Ensembling in deep learning improves accuracy and calibration over single networks. The traditional aggregation approach, ensemble averaging, treats all individual networks equally by averaging their outputs. Inspired by crowdsourcing we propose an aggregation method called soft Dawid Skene for deep ensembles that estimates confusion matrices of ensemble members and weighs them according to their inferred performance. Soft Dawid Skene aggregates soft labels in contrast to hard labels often used in crowdsourcing. We empirically show the superiority of soft Dawid Skene in accuracy, calibration and out of distribution detection in comparison to ensemble averaging in extensive experiments.

We consider standard gradient descent, gradient flow and conjugate gradients as iterative algorithms for minimizing a penalized ridge criterion in linear regression. While it is well known that conjugate gradients exhibit fast numerical convergence, the statistical properties of their iterates are more difficult to assess due to inherent nonlinearities and dependencies. On the other hand, standard gradient flow is a linear method with well known regularizing properties when stopped early. By an explicit non-standard error decomposition we are able to bound the prediction error for conjugate gradient iterates by a corresponding prediction error of gradient flow at transformed iteration indices. This way, the risk along the entire regularisation path of conjugate gradient iterations can be compared to that for regularisation paths of standard linear methods like gradient flow and ridge regression. In particular, the oracle conjugate gradient iterate shares the optimality properties of the gradient flow and ridge regression oracles up to a constant factor. Numerical examples show the similarity of the regularisation paths in practice.

Online grooming is a severe social threat where sexual predators gradually entrap child victims with subtle and gradual manipulation. Therefore, timely intervention for online grooming is critical for proactive protection. However, previous methods fail to determine the optimal intervention points (i.e., jump to conclusions) as they rely on chat-level risk labels by causing weak supervision of risky utterances. For timely detection, we propose speed control reinforcement learning (SCoRL) (The code and supplementary materials are available at https://github.com/jinmyeongAN/SCoRL), incorporating a practical strategy derived from luring communication theory (LCT). To capture the predator's turn-level entrapment, we use a turn-level risk label based on the LCT. Then, we design a novel speed control reward function that balances the trade-off between speed and accuracy based on turn-level risk label; thus, SCoRL can identify the optimal intervention moment. In addition, we introduce a turn-level metric for precise evaluation, identifying limitations in previously used chat-level metrics. Experimental results show that SCoRL effectively preempted online grooming, offering a more proactive and timely solution. Further analysis reveals that our method enhances performance while intuitively identifying optimal early intervention points.

Objective. Large vessel occlusion (LVO) stroke presents a major challenge in clinical practice due to the potential for poor outcomes with delayed treatment. Treatment for LVO involves highly specialized care, in particular endovascular thrombectomy, and is available only at certain hospitals. Therefore, prehospital identification of LVO by emergency ambulance services, can be critical for triaging LVO stroke patients directly to a hospital with access to endovascular therapy. Clinical scores exist to help distinguish LVO from less severe strokes, but they are based on a series of examinations that can take minutes and may be impractical for patients with dementia or those who cannot follow commands due to their stroke. There is a need for a fast and reliable method to aid in the early identification of LVO. In this study, our objective was to assess the feasibility of using 30-second photoplethysmography (PPG) recording to assist in recognizing LVO stroke. Method. A total of 88 patients, including 25 with LVO, 27 with stroke mimic (SM), and 36 non-LVO stroke patients (NL), were recorded at the Liverpool Hospital emergency department in Sydney, Australia. Demographics (age, sex), as well as morphological features and beating rate variability measures, were extracted from the PPG. A binary classification approach was employed to differentiate between LVO stroke and NL+SM (NL.SM). A 2:1 train-test split was stratified and repeated randomly across 100 iterations. Results. The best model achieved a median test set area under the receiver operating characteristic curve (AUROC) of 0.77 (0.71--0.82). \textit{Conclusion.} Our study demonstrates the potential of utilizing a 30-second PPG recording for identifying LVO stroke.

Robotics researchers increasingly leverage large language models (LLM) in robotics systems, using them as interfaces to receive task commands, generate task plans, form team coalitions, and allocate tasks among multi-robot and human agents. However, despite their benefits, the growing adoption of LLM in robotics has raised several safety concerns, particularly regarding executing malicious or unsafe natural language prompts. In addition, ensuring that task plans, team formation, and task allocation outputs from LLMs are adequately examined, refined, or rejected is crucial for maintaining system integrity. In this paper, we introduce SafePlan, a multi-component framework that combines formal logic and chain-of-thought reasoners for enhancing the safety of LLM-based robotics systems. Using the components of SafePlan, including Prompt Sanity COT Reasoner and Invariant, Precondition, and Postcondition COT reasoners, we examined the safety of natural language task prompts, task plans, and task allocation outputs generated by LLM-based robotic systems as means of investigating and enhancing system safety profile. Our results show that SafePlan outperforms baseline models by leading to 90.5% reduction in harmful task prompt acceptance while still maintaining reasonable acceptance of safe tasks.

