检索结果-内蒙古大学图书馆

TOWARDS GENERAL FUNCTION APPROXIMATION IN ZERO-SUM MARKOV GAMES 10

学校读者我要写书评

暂无评论

TOWARDS GENERAL FUNCTION APPROXIMATION IN ZERO-SUM MARKOV GA...

10th International Conference on Learning Representations, ICLR 2022

作者： Huang, Baihe Lee, Jason D. Wang, Zhaoran Yang, Zhuoran School of Mathematical Sciences Peking University China Department of Electrical and Computer Engineering Princeton University United States Departments of Industrial Engineering & Management Sciences Northwestern University United States Department of Statistics and Data Science Yale University United States

This paper considers two-player zero-sum finite-horizon Markov games with simultaneous moves. The study focuses on the challenging settings where the value function or the model is parameterized by general function classes. Provably efficient algorithms for both decoupled and coordinated settings are developed. In the decoupled setting where the agent controls a single player and plays against an arbitrary opponent, we propose a new model-free algorithm. The sample complexity is governed by the Minimax Eluder dimension-a new dimension of the function class in Markov games. As a special case, this method improves the state-of-the-art algorithm by a √d factor in the regret when the reward function and transition kernel are parameterized with d-dimensional linear features. In the coordinated setting where both players are controlled by the agent, we propose a model-based algorithm and a model-free algorithm. In the model-based algorithm, we prove that sample complexity can be bounded by a generalization of Witness rank to Markov games. The model-free algorithm enjoys a √K-regret upper bound where K is the number of episodes. © 2022 ICLR 2022 - 10th International Conference on Learning Representationss. All rights reserved.

关键词：

Learnability Gaps of Strategic Classification

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Cohen, Lee Mansour, Yishay Moran, Shay Shao, Han Stanford United States Tel Aviv University Google Research United States Departments of Mathematics Computer Science and Data and Decision Sciences Technion and Google Research Toyota Technological Institute of Chicago United States

In contrast with standard classification tasks, strategic classification involves agents strategically modifying their features in an effort to receive favorable predictions. For instance, given a classifier determining loan approval based on credit scores, applicants may open or close their credit cards and bank accounts to fool the classifier. The learning goal is to find a classifier robust against strategic manipulations. Various settings, based on what and when information is known, have been explored in strategic classification. In this work, we focus on addressing a fundamental question: the learnability gaps between strategic classification and standard learning. We essentially show that any learnable class is also strategically learnable: we first consider a fully informative setting, where the manipulation structure (which is modeled by a manipulation graph G) is known and during training time the learner has access to both the pre-manipulation data and post-manipulation data. We provide nearly tight sample complexity and regret bounds, offering significant improvements over prior results. Then, we relax the fully informative setting by introducing two natural types of uncertainty. First, following Ahmadi et al. [2023], we consider the setting in which the learner only has access to the post-manipulation data. We improve the results of Ahmadi et al. [2023] and close the gap between mistake upper bound and lower bound raised by them. Our second relaxation of the fully informative setting introduces uncertainty to the manipulation structure. That is, we assume that the manipulation graph is unknown but belongs to a known class of graphs. We provide nearly tight bounds on the learning complexity in various unknown manipulation graph settings. Notably, our algorithm in this setting is of independent interest and can be applied to other problems such as multi-label learning. Copyright © 2024, The Authors. All rights reserved.

关键词： Classification (of information)

Statistical Distance-Guided Unsupervised Domain Adaptation for Automated Multi-Class Cardiovascular Magnetic Resonance Image Quality Assessment

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Nabavi, Shahabedin Hamedani, Kian Anvari Moghaddam, Mohsen Ebrahimi Abin, Ahmad Ali Frangi, Alejandro F. Faculty of Computer Science and Engineering Shahid Beheshti University Tehran Iran Division of Informatics Imaging and Data Sciences Schools of Computer Science and Health Sciences The University of Manchester Manchester United Kingdom Electrical Engineering and Cardiovascular Sciences Departments KU Leuven Leuven Belgium Alan Turing Institute London United Kingdom

This study proposes an attention-based statistical distance-guided unsupervised domain adaptation model for multi-class cardiovascular magnetic resonance (CMR) image quality assessment. The proposed model consists of a feature extractor, a label predictor and a statistical distance estimator. An annotated dataset as the source set and an unlabeled dataset as the target set with different statistical distributions are considered inputs. The statistical distance estimator approximates the Wasserstein distance between the extracted feature vectors from the source and target data in a mini-batch. The label predictor predicts data labels of source data and uses a combinational loss function for training, which includes cross entropy and centre loss functions plus the estimated value of the distance estimator. Four datasets, including imaging and k-space data, were used to evaluate the proposed model in identifying four common CMR imaging artefacts: respiratory and cardiac motions, Gibbs ringing and Aliasing. The results of the extensive experiments showed that the proposed model, both in image and k-space analysis, has an acceptable performance in covering the domain shift between the source and target sets. The model explainability evaluations and the ablation studies confirmed the proper functioning and effectiveness of all the model's modules. The proposed model outperformed the previous studies regarding performance and the number of examined artefacts. The proposed model can be used for CMR post-imaging quality control or in large-scale cohort studies for image and k-space quality assessment due to the appropriate performance in domain shift coverage without a tedious data-labelling process. Copyright © 2024, The Authors. All rights reserved.

