inference and wall-clock budget. On Karpathy's GPT pretraining benchmark, the meta-autoresearch outer loop achieves a 5x improvement over the standard inner loop alone (-0.045 vs. -0.009 val_bpb), while parameter-level adjustment without mechanism change yields no reliable gain. The outer loop instantiates mechanisms from adjacent search domains, including combinatorial optimization, multi-armed bandits, and design of experiments, without human specification of the final mechanism design. Trace analysis suggests that these mechanisms break deterministic search patterns and force exploration of directions the LLM's priors avoid. The experiments demonstrate, on this benchmark, a first bilevel step: an outer loop improves the search behavior of an inner loop. Code is the mechanism carrier in this implementation, but skills, prompts, workflows, evaluators, domain principles, world-model assumptions, and memory schemas can also encode mechanisms that shape future agent behavior. This suggests a path toward recursive bootstrapping, where mechanisms discovered for the inner loop can be fed back to improve the meta-level loop itself.

cs.AI↗

arXiv2026

MEDVISTAGYM: A Scalable Training Environment for Thinking with Medical Images via Tool-Integrated Reinforcement Learning

Vision language models (VLMs) achieve strong performance on general image understanding but struggle to think with medical images, especially when performing multi-step reasoning through iterative visual interaction. Medical VLMs often rely on static visual embeddings and single-pass inference, preventing models from re-examining, verifying, or refining visual evidence during reasoning. While tool-integrated reasoning offers a promising path forward, open-source VLMs lack the training infrastructure to learn effective tool selection, invocation, and coordination in multi-modal medical reasoning. We introduce MedVistaGym, a scalable and interactive training environment that incentivizes tool-integrated visual reasoning for medical image analysis. MedVistaGym equips VLMs to determine when and which tools to invoke, localize task-relevant image regions, and integrate single or multiple sub-image evidence into interleaved multimodal reasoning within a unified, executable interface for agentic training. Using MedVistaGym, we train MedVistaGym-R1 to interleave tool use with agentic reasoning through trajectory sampling and end-to-end reinforcement learning. Across six medical VQA benchmarks, MedVistaGym-R1-8B exceeds comparably sized tool-augmented baselines by 19.10% to 24.21%, demonstrating that structured agentic training--not tool access alone--unlocks effective tool-integrated reasoning for medical image analysis.

cs.CV↗

arXiv2025

Scaling Agentic Reinforcement Learning for Tool-Integrated Reasoning in VLMs

While recent vision-language models (VLMs) demonstrate strong image understanding, their ability to "think with images", i.e., to reason through multi-step visual interactions, remains limited. We introduce VISTA-Gym, a scalable training environment for incentivizing tool-integrated visual reasoning capabilities in VLMs. VISTA-Gym unifies diverse real-world multimodal reasoning tasks (7 tasks from 13 datasets in total) with a standardized interface for visual tools (e.g., grounding, parsing), executable interaction loops, verifiable feedback signals, and efficient trajectory logging, enabling visual agentic reinforcement learning at scale. While recent VLMs exhibit strong text-only reasoning, both proprietary and open-source models still struggle with tool selection, invocation, and coordination. With VISTA-Gym, we train VISTA-R1 to interleave tool-use with agentic reasoning via multi-turn trajectory sampling and end-to-end reinforcement learning. Extensive experiments across 11 public reasoning-intensive VQA benchmarks show that VISTA-R1-8B outperforms state-of-the-art baselines with similar sizes by 9.51%-18.72%, demonstrating VISTA-Gym as an effective training ground to unlock the tool-integrated reasoning capabilities for VLMs.

cs.AI↗

arXiv2025

JudgeBoard: Benchmarking and Enhancing Small Language Models for Reasoning Evaluation

