Skip to main content
Search

University of Edinburgh

10 related articles

Subscribe to updates
  1. arXiv

    Multiple objectives stop being extra baggage: each agent learns its own preference, and the dynamic variant tops the Pareto front on four cooperative tasks

    Multi-objective multi-agent reinforcement learning has treated multiple objectives as an extra optimisation dimension stacked on the multi-agent problem. MIRL instead uses that objective structure to guide cooperation: a team preference fixes the system-level trade-off while each agent carries a learnable individual preference, adjusted by meta-gradient according to subsequent team performance. On four multi-objective particle-environment tasks, MIRL-S and MIRL-D improve Pareto-front quality over five baselines, with the dynamic variant MIRL-D strongest overall.

  2. A single-cell reference can flip the verdict: CD8 T-cell age link reverses in all 80 six-type references

    Cell proportions from bulk RNA sequencing are inferred through a single-cell reference, and that reference can change the conclusion drawn from them. In 981 PBMC donors the CD8 T-cell age association reversed in all 80 six-type references and the CD4 slope was inflated about 6.3-fold; in 48 islet donors a borderline alpha-cell association became significant in 54% to 100% of references. The error recurred across four estimator classes, its direction was set by the estimator, and fit accuracy did not flag it.

  3. arXiv

    LoRA stops giving every token a fixed rank: noise-floor squelch activates 8.6 of 32 components

    SQUELCH treats how many rank-one components a token needs as a detection question, soft-thresholding each token at its own noise level so 8.6 of 32 components activate on average, and beats LoRA by 2.2 to 3.2 points on commonsense reasoning, knowledge, math, code and GLUE.

  4. Alzheimer's classifiers stop using age as a shortcut: residualizing age in feature space nearly zeroes the age-prediction link in cognitively normal and mild cognitive impairment groups

    On structural MRI from three independent cohorts, Alzheimer's disease classifiers use age as a shortcut: older patients receive higher predicted risk. Residualizing age linearly in feature space (PMDN) brings the age-prediction correlation near zero in cognitively normal and mild cognitive impairment groups, with balanced accuracy 0.71 and Brier score about 0.22, and far less fold-to-fold spread than the adversarial CF-Net. Post hoc calibration further lowers probability error, with nonparametric isotonic regression and Bayesian binning performing best.

  5. arXiv

    Delegating quantum generative models to a remote server: the sampling protocol now estimates the corrupted-sample rate, not just accept or abort

    When the inference phase of a quantum generative model is delegated to a remote quantum server, the client can now verify the sampling result without revealing inputs, circuit, or outputs. Earlier verifiable blind sampling rested on computational assumptions, whereas this protocol builds on universal blind quantum computing and trap circuits, amortizing verification over rounds and returning an estimate of the corrupted-sample ratio beyond accept or abort. Numerical simulations run on a 4-qubit IQP Born machine for the Bars&Stripes benchmark, where trap detections correlate with wrong samples.

  6. bioRxiv

    Molecular property models skip the sparse assay data: 1.54B structure presentations, then 642 ChEMBL assays as a middle stage

    Molecular property prediction has mostly gone from structural self-supervision straight to endpoint fine-tuning, skipping the sparse experimental activity data in between. GRASP is a 93.5M-parameter graph Transformer trained in two stages: replaced-token detection on 1.54 billion ZINC20 molecule presentations, then adaptation on 642 ChEMBL assays. On 23 cluster-held-out OpenADMET endpoints, full fine-tuning reaches a mean MAE of 0.374, versus 0.383 for CheMeleon, 0.408 for ECFP4-LightGBM and 0.440 for Chemprop; adding ChEMBL adaptation lowers mean MAE from 0.407 to 0.374 and improves 21 of 23 endpoints.

  7. arXiv

    How Retrieval Credit Is Assigned Decides Search-Agent Training Gains: Local Credit Lifts Qwen3-4B Average F1 from 51.25 to 54.34

    The study builds corpus-grounded QA data with intermediate evidence annotations and compares three retrieval signals (evidence coverage Cov, coverage with dependencies Cov-Dep, answer match AM) under scalar reward bonuses versus action-aligned local credit, training Qwen3-4B on seven QA benchmarks and finding local credit outperforms scalar, with Cov-local reaching 54.34 average F1 versus 51.25 for outcome-only training.

  8. arXiv

    BIRD Unifies Iterative Reward Design Evaluation: Simple Design Choices Are Most Robust, and ERA-U/ERA-S Rank First and Second Under Matched Budgets

    The authors introduce BIRD, which expresses 14 published iterative reward design (IRD) methods as configurations of one loop, compares 6 published methods and ablates 30 design choices across MuJoCo, Meta-World, Assistax and HumanoidBench under matched feedback conditions and policy-training budgets, finding that most choices (including the authors' own proposals) have small or inconclusive effects while a few simple choices (such as greedily retaining checkpoints and summarizing feedback) help consistently; a small hill-climb from RDA yields ERA-U and ERA-S, which take first and second place in the pooled Bradley–Terry ranking, though a small human study shows high benchmark scores do not imply natural behavior.

  9. Coronary calcium is detectable opportunistically on thoracic CT: pooled analyses show 52% prevalence, and AI models can automate scoring and predict cardiovascular death risk

    This review synthesizes evidence on opportunistically identifying and quantifying coronary artery calcium (CAC) on thoracic CT performed for other clinical indications (routine thoracic CT, lung cancer screening CT, attenuation correction CT), reporting a pooled CAC prevalence of 52%, independent associations with all-cause mortality (pooled relative risk 2.13) and major adverse cardiovascular events (pooled relative risk 2.61), and reviewing novel applications including AI/machine learning automated scoring, radiomics, photon-counting CT, and CAC as a multi-system biomarker, while noting that no randomized trials assessing the clinical impact of this approach have yet been published.

  10. Unmet needs in functional neurological disorder related to digital healthcare

    This discussion article systematically maps six ways digital healthcare could address unmet needs in functional neurological disorder (FND)—using large biobank and clinic datasets to reveal epidemiology, comorbidity, mechanisms, economic burden and possibly treatment effects; augmenting complex and time-consuming assessment processes that exceed clinical capacity with artificial intelligence; improving diagnostic precision through automated tremor analysis, quantification of functional motor signs and speech recognition; improving self-management and access to therapy via online tools and telehealth; improving outcome measurement and existing therapy with wearables and telehealth; and developing new therapies such as AI-assisted therapies, biofeedback and virtual or augmented reality—while

You’re all caught up