Annealed Co-Generation: Disentangling Variables via Progressive Pairwise Modeling

Prediction of body mass index status from voice signals based on machine learning for automated medical applications

We present a novel approach to the problem of text style transfer. Unlike previous approaches requiring style-labeled training data, our method makes use of readily-available unlabeled text by relying on the implicit connection in style between adjacent sentences, and uses labeled data only at inference time. We adapt T5 (Raffel et al., 2020), a strong pretrained text-to-text model, to extract a style vector from text and use it to condition the decoder to perform style transfer. As our label-free training results in a style vector space encoding many facets of style, we recast transfers as ``targeted restyling'' vector operations that adjust specific attributes of the input while preserving others. We demonstrate that training on unlabeled Amazon reviews data results in a model that is competitive on sentiment transfer, even compared to models trained fully on labeled data. Furthermore, applying our novel method to a diverse corpus of unlabeled web text results in a single model capable of transferring along multiple dimensions of style (dialect, emotiveness, formality, politeness, sentiment) despite no additional training and using only a handful of exemplars at inference time.

With the development of communication telecom operators like China Mobile, relevant recommendation systems have achieved relatively diverse and precise user-item interactions via incorporating complicated knowledge graph with multi-modal data including text and other modals.However, modern telecom operators recommendation systems still face great challenges posed by a series of noise from sparse interactions, ambiguous user preferences, and data imbalances. Although diffusion models have been viewed as powerful denoising tools, existing methods rely on static denoising strategies definitely limiting adaptability to diverse noise types. Based on that issue, this paper introduces Dual Trainable Diffusion Models (DTDM), a framework unifying multi-model denoising for adaptive denoising policy selection. Our architecture pre-trains specialized denoisers for distinct noise profiles like sparsity and temporal drift, and employs a Deep Q-Network (DQN) to dynamically select denoising weights during the process of reverse diffusion.Our work bridges diffusion models and RL, offering a principled solution for multi-source noise in the sceneries of real-world recommendations.

Image denoising is a fundamental problem in computational photography, where achieving high perception with low distortion is highly demanding. Current methods either struggle with perceptual quality or suffer from significant distortion. Recently, the emerging diffusion model has achieved state-of-the-art performance in various tasks and demonstrates great potential for image denoising. However, stimulating diffusion models for image denoising is not straightforward and requires solving several critical problems. For one thing, the input inconsistency hinders the connection between diffusion models and image denoising. For another, the content inconsistency between the generated image and the desired denoised image introduces distortion. To tackle these problems, we present a novel strategy called the Diffusion Model for Image Denoising (DMID) by understanding and rethinking the diffusion model from a denoising perspective. Our DMID strategy includes an adaptive embedding method that embeds the noisy image into a pre-trained unconditional diffusion model and an adaptive ensembling method that reduces distortion in the denoised image. Our DMID strategy achieves state-of-the-art performance on both distortion-based and perception-based metrics, for both Gaussian and real-world image denoising.

With rapid advances in image-generative AI, interest has grown in applying these technologies across diverse domains. While current models excel at producing high-quality final images, they rarely model the intermediate stages of the drawing process. This study proposes a drawing support system that leverages generative AI to sequentially generate and modify images during the drawing process. To train the system, we constructed a custom dataset of time-series drawing images, which are typically unavailable in existing AI models. We developed an encoder–decoder model based on convolutional neural networks to predict the next frame from a current input image. To address error accumulation during recursive generation, we introduced a noise-augmented training method, incorporating noisy images into the dataset. Experimental results show that standard training suffers from image degradation over time, while the noise-augmented approach significantly improves stability and quality throughout the sequence. Averaged across all generated frames, the noise-augmented training achieved a PSNR of 20.48, SSIM of 0.81, and LPIPS of 0.13. As a benchmark, we compared this approach to PredRNN, a representative temporal model, which achieved a PSNR of 24.05, SSIM of 0.88, and LPIPS of 0.08. These results demonstrate the effectiveness of noise-augmented learning while also highlighting potential performance gains using temporal architectures. PredRNN also required more computation, with 6.41 M parameters and a 0.188 s inference time per sequence, compared to 5.20 M and 0.005 s for our model. Furthermore, in a sample-level analysis of 10 final images, the proposed model outperformed PredRNN in three cases across all metrics, suggesting its robustness in certain sequences despite architectural simplicity. Future directions include using more advanced models such as Variational Autoencoders and diffusion models, and enhancing consistency in long-term sequences. Our work confirms the feasibility of generating interactive drawing sequences using AI and sets the foundation for more robust and creative drawing support tools.

