arXiv subjects

Geoffrey Martin

Publications and source records attributed to Geoffrey Martin.

8 recordsLinked to original sources

arXiv2026

Capacity Planning for Resource-Constrained Clinical LLM Serving

Deploying large language models (LLMs) under a fixed accelerator budget requires jointly deciding how computational capacity is allocated across model sizes and how incoming requests are routed across that capacity. Although prior work has extensively studied LLM routing and serving efficiency, the relative importance of fleet composition and routing under resource constraints remains less understood. We investigate this question using a controlled four-GPU serving environment with heterogeneous clinical NLP workloads spanning short factual verification, biomedical semantic matching, and long-context clinical summarization. We systematically vary fleet composition, routing policy, and offered load while holding the workload trace and hardware budget fixed. Across 24 configurations, descriptive main-effect shares attribute more variation to fleet composition than to the evaluated routing policies in time to first token, latency, and long-context coverage. All-small fleets provide the best responsiveness and throughput but substantially lower long-context coverage, whereas larger-model capacity improves coverage at the cost of serving efficiency. Heterogeneous fleets provide intermediate operating points but can introduce strong-model bottlenecks when large-model capacity is scarce. In a focused evaluation of this bottleneck, capacity-aware routing reduces mean latency and increases throughput while incurring a small reduction in long-context coverage. Within this measured regime, the results support prioritizing capacity allocation, with routing managing demand within the provisioned capacity.

cs.AI↗

arXiv2026

Explainable Suicide Risk Assessment on Social Media with Multi-Task QLoRA

Explainable suicide-risk assessment requires models not only to estimate risk severity, but also to identify supporting language and the risk and protective factors expressed in a post. We present our system for the IEEE BigData 2026 Cup on Explainable Suicide Risk Assessment on Social Media, which addresses three tasks: risk-level classification, evidence phrase extraction, and multi-label factor identification. Our approach adapts Qwen2.5-Instruct models using quantized low-rank adaptation (QLoRA) and an answer-masked causal language-model objective. We jointly train across all three tasks for risk classification, jointly train on Tasks 1a and 1b for evidence extraction, and adapt Task 2 separately for factor identification. We also tailor aggregation to each output: we average risk-level probabilities from the 32B and 72B models, combine evidence phrases through cross-fold consensus, and calibrate factor-specific decisions through rate matching based on out-of-fold operating points. On the official leaderboard, the final system achieved a composite score of 0.7738, with 0.8089 on Task 1 and 0.6919 on Task 2. Across the evaluated configurations, three-task training performed best for Task 1a, joint training on Tasks 1a and 1b performed best for Task 1b, and task-specific training performed best for Task 2. Probability averaging further improved Task 1a when component models had complementary errors. These findings highlight the value of tailoring both training objectives and aggregation strategies to the output structure of each task within a unified language-model framework.

cs.CL↗

arXiv2026

Comparing LLM and Fine-Tuned Model Performance on NVDRS Circumstance Extraction with Varying Prompt Complexity

Suicide is a leading cause of death in the United States, and understanding the circumstances that precede it requires extracting structured information from death investigation narratives. Many of these circumstances require semantic inference beyond simple keyword matching. We develop a ``Complexity Score'' algorithm that analyzes coding manual structure to predict when detailed prompts with full coding guidelines improve over name-only prompts. We then construct a hybrid approach that selects prompt strategy per circumstance. We evaluate large language models (LLMs) against fine-tuned RoBERTa on 25 inferentially complex circumstances from the National Violent Death Reporting System (NVDRS). We found that LLMs substantially outperform on low-prevalence circumstances where training data is insufficient. We further demonstrate that our framework generalizes across frontier LLMs, with GPT-5.2, Gemini 2.5 Pro and Llama-3 70B showing consistent performance patterns. These findings support a hybrid architecture where LLMs handle rare, inferentially complex circumstances while fine-tuned models handle common ones.

cs.CL↗

arXiv2026

Budget-Aware Routing for Long Clinical Text

A key challenge for large language models is token cost per query and overall deployment cost. Clinical inputs are long, heterogeneous, and often redundant, while downstream tasks are short and high stakes. We study budgeted context selection, where a subset of document units is chosen under a strict token budget so an off-the-shelf generator can meet fixed cost and latency constraints. We cast this as a knapsack-constrained subset selection problem with two design choices, unitization that defines document segmentation and selection that determines which units are kept. We propose \textbf{RCD}, a monotone submodular objective that balances relevance, coverage, and diversity. We compare sentence, section, window, and cluster-based unitization, and introduce a routing heuristic that adapts to the budget regime. Experiments on MIMIC discharge notes, Cochrane abstracts, and L-Eval show that optimal strategies depend on the evaluation setting. Positional heuristics perform best at low budgets in extractive tasks, while diversity-aware methods such as MMR improve LLM generation. Selector choice matters more than unitization, with cluster-based grouping reducing performance and other schemes behaving similarly. ROUGE saturates for LLM summaries, while BERTScore better reflects quality differences. We release our code at https://github.com/stone-technologies/ACL_budget_paper.