While small language models (SLMs) have shown promise on various reasoning tasks, their ability to judge the correctness of answers remains unclear compared to large language models (LLMs). Prior work on LLM-as-a-judge frameworks typically relies on comparing candidate answers against ground-truth labels or other candidate answers using predefined metrics like entailment. However, this approach is inherently indirect and difficult to fully automate, offering limited support for fine-grained and scalable evaluation of reasoning outputs. In this work, we propose JudgeBoard, a novel evaluation pipeline that directly queries models to assess the correctness of candidate answers without requiring extra answer comparisons. We focus on two core reasoning domains: mathematical reasoning and science/commonsense reasoning, and construct task-specific evaluation leaderboards using both accuracy-based ranking and an Elo-based rating system across five benchmark datasets, enabling consistent model comparison as judges rather than comparators. To improve judgment performance in lightweight models, we propose MAJ (Multi-Agent Judging), a novel multi-agent evaluation framework that leverages multiple interacting SLMs with distinct reasoning profiles to approximate LLM-level judgment accuracy through collaborative deliberation. Experimental results reveal a significant performance gap between SLMs and LLMs in isolated judging tasks. However, our MAJ framework substantially improves the reliability and consistency of SLMs. On the MATH dataset, MAJ using smaller-sized models as backbones performs comparatively well or even better than their larger-sized counterparts. Our findings highlight that multi-agent SLM systems can potentially match or exceed LLM performance in judgment tasks, with implications for scalable and efficient assessment.

cs.AI↗

arXiv2025

OPTAGENT: Optimizing Multi-Agent LLM Interactions Through Verbal Reinforcement Learning for Enhanced Reasoning

Large Language Models (LLMs) have shown remarkable reasoning capabilities in mathematical and scientific tasks. To enhance complex reasoning, multi-agent systems have been proposed to harness the collective intelligence of LLM agents. However, existing collaboration structures are either predefined or rely on majority voting or round-table debates, which can suppress correct but less dominant agent contributions. Recent approaches model multi-agent systems as graph networks but optimize purely for agent performance, neglecting the quality of interactions. We hypothesize that effective agent communication is crucial for multi-agent reasoning and that debating quality plays a significant role. To address this, we propose $\ours$, a multi-agent verbal reinforcement learning algorithm that dynamically constructs and refines multi-agent collaboration structures. Our method defines action spaces and a feedback mechanism that evaluates communication robustness and coherence throughout the debate. The final decision is achieved through a majority vote over all the agents. We assess $\ours$ on various reasoning tasks, including mathematical reasoning, creative writing, scientific reasoning, and numerical sorting. Results demonstrate that our approach significantly outperforms single-agent prompting methods and state-of-the-art multi-agent frameworks on diverse tasks.

cs.AI↗

arXiv2025

BeyondBench: Contamination-Resistant Evaluation of Reasoning in Language Models

Evaluating language models fairly is increasingly difficult as static benchmarks risk contamination by training data, obscuring whether models truly reason or recall. We introduce BeyondBench, an evaluation framework using algorithmic problem generation to create mathematically grounded problems on the fly, ensuring each test remains uncontaminated. Our framework covers 44 algorithmic tasks with 117 variations across three difficulty levels: the Easy Suite (29 tasks) for arithmetic and statistics, the Medium Suite (5 tasks, 49 variations) for sequence patterns and reasoning, and the Hard Suite (10 tasks, 68 variations) for NP-complete and constraint satisfaction problems. Each task draws from a space exceeding 10^15 unique instances, with deterministically verified solutions. We evaluated 101 language models (85 open-source, 16 closed-source), spanning 0.5B to 141B parameters and multiple quantization schemes, using three-fold evaluation for robustness. Results reveal consistent reasoning deficiencies, with performance degrading sharply as complexity increases. In Hard Suite evaluations, Gemini-2.5-pro, Llama-3.3-70B, and Qwen2.5-72B achieved accuracies of 56.21%, 27.16%, and 33.37% respectively. Performance drops significantly without tool usage, with GPT-5, GPT-5-mini, and GPT-5-nano showing declines of 16.81%, 15.86%, and 43.95% in overall accuracy. Contamination resistance rests on three guarantees: (i) the problem space vastly exceeds any static dataset, (ii) every instance has a deterministically verifiable solution, and (iii) isomorphic transformations yield semantically equivalent but syntactically novel problems. BeyondBench redefines reasoning evaluation via genuine algorithmic problem-solving. Our leaderboard is at https://ctrl-gaurav.github.io/BeyondBench/, Python package at https://pypi.org/project/beyondbench/, and codebase at https://github.com/ctrl-gaurav/BeyondBench.