Dexterous grasp generation that considers armrelated constraints is crucial in real-world scenarios involving armenvironment collision avoidance, workspace boundary grasps, and consecutive grasping. Existing hand-centric grasp models, which primarily focus on the floating hand's pose, are insufficient for such cases. Conventional arm-aware methods either rely on rejection sampling to discard infeasible samples or require retraining on arm-specific data, leading to low sample efficiency under adverse conditions or limited generalization across different robots and environments. To overcome these limitations, this letter presents an arm-aware dexterous grasp generation framework that leverages pretrained arm-agnostic grasp models while integrating arm and environmental information only at inference time. Specifically, we formulate arm-aware constrained grasp generation as a joint optimization of hand pose and arm configuration, and derive closed-form gradients for arm-related constraints. Assuming the hand pose distribution is represented by a diffusion model, we prove that gradient-based optimization is equivalent to guided diffusion sampling, steering near-feasible samples toward the feasible region. Through comprehensive evaluation involving 10k objects across 6 scenarios, we demonstrate that the proposed framework generates feasible grasps in highly constrained settings with significantly higher probability, highlighting its advantages in realworld applications. Supplementary materials and appendix are available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://arm-aware-dexgrasp.github.io/</uri>.

Brain age, a biomarker of neurological health, is widely used in neuroimaging for early detection of neurodegenerative diseases. While deep learning models have shown promise in brain age prediction from MRI, data imbalance and model interpretability remain key challenges. This study investigates the impact of data augmentation (DA) on both predictive accuracy and explanation stability in convolutional neural networks (CNNs) for brain age prediction. We compare three training strategies: (i) a baseline model, (ii) a model augmented with real MRI scans from OASIS-3, and (iii) a model trained with synthetic data generated by a diffusion model. Model performance is evaluated using mean absolute error (MAE), while interpretability is assessed through Explainable AI (XAI) methods, including DeepSHAP, Grad-CAM, and Occlusion. Our findings indicate that synthetic augmentation improves predictive accuracy, particularly for underrepresented age groups (individuals aged 40–80 years), while real-data augmentation provides more stable feature attributions. However, differences in XAI methods suggest that explanation reliability varies across training strategies. These results highlight the trade-offs between accuracy and interpretability in AI-driven neuroimaging, emphasizing the need for balanced augmentation strategies to develop clinically trustworthy models.

Learning from imbalanced data occurs frequently in many machine learning applications. One positive example to thousands of negative instances is common in scientific applications. Unfortunately, traditional machine learning techniques often treat rare instances as noise. One popular approach for this difficulty is to resample the training data. However, this results in high false positive predictions. Hence, we propose preprocessing training data by partitioning them into clusters. This greatly reduces the imbalance between minority and majority instances in each cluster. For moderate imbalance ratio, our technique gives better prediction accuracy than other resampling method. For extreme imbalance ratio, this technique serves as a good filter that reduces the amount of imbalance so that traditional classification techniques can be deployed. More importantly, we have successfully applied our techniques to splice site prediction and protein subcellular localization problem, with significant improvements over previous predictors.

Data imputation and data generation have important applications across many domains where incomplete or missing data can hinder accurate analysis and decision-making. Diffusion models have emerged as powerful generative models capable of capturing complex data distributions across various data modalities such as image, audio, and time series. Recently, they have been also adapted to generate tabular data. In this article, we propose a diffusion model for tabular data that introduces three key enhancements: (1) a conditioning attention mechanism, (2) an encoder–decoder transformer as the denoising network, and (3) dynamic masking. The conditioning attention mechanism is designed to improve the model’s ability to capture the relationship between the condition and synthetic data. The transformer layers help model interactions within the condition (encoder) or synthetic data (decoder), while dynamic masking enables our model to efficiently handle both missing data imputation and synthetic data generation tasks within a unified framework. We conduct a comprehensive evaluation by comparing the performance of diffusion models with transformer conditioning against state-of-the-art techniques such as Variational Autoencoders, Generative Adversarial Networks, and Diffusion Models, on benchmark datasets. Our evaluation focuses on the assessment of the generated samples with respect to three important criteria, namely: (1) machine learning efficiency, (2) statistical similarity, and (3) privacy risk mitigation. For the task of data imputation, we consider the efficiency of the generated samples across different levels of missing features. The results demonstrate average superior machine learning efficiency and statistical accuracy compared to the baselines, while maintaining privacy risks at a comparable level, particularly showing increased performance in datasets with a large number of features. By conditioning the data generation on a desired target variable, the model can mitigate systemic biases, generate augmented datasets to address data imbalance issues, and improve data quality for subsequent analysis. This has significant implications for domains such as healthcare and finance, where accurate, unbiased, and privacy-preserving data are critical for informed decision-making and fair model outcomes.