Accurate, noninvasive glioma characterization is crucial for effective clinical management. Traditional methods, dependent on invasive tissue sampling, often fail to capture the spatial heterogeneity of the tumor. While deep learning has improved segmentation and molecular profiling, few approaches simultaneously integrate tumor morphology and molecular features. Foundation deep learning models, which learn robust, task-agnostic representations from large-scale datasets, hold great promise but remain underutilized in glioma imaging biomarkers. We propose the Multi-Task SWIN-UNETR (MTS-UNET) model, a novel foundation-based framework built on the BrainSegFounder model, pretrained on large-scale neuroimaging data. MTS-UNET simultaneously performs glioma segmentation, histological grading, and molecular subtyping (IDH mutation and 1p/19q co-deletion). It incorporates two key modules: Tumor-Aware Feature Encoding (TAFE) for multi-scale, tumor-focused feature extraction and Cross-Modality Differential (CMD) for highlighting subtle T2-FLAIR mismatch signals associated with IDH mutation. The model was trained and validated on a diverse, multi-center cohort of 2,249 glioma patients from seven public datasets. MTS-UNET achieved a mean Dice score of 84% for segmentation, along with AUCs of 90.58% for IDH mutation, 69.22% for 1p/19q co-deletion prediction, and 87.54% for grading, significantly outperforming baseline models (p<=0.05). Ablation studies validated the essential contributions of the TAFE and CMD modules and demonstrated the robustness of the framework. The foundation-based MTS-UNET model effectively integrates tumor segmentation with multi-level classification, exhibiting strong generalizability across diverse MRI datasets. This framework shows significant potential for advancing noninvasive, personalized glioma management by improving predictive accuracy and interpretability.

Current techniques for privacy auditing of large language models (LLMs) have limited efficacy -- they rely on basic approaches to generate canaries which leads to weak membership inference attacks that in turn give loose lower bounds on the empirical privacy leakage. We develop canaries that are far more effective than those used in prior work under threat models that cover a range of realistic settings. We demonstrate through extensive experiments on multiple families of fine-tuned LLMs that our approach sets a new standard for detection of privacy leakage. For measuring the memorization rate of non-privately trained LLMs, our designed canaries surpass prior approaches. For example, on the Qwen2.5-0.5B model, our designed canaries achieve $49.6\%$ TPR at $1\%$ FPR, vastly surpassing the prior approach's $4.2\%$ TPR at $1\%$ FPR. Our method can be used to provide a privacy audit of $\varepsilon \approx 1$ for a model trained with theoretical $\varepsilon$ of 4. To the best of our knowledge, this is the first time that a privacy audit of LLM training has achieved nontrivial auditing success in the setting where the attacker cannot train shadow models, insert gradient canaries, or access the model at every iteration.

We examine LLM representations of gender for first names in various occupational contexts to study how occupations and the gender perception of first names in LLMs influence each other mutually. We find that LLMs' first-name gender representations correlate with real-world gender statistics associated with the name, and are influenced by the co-occurrence of stereotypically feminine or masculine occupations. Additionally, we study the influence of first-name gender representations on LLMs in a downstream occupation prediction task and their potential as an internal metric to identify extrinsic model biases. While feminine first-name embeddings often raise the probabilities for female-dominated jobs (and vice versa for male-dominated jobs), reliably using these internal gender representations for bias detection remains challenging.

Data in the real world often has an evolving distribution. Thus, machine learning models trained on such data get outdated over time. This phenomenon is called model drift. Knowledge of this drift serves two purposes: (i) Retain an accurate model and (ii) Discovery of knowledge or insights about change in the relationship between input features and output variable w.r.t. the model. Most existing works focus only on detecting model drift but offer no interpretability. In this work, we take a principled approach to study the problem of interpretable model drift detection from a risk perspective using a feature-interaction aware hypothesis testing framework, which enjoys guarantees on test power. The proposed framework is generic, i.e., it can be adapted to both classification and regression tasks. Experiments on several standard drift detection datasets show that our method is superior to existing interpretable methods (especially on real-world datasets) and on par with state-of-the-art black-box drift detection methods. We also quantitatively and qualitatively study the interpretability aspect including a case study on USENET2 dataset. We find our method focuses on model and drift sensitive features compared to baseline interpretable drift detectors.

With the booming development of prosumers, there is an urgent need for a prosumer energy management system to take full advantage of the flexibility of prosumers and take into account the interests of other parties. However, building such a system will undoubtedly reveal users' privacy. In this paper, by solving the non-independent and identical distribution of data (Non-IID) problem in federated learning with federated cluster average(FedClusAvg) algorithm, prosumers' information can efficiently participate in the intelligent decision making of the system without revealing privacy. In the proposed FedClusAvg algorithm, each client performs cluster stratified sampling and multiple iterations. Then, the average weight of the parameters of the sub-server is determined according to the degree of deviation of the parameter from the average parameter. Finally, the sub-server multiple local iterations and updates, and then upload to the main server. The advantages of FedClusAvg algorithm are the following two parts. First, the accuracy of the model in the case of Non-IID is improved through the method of clustering and parameter weighted average. Second, local multiple iterations and three-tier framework can effectively reduce communication rounds.