cs.CL↗

arXiv2026

A Survey on MLLM-based Visually Rich Document Understanding: Methods, Challenges, and Emerging Trends

Visually Rich Document Understanding (VRDU) has become a pivotal area of research, driven by the need to automatically interpret documents that contain intricate visual, textual, and structural elements. Recently, Multimodal Large Language Models (MLLMs) have demonstrated significant promise in this domain, including both OCR-based and OCR-free approaches for information extraction from document images. This survey reviews recent advances in MLLM-based VRDU, highlighting emerging trends and promising research directions with a focus on two key aspects: (1) techniques for representing and integrating textual, visual, and layout features; (2) training paradigms, including pretraining, instruction tuning, and training strategies. Moreover, we address challenges such as data scarcity, handling multi-page and multilingual documents, and integrating emerging trends such as Retrieval-Augmented Generation and agentic frameworks. Our analysis offers a roadmap for advancing MLLM-based VRDU toward more scalable, reliable, and adaptable systems.

cs.CV↗

arXiv2026

CPGPrompt: Translating Clinical Guidelines into LLM-Executable Decision Support

Clinical practice guidelines (CPGs) provide evidence-based recommendations for patient care; however, integrating them into Artificial Intelligence (AI) remains challenging. Previous approaches, such as rule-based systems, face significant limitations, including poor interpretability, inconsistent adherence to guidelines, and narrow domain applicability. To address this, we develop and validate CPGPrompt, an auto-prompting system that converts narrative clinical guidelines into large language models (LLMs). Our framework translates CPGs into structured decision trees and utilizes an LLM to dynamically navigate them for patient case evaluation. Synthetic vignettes were generated across three domains (headache, lower back pain, and prostate cancer) and distributed into four categories to test different decision scenarios. System performance was assessed on both binary specialty-referral decisions and fine-grained pathway-classification tasks. The binary specialty referral classification achieved consistently strong performance across all domains (F1: 0.85-1.00), with high recall (1.00 $\pm$ 0.00). In contrast, multi-class pathway assignment showed reduced performance, with domain-specific variations: headache (F1: 0.47), lower back pain (F1: 0.72), and prostate cancer (F1: 0.77). Domain-specific performance differences reflected the structure of each guideline. The headache guideline highlighted challenges with negation handling. The lower back pain guideline required temporal reasoning. In contrast, prostate cancer pathways benefited from quantifiable laboratory tests, resulting in more reliable decision-making.

cs.AI↗

arXiv2025

X-SYCON: Xylem-Inspired Passive Gradient Control for Communication-Free Swarm Response in Dynamic Disaster Environments

We present X-SYCON, a xylem-inspired multi-agent architecture in which coordination emerges from passive field dynamics rather than explicit planning or communication. Incidents (demands) and obstructions (hazards) continually write diffusing and decaying scalar fields, and agents greedily ascend a local utility $U=ϕ_{\mathrm{DE}}-κ\,ϕ_{\mathrm{HZ}}$ with light anti-congestion and separation. A beaconing rule triggered on first contact temporarily deepens the local demand sink, accelerating completion without reducing time-to-first-response. Across dynamic, partially blocked simulated environments, we observe low miss rates and stable throughput with interpretable, tunable trade-offs over carrier count, arrival rate, hazard density, and hazard sensitivity $κ$. We derive that a characteristic hydraulic length scale $\ell\approx\sqrt{D/λ}$ predicts recruitment range in a continuum approximation, and we provide a work-conservation (Ohm-law) bound consistent with sublinear capacity scaling with team size. Empirically: (i) soft hazard penalties yield fewer misses when obstacles already block motion; (ii) throughput saturates sublinearly with carriers while reliability improves sharply; (iii) stronger arrivals can reduce misses by sustaining sinks that recruit help; and (iv) phase-stability regions shrink with hazard density but are recovered by more carriers or higher arrivals. We refer to X-SYCON as an instance of Distributed Passive Computation and Control, and we evaluate it in simulations modeling communication-denied disaster response and other constrained sensing-action regimes.

physics.soc-ph↗

arXiv2001

Inflation and Global Equivalence

This article investigates an extension of General Relativity based upon a class of lifted metrics on the cotangent bundle of space-time. The dynamics of the theory is determined by a fixed section of the cotangent bundle, representing the momentum of a fluid flow, and Einstein's equations for the fluid applied to the induced space-time metric on the submanifold of the cotangent bundle defined by the image of the section. This construction is formally analogous to the extension of Galilean Relativity by Special Relativity, and is shown to reduce to General Relativity as the gravitational constant approaches zero. By examining the consequences of the model for homogeneous cosmologies, it is demonstrated that this construction globalizes the equivalence principle, in that, the perfect fluid model of Special Relativity is sufficient to predict both the inflationary and the current era.

gr-qc↗