The goal of this project was to work with rural electric cooperatives to facilitate the diffusion of solar energy adoption in households located in the rural and semi-urban areas of Virginia by identifying social and behavioral factors that might be unique to rural regions; and develop a model to calculate the solar adoption propensity score for household based on their demographics, social and behavioral characteristics which would provide an objective metric to cooperatives that can be further used to do targeted marketing of rooftop solar panels. This was achieved through the following tasks: (1) Conducted a survey of the members of Virginia electric cooperatives to identify demographic, social, financial and behavioral attributes of individuals who are likely to adopt rooftop solar panels. (2) Developed a highly detailed, data-driven, agent-based model of the population of Virginia, focusing on the rural regions. (3) Developed diffusion models that use social, behavioral, and demographic factors, and peer effects to study their impact on solar adoption in rural areas. (4) Built a prototype tool based on the diffusion model to help study market segmentation in rural areas and made it available to National Rural Electric Cooperative Association (NRECA). (5) Results and recommendations derived from the model were provided to NRECA to be shared with participating cooperatives. (6) Results were published in peer reviewed journals, conference proceedings and book chapters, and ideas disseminated through presentations and newsletters. There were several important methodological contributions made under this project which are detailed in the published papers, including: (1) Built a decision-adjusted model for predicting adoptors with imbalanced training data; (2) designed seeding strategies to maximize adoption given a fixed budget; (3) built a methodology to compare different agent based models; (4) created models to identify important factors that influence decision to adopt solar panels; and (5) built a methodology for building household profiles of solar generation to study the duck curve phenomenon. The team included members from the University of Virginia (lead), National Rural Electric Cooperative Association (NRECA), Arizona State University, Virginia Tech and Sandia National Laboratory. Note that no individual entity or stakeholder has incentive to promote solar in rural regions. Most of the research and work focuses around urban regions where the potential for growth in solar adoption is higher due to higher population density. This puts rural areas at a disadvantage. By improving the diffusion of solar adoption in rural parts of the country, we can not only provide clean energy to rural areas but also promote job growth and improves energy independence.

RETRACTED: An Improved Dynamic Process Neural Network Prediction Model Identification Method

Gaussian steering in continuousvariable (CV) systems, as a quantum correlation between nonlocality and entanglement, is an important quantum resource. Rapid detection of Gaussian steering is a significant challenge in quantum information process. In this paper, we employ a combination of machine learning methods, including Support Vector Machine (SVM), Backpropagation Neural Network (BPNN) and Meta-Weight-Net Neural Network (MWN) to speed up the detection. An ensemble learning approach that integrates these methods is also utilized to increase the accuracy of detection. A computable Gaussian steering quantification introduced recently in [Phys. Rev. A 110, 052427] serves as a pivotal tool for labeling the samples. A key observation is that steerable Gaussian states are vastly outnumbered by unsteerable ones, particularly in configurations where the untrusted party possesses significantly more modes than the trusted party. This leads to a highly skewed distribution of sample states in the dataset. In response to this phenomenon and to make comparison, we propose the imbalance factor and prepare three types of datasets to be trained: balanced datasets, naturally generated datasets and augmented datasets with via a data augmentation strategy. Numerical experiments for seven low modes scenarios reveal that the classifiers obtained by utilizing the ensemble learning method training on augmented dataset have the best overall performance, significantly improving generalization capabilities with low cost and high test accuracy, achieving detection times as fast as seconds, at least 100 times faster than calculating . The speed advantage of machine learning detection will be more obvious in the case of higher modes. Thus the approach is both efficient and reliable, offering valuable insights into the broader potential of machine learning applications in quantum information science and providing a robust framework for machine learning utilized to classification tasks, especially in data-imbalanced scenarios.