关键词： Magnetic resonance imaging

Multiple Teachers-Meticulous Student: A Domain Adaptive Meta-Knowledge Distillation Model for Medical Image Classification

学校读者我要写书评

暂无评论

arXiv 2024年

Background: Image classification can be considered one of the key pillars of medical image analysis. Deep learning (DL) faces challenges that prevent its practical applications despite the remarkable improvement in medical image classification. The data distribution differences can lead to a drop in the efficiency of DL, known as the domain shift problem. Besides, requiring bulk annotated data for model training, the large size of models, and the privacy-preserving of patients are other challenges of using DL in medical image classification. This study presents a strategy that can address the mentioned issues simultaneously. Method: The proposed domain adaptive model based on knowledge distillation can classify images by receiving limited annotated data of different distributions. The designed multiple teachers-meticulous student model trains a student network that tries to solve the challenges by receiving the parameters of several teacher networks. The proposed model was evaluated using six available datasets of different distributions by defining the respiratory motion artefact detection task. Results: The results of extensive experiments using several datasets show the superiority of the proposed model in addressing the domain shift problem and lack of access to bulk annotated data. Besides, the privacy preservation of patients by receiving only the teacher network parameters instead of the original data and consolidating the knowledge of several DL models into a model with almost similar performance are other advantages of the proposed model. Conclusions: The proposed model can pave the way for practical clinical applications of deep classification methods by achieving the mentioned objectives simultaneously. Copyright © 2024, The Authors. All rights reserved.

关键词： Distillation

Text Generation to Aid Depression Detection: A Comparative Study of Conditional Sequence Generative Adversarial Networks

学校读者我要写书评

暂无评论

Text Generation to Aid Depression Detection: A Comparative S...

2022 IEEE International Conference on Big data, Big data 2022

作者： Tlachac, M.L. Gerych, Walter Agrawal, Kratika Litterer, Benjamin Jurovich, Nicholas Thatigotla, Saitheeraj Thadajarassiri, Jidapa Rundensteiner, Elke A. Bryant University Center for Health and Behavioral Sciences Department of Information Systems and Analytics SmithfieldRI02911 United States Data Science and Computer Science Departments WorcesterMA01609 United States University of Michigan School of Information Ann ArborMI48109 United States University of Tennessee KnoxvilleTN37996 United States

ISBN: (纸本)9781665480451

Corpuses of unstructured textual data, such as text messages between individuals, are often predictive of medical issues such as depression. The text data usually used in healthcare applications has high value and great variety, but is typically small in volume. Generating labeled unstructured text data is important to improve models by augmenting these small datasets, as well as to facilitate anonymization. While methods for labeled data generation exist, not all of them generalize well to small datasets. In this work, we thus perform a much needed systematic comparison of conditional text generation models that are promising for small datasets due to their unified architectures. We identify and implement a family of nine conditional sequence generative adversarial networks for text generation, which we collectively refer to as cSeqGAN models. These models are characterized along two orthogonal design dimensions: weighting strategies and feedback mechanisms. We conduct a comparative study evaluating the generation ability of the nine cSeqGAN models on three diverse text datasets with depression and sentiment labels. To assess the quality and realism of the generated text, we use standard machine learning metrics as well as human assessment via a user study. While the unconditioned models produced predictive text, the cSeqGAN models produced more realistic text. Our comparative study lays a solid foundation and provides important insights for further text generation research, particularly for the small datasets common within the healthcare domain. © 2022 IEEE.

关键词： Generative adversarial networks

Dynamic Allocation Hypernetwork with Adaptive Model Recalibration for Federated Continual Learning

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Qi, Xiaoming Zhang, Jingyang Fu, Huazhu Yang, Guanyu Li, Shuo Jin, Yueming Department of Biomedical Engineering Department of Electrical and Computer Engineering National University of Singapore Singapore Singapore Ministry of Education Nanjing China Institute of High Performance Computing A*STAR Singapore Departments Biomedical Engineering and Computer and Data Science Case Western Reserve University Cleveland United States

Federated continual learning (FCL) offers an emerging pattern to facilitate the applicability of federated learning (FL) in real-world scenarios, where tasks evolve dynamically and asynchronously across clients, especially in medical scenario. Existing server-side FCL methods in nature domain construct a continually learnable server model by client aggregation on all-involved tasks. However, they are challenged by: (1) Catastrophic forgetting for previously learned tasks, leading to error accumulation in server model, making it difficult to sustain comprehensive knowledge across all tasks. (2) Biased optimization due to asynchronous tasks handled across different clients, leading to the collision of optimization targets of different clients at the same time steps. In this work, we take the first step to propose a novel server-side FCL pattern in medical domain, Dynamic Allocation Hypernetwork with adaptive model recalibration (FedDAH). It is to facilitate collaborative learning under the distinct and dynamic task streams across clients. To alleviate the catastrophic forgetting, we propose a dynamic allocation hypernetwork (DAHyper) where