Skip to main content
Search

Imperial College London

32 related articles

Subscribe to updates
  1. arXiv

    The edge of chaos shifts at finite width: Crinit places deep networks on it without a search and trains them faster

    Deep networks have an edge of chaos in the plane of weight and bias variances where signals and gradients propagate with depth; its location was known only in the infinite-width limit, and whether and how it moves at finite width was disputed. Treating the logarithm of the distance between two inputs as a random walk, this work computes its drift and variance layer by layer and finds the finite-width edge shifted to a larger weight variance, by an amount that falls as width grows. The resulting initialisation rule, Crinit, places a network on that edge without a search and is more accurate and faster to train than standard rules on MNIST and Fashion-MNIST.

  2. arXiv

    PoreTimeNet segments a whole micro-CT sequence at once, cutting carbonate porosity error from 0.97 to 0.28 points

    A single time-resolved micro-CT experiment can yield tens to hundreds of three-dimensional images, and segmenting each one independently lets errors drift between frames and be mistaken for physical change. PoreTimeNet feeds several images from one temporal window through a shared spatial encoder and borrows information across time steps with temporal attention and decoder fusion. On a carbonate dissolution sequence its mean absolute porosity error was 0.28 percentage points against 0.97 for a 3D model that ignores temporal information, and rock boundary F1 rose from 0.833 to 0.941.

  3. arXiv

    A multi-disease simulation swaps its road-injury module for a generative model, cutting runtime 86.6% while DALY error stays at 13%

    The Malawi multi-disease agent-based simulation Thanzi La Onse replaces step-by-step event simulation of its road traffic injury module with a conditional tabular generative adversarial network that samples outcomes directly at the agent level. Across 50 independent runs, mean CPU runtime improves by 86.6% (95% bootstrap confidence interval [85.2, 88.1]), mean disability-adjusted life years averaged over age, sex and year carry a 13% error, and multi-disease dynamics are still recovered.

  4. arXiv

    Airlock stops multi-agent faults at the data path, not the whole workflow, keeping 68 to 96 points more healthy work

    In multi-agent systems, one agent's faulty output can be read by peers and spread into external actions before anyone detects it. Airlock narrows permissions down a delegation tree and records cross-agent data consumption in an append-only provenance graph; after an alert it traces forward to isolate tainted executions and pending actions. Across 1,832 testbed runs, 161 physical smart-home runs and 24,266 simulation records, Airlock blocked all post-alert faulty actuations on its main testbed and smart-home grids with zero post-convergence leaks, retaining 68 to 96 percentage points more healthy work than native A2A and LangGraph controls.

  5. arXiv

    Liver CT diagnosis switches to composing findings: diseases unseen in training are detected and localized

    Liver CT diagnosis models usually predict disease labels directly from images, so they only recognize diseases seen in training. DbC splits diagnosis into two steps: it learns a set of radiological findings as dense maps over the scan, then defines each disease by the findings its guidelines require and composes the diagnosis map with a fixed, parameter-free soft-minimum operator. On multi-phase liver CT, for six diseases never seen in training, DbC localized lesions significantly better than every comparator, including models that align images with disease names, and kept its advantage on rare diseases from another institution and on a public external cohort.

  6. medRxiv

    Bedside monitors stop guessing from 10-second clips: BioLT streams ECG, PPG and EEG into one trajectory, topping detection AUROC in 15 of 21 settings

    Clinical ECG, PPG and EEG now arrive as hours-long streams, yet most analysis still runs on isolated windows of about 10 seconds and cannot handle long, irregular, frequently interrupted signals. BioLT turns any window-level feature extractor or pretrained biosignal foundation model into a streaming model: it feeds the elapsed time between windows as an explicit input channel and keeps long-range context in a fixed-size compressed memory bank, so training and inference stay linear in recording length. Across CPSC-2021, VitalDB, MC-MED and CHB-MIT, covering arrhythmia, sepsis and seizure detection plus forecasting up to 20 minutes, BioLT had the highest detection AUROC in 15 of 21 dataset-feature settings and the highest 20-minute forecasting AUROC in 13.

  7. Finite-element solvers become differentiable graph programs: one H200 runs a step up to 181 times faster than 32 CPU cores

    A finite-element solver for conjugate heat transfer and buoyant flow is recast as fixed-weight graph programs inside an automatic-differentiation framework, with assembled operators and the algebraic-multigrid cycle executed as sparse graph operations and nothing trained. With discretization, mesh, time step and tolerances held fixed, the same problem runs up to 181 times faster per step on one H200 than on 32 CPU cores, and the pressure solve alone 18.7 times faster. Reverse-mode gradients match central differences, and implicit differentiation holds one gradient near 4.9 forward solves.

  8. arXiv

    Multifidelity optimization stops letting cheap runs eat the budget: on DTMB 5415 the hull resistance falls further than the single-fidelity baseline

    Multifidelity active learning now folds the choice of fidelity into the choice of where to sample, instead of fixing the location first. MF-CAS scores each candidate by the high-fidelity improvement opportunity per unit cost using decomposed uncertainty intervals; on the constrained 14-dimensional DTMB 5415 hull hydrodynamic optimization with a 100 equivalent-high-fidelity budget, its resistance reduction exceeds that of the single-fidelity LCB baseline, while the multifidelity LCB baseline barely improves on the initial design.

  9. arXiv

    Long-horizon agent training stops grouping by repeated states alone: adding task progress to the comparison lifts WebShop success by 3.96 and 5.16 points

    Long-horizon language-model agent training now builds action preferences inside existing rollout groups, matching on pre-action state plus task progress, and raises mean WebShop success by 3.96 and 5.16 points on Qwen3-4B and 8B.

  10. 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.