Skip to main content
Search

The University of Texas at El Paso

4 related articles

Subscribe to updates
  1. arXiv

    Power sampling no longer takes one step per call: prefetching the accept/reject subtree cuts PowerMH time by a median 1.71×

    PowerMH must decide each accept/reject before it can generate the next proposal; PreSTO prefetches several future proposals in one batched call, giving a median 1.71× speedup over PowerMH across seven datasets and five base LLMs while leaving its transition kernel unchanged.

  2. medRxiv

    PCOS screening stops hiding behind one accuracy number: rarest subtype recall climbs from 75% to 87.5%

    Machine learning screening for polycystic ovary syndrome has long reported a single binary accuracy, hiding differences between clinical subtypes. On a public dataset of 541 patients, a tuned random forest reached per-subtype accuracy ranging from 75.0% on the rarest subtype to 97.8% on the most common, a 22.8-point spread that stayed at 23.3 points after removing the seven variables used to define subtypes. A multi-task neural network with random-forest feature ranking and genetic-algorithm tuning predicts the binary label and the subtype together, holding 90.2% binary accuracy while lifting rarest-subtype recall from 75.0% to 87.5% and unclassified-case recall from the multi-class model's 31.5% to 79.6%.

  3. arXiv

    A 9B model matches frontier reasoning on GPQA, AIME and LCB via parallel tempering sampling, with no parameter updates

    The work introduces Parallel Power Tempering (PPT), which adapts classical parallel tempering (replica exchange) to sequence-level power-sharpened LLM sampling: multiple replicas at different sharpening levels run in parallel so that low-power replicas explore diverse reasoning trajectories while high-power chains exploit higher-likelihood responses, with swaps carrying promising traces to the sharpest output rung, and a fixed-horizon construction removes a structural truncation bias identified in early-stopped power samplers; across MATH500, GPQA, HumanEval, GSM8K, AIME 24&25 and LiveCodeBench v5 on Qwen3-4B, Qwen3-8B and Qwen3.5-9B, PPT attains best or tied-best accuracy in all model-benchmark settings, outperforming Power Sampling, PowerSMC and GRPO, and on Qwen3.

  4. bioRxiv

    RSAUNet: A Hybrid Residual–Swin Transformer Design with Attention for Prostate Cancer Segmentation

    The study proposes RSAUNet, a deep learning architecture that combines residual convolutional blocks, Swin Transformer blocks, and attention mechanisms inside a U-Net backbone for MRI prostate cancer segmentation, reporting a Dice coefficient of 0.998 and a Jaccard index (IoU) of 0.965 on a public Kaggle prostate annotation dataset, together with an ablation study tracking loss, accuracy, Dice, and mean IoU as each component is added.

You’re all caught up