Recently, enhancing the generative capability of text-to-image (T2I) models has become a promising direction in both academia and industry. Prior studies often focused on either improving generative quality or reducing inference latency, but typically failed to improve both quality and speed simultaneously. Moreover, existing inference-enhancement methods do not achieve significant improvements simultaneously across both diffusion models (DMs) and autoregressive models (ARMs). In this paper, we introduce a general tuning-based inference-enhancement framework, named ${\bm {CoRe}^{2}}$, which is the first to simultaneously achieve significant generative quality and reduced inference overhead across DMs and ARMs, to the best of our knowledge. ${\bm {CoRe}^{2}}$ comprises three stages: ${\bm {Co}llect}$, ${\bm {Re}flect}$, and ${\bm {Re}fine}$. During the Collect stage, classifier-free guidance (CFG) trajectories are collected and subsequently used in the Reflect stage to train a weak model capable of reflecting the "easy-to-learn" content. Finally, during the Refine stage, CoRe$^{2}$ can utilize the trained weak model to achieve speedup and performance gain in inference. Specifically, in the early sampling steps, CoRe$^{2}$ employs weak-to-strong guidance to refine the "difficult-to-learn" and realistic content, thereby improving generative quality. In the later sampling steps, CoRe$^{2}$ can use the weak model to generate "easy-to-learn" content instead of CFG, dramatically reducing inference time. Experimental outcomes substantiates CoRe$^{2}$ achieve significant performance improvements on HPD v2, Pick-of-Pic, Drawbench, GenEval, and T2I-Compbench across SDXL, SD3.5, FLUX and LlamaGen. Notably, for SD3.5, CoRe$^{2}$ can be seamlessly integrated with the state-of-the-art inference-enhancement algorithm Z-Sampling, outperforming it with winning rates of 73% and 69% on PickScore and AES even with less time.

PurposeUnderstanding anatomical structures in laparoscopic images is crucial for various types of laparoscopic surgery. However, creating specialized datasets for each type is both inefficient and challenging. This highlights the clinical significance of exploring class-incremental semantic segmentation (CISS) for laparoscopic images. Although CISS has been widely studied in diverse image datasets, in clinical settings, incremental data typically consists of new patient images rather than reusing previous images, necessitating a novel algorithm.MethodsWe introduce a distillation approach driven by a diffusion model for CISS of laparoscopic images. Specifically, an unconditional diffusion model is trained to generate synthetic laparoscopic images, which are then incorporated into subsequent training steps. A distillation network is employed to extract and transfer knowledge from networks trained in earlier steps. Additionally, to address the challenge posed by the limited semantic information available in individual laparoscopic images, we employ cross-image contrastive learning, enhancing the model’s ability to distinguish subtle variations across images.ResultsOur method was trained and evaluated on all 11 anatomical structures from the Dresden Surgical Anatomy Dataset, which presents significant challenges due to its dispersed annotations. Extensive experiments demonstrate that our approach outperforms other methods, especially in difficult categories such as the ureter and vesicular glands, where it surpasses even supervised offline learning.ConclusionThis study is the first to address class-incremental semantic segmentation for laparoscopic images, significantly improving the adaptability of segmentation models to new anatomical classes in surgical procedures.

This paper presents an unsupervised method for single-channel blind dereverberation and room impulse response (RIR) estimation, called BUDDy. The algorithm is rooted in Bayesian posterior sampling: it combines a likelihood model enforcing fidelity to the reverberant measurement, and an anechoic speech prior implemented by an unconditional diffusion model. We design a parametric filter representing the RIR, with exponential decay for each frequency subband. Room acoustics estimation and speech dereverberation are jointly carried out, as the filter parameters are iteratively estimated and the speech utterance refined along the reverse diffusion trajectory. In a blind scenario where the RIR is unknown, BUDDy successfully performs speech dereverberation in various acoustic scenarios, significantly outperforming other blind unsupervised baselines. Unlike supervised methods, which often struggle to generalize, BUDDy seamlessly adapts to different acoustic conditions. This paper extends our previous work by offering new experimental results and insights into the algorithm's versatility. We demonstrate the robustness of our proposed method to new acoustic and speaker conditions, as well as its adaptability to highresolution singing voice dereverberation, using both instrumental metrics and subjective listening evaluation. We study BUDDy's performance for RIR estimation and observe it surpasses a state-of-the-art supervised DNN-based estimator on mismatched acoustic conditions. Finally, we investigate the sensitivity of informed dereverberation methods to RIR estimation errors, thereby motivating the joint acoustic estimation and dereverberation design. Audio examples and code can be found online.