cs.CL↗

arXiv2025

Evaluating the Quality of Open Building Datasets for Mapping Urban Inequality: A Comparative Analysis Across 5 Cities

While informal settlements lack focused development and are highly dynamic, the quality of spatial data for these places may be uncertain. This study evaluates the quality and biases of AI-generated Open Building Datasets (OBDs) generated by Google and Microsoft against OpenStreetMap (OSM) data, across diverse global cities including Accra, Nairobi, Caracas, Berlin, and Houston. The Intersection over Union (IoU), overlap analysis and a positional accuracy algorithm are used to analyse the similarity and alignment of the datasets. The paper also analyses the size distribution of the building polygon area, and completeness using predefined but regular spatial units. The results indicate significant variance in data quality, with Houston and Berlin demonstrating high alignment and completeness, reflecting their structured urban environments. There are gaps in the datasets analysed, and cities like Accra and Caracas may be under-represented. This could highlight difficulties in capturing complex or informal regions. The study also notes different building size distributions, which may be indicative of the global socio-economic divide. These findings may emphasise the need to consider the quality of global building datasets to avoid misrepresentation, which is an important element of planning and resource distribution.

cs.DB↗

arXiv2025

Hierarchical Reasoning Model

Reasoning, the process of devising and executing complex goal-oriented action sequences, remains a critical challenge in AI. Current large language models (LLMs) primarily employ Chain-of-Thought (CoT) techniques, which suffer from brittle task decomposition, extensive data requirements, and high latency. Inspired by the hierarchical and multi-timescale processing in the human brain, we propose the Hierarchical Reasoning Model (HRM), a novel recurrent architecture that attains significant computational depth while maintaining both training stability and efficiency. HRM executes sequential reasoning tasks in a single forward pass without explicit supervision of the intermediate process, through two interdependent recurrent modules: a high-level module responsible for slow, abstract planning, and a low-level module handling rapid, detailed computations. With only 27 million parameters, HRM achieves exceptional performance on complex reasoning tasks using only 1000 training samples. The model operates without pre-training or CoT data, yet achieves nearly perfect performance on challenging tasks including complex Sudoku puzzles and optimal path finding in large mazes. Furthermore, HRM outperforms much larger models with significantly longer context windows on the Abstraction and Reasoning Corpus (ARC), a key benchmark for measuring artificial general intelligence capabilities. These results underscore HRM's potential as a transformative advancement toward universal computation and general-purpose reasoning systems.

cs.AI↗

arXiv2025

Programming Geotechnical Reliability Algorithms using Generative AI

Programming reliability algorithms is crucial for risk assessment in geotechnical engineering. This study explores the possibility of automating and accelerating this task using Generative AI based on Large Language Models (LLMs). Specifically, the most popular LLM, i.e., ChatGPT, is used to test the ability to generate MATLAB codes for four classical reliability algorithms. The four specific examples considered in this study are: (1) First Order Reliability Method (FORM); (2) Subset simulation; (3) Random field simulation; and (4) Bayesian update using Gibbs sampling. The results obtained using the generated codes are compared with benchmark methods. It is found that the use of LLMs can be promising for generating reliability codes. Failure, limitations, and challenges of adopting LLMs are also discussed. Overall, this study demonstrates that existing LLMs can be leveraged powerfully and can contribute toward accelerating the adoption of reliability techniques in routine geotechnical engineering.

stat.AP↗

arXiv2025

Paths Not Taken: Understanding and Mending the Multilingual Factual Recall Pipeline

Multilingual large language models (LLMs) often exhibit factual inconsistencies across languages, with significantly better performance in factual recall tasks in English than in other languages. The causes of these failures, however, remain poorly understood. Using mechanistic analysis techniques, we uncover the underlying pipeline that LLMs employ, which involves using the English-centric factual recall mechanism to process multilingual queries and then translating English answers back into the target language. We identify two primary sources of error: insufficient engagement of the reliable English-centric mechanism for factual recall, and incorrect translation from English back into the target language for the final answer. To address these vulnerabilities, we introduce two vector interventions, both independent of languages and datasets, to redirect the model toward better internal paths for higher factual consistency. Our interventions combined increase the recall accuracy by over 35 percent for the lowest-performing language. Our findings demonstrate how mechanistic insights can be used to unlock latent multilingual capabilities in LLMs.

cs.CL↗

arXiv2025

DEBATE, TRAIN, EVOLVE: Self Evolution of Language Model Reasoning

Large language models (LLMs) have improved significantly in their reasoning through extensive training on massive datasets. However, relying solely on additional data for improvement is becoming increasingly impractical, highlighting the need for models to autonomously enhance their reasoning without external supervision. In this paper, we propose Debate, Train, Evolve (DTE), a novel ground truth-free training framework that uses multi-agent debate traces to evolve a single language model. We also introduce a new prompting strategy Reflect-Critique-Refine, to improve debate quality by explicitly instructing agents to critique and refine their reasoning. Extensive evaluations on seven reasoning benchmarks with six open-weight models show that our DTE framework achieve substantial improvements, with an average accuracy gain of 8.92% on the challenging GSM-PLUS dataset. Furthermore, we observe strong cross-domain generalization, with an average accuracy gain of 5.8% on all other benchmarks, suggesting that our method captures general reasoning capabilities. Our framework code and trained models are publicly available at https://github.com/ctrl-gaurav/Debate-Train-Evolve

cs.CL↗

arXiv2025

Geoinformation dependencies in geographic space and beyond

The use of geospatially dependent information, which has been stipulated as a law in geography, to model geographic patterns forms the cornerstone of geostatistics, and has been inherited in many data science based techniques as well, such as statistical learning algorithms. Still, we observe hesitations in interpreting geographic dependency scientifically as a property in geography, since interpretations of such dependency are subject to model choice with different hypotheses of trends and stationarity. Rather than questioning what can be considered as trends or why it is non-stationary, in this work, we share and consolidate a view that the properties of geographic dependency, being it trending or stationary, are essentially variations can be explained further by unobserved or unknown predictors, and not intrinsic to geographic space. Particularly, geoinformation dependency properties are in fact a projection of high dimensional feature space formed by all potential predictors into the lower dimension of geographic space, where geographic coordinates are equivalent to other predictors for modelling geographic patterns. This work brings together different aspects of geographic dependency, including similarity and heterogeneity, under a coherent framework, and aligns with the understanding of modelling in high dimensional feature space with different modelling concept including the classical geostatistics, Gaussian Process Regression and popular data science based spatial modelling techniques.

stat.AP↗

arXiv2025

Pathway to Relevance: How Cross-Encoders Implement a Semantic Variant of BM25

Mechanistic interpretation has greatly contributed to a more detailed understanding of generative language models, enabling significant progress in identifying structures that implement key behaviors through interactions between internal components. In contrast, interpretability in information retrieval (IR) remains relatively coarse-grained, and much is still unknown as to how IR models determine whether a document is relevant to a query. In this work, we address this gap by mechanistically analyzing how one commonly used model, a cross-encoder, estimates relevance. We find that the model extracts traditional relevance signals, such as term frequency and inverse document frequency, in early-to-middle layers. These concepts are then combined in later layers, similar to the well-known probabilistic ranking function, BM25. Overall, our analysis offers a more nuanced understanding of how IR models compute relevance. Isolating these components lays the groundwork for future interventions that could enhance transparency, mitigate safety risks, and improve scalability.

cs.IR↗

arXiv2024

A mathematical framework of intelligence and consciousness based on Riemannian Geometry

Understanding intelligence is a central pursuit in neuroscience, cognitive science, and artificial intelligence. Intelligence encompasses learning, problem-solving, creativity, and even consciousness. Recent advancements in geometric analysis have revealed new insights into high-dimensional information representation and organisation, exposing intrinsic data structures and dynamic processes within neural and artificial systems. However, a comprehensive framework that unifies the static and dynamic aspects of intelligence is still lacking. This manuscript proposes a mathematical framework based on Riemannian geometry to describe the structure and dynamics of intelligence and consciousness. Intelligence elements are conceptualised as tokens embedded in a high-dimensional space. The learned token embeddings capture the interconnections of tokens across various scenarios and tasks, forming manifolds in the intelligence space. Thought flow is depicted as the sequential activation of tokens along geodesics within these manifolds. During the navigation of geodesics, consciousness, as a self-referential process, perceives the thought flow, evaluates it against predictions, and provides feedback through prediction errors, adjusting the geodesic: non-zero prediction errors, such as learning, lead to the restructuring of the curved manifolds, thus changing the geodesic of thought flow. This dynamic interaction integrates new information, evolves the geometry and facilitates learning. The geometry of intelligence guides consciousness, and consciousness structures the geometry of intelligence. By integrating geometric concepts, this proposed theory offers a unified, mathematically framework for describing the structure and dynamics of intelligence and consciousness. Applicable to biological and artificial intelligence, this framework may pave the way for future research and empirical validation.

cs.AI↗

arXiv2024

Study of charged Lepton Flavor Violation in electron muon interactions

With the improvement of muon acceleration technology, it has received great interest to exploit high-energy muon beams for collision or target experiments. We investigate possible charged Lepton Flavor Violation (cLFV) processes mediated by an extra massive neutral gauge boson Zprime in electron muon interactions, either at a proposed electron muon collider or in a fixed target experiment with high-energy muon beam hitting electrons in the target. Based on Monte Carlo calculations and fast detector simulations, we study in detail our signal and possible backgrounds, giving the sensitivity results of cLFV signals at the 90% confidence level. Compared with current and prospective limits set by other experiments, electron muon interactions demonstrate significant advantages in the cLFV coupling strength sensitivity with tau in the final states. In addition, a special cLFV coupling combination, lambda_emu * lambda_mumu, can also be probed in our proposal.

hep-ex↗

arXiv2024

AI for Biomedicine in the Era of Large Language Models

The capabilities of AI for biomedicine span a wide spectrum, from the atomic level, where it solves partial differential equations for quantum systems, to the molecular level, predicting chemical or protein structures, and further extending to societal predictions like infectious disease outbreaks. Recent advancements in large language models, exemplified by models like ChatGPT, have showcased significant prowess in natural language tasks, such as translating languages, constructing chatbots, and answering questions. When we consider biomedical data, we observe a resemblance to natural language in terms of sequences: biomedical literature and health records presented as text, biological sequences or sequencing data arranged in sequences, or sensor data like brain signals as time series. The question arises: Can we harness the potential of recent large language models to drive biomedical knowledge discoveries? In this survey, we will explore the application of large language models to three crucial categories of biomedical data: 1) textual data, 2) biological sequences, and 3) brain signals. Furthermore, we will delve into large language model challenges in biomedical research, including ensuring trustworthiness, achieving personalization, and adapting to multi-modal data representation

cs.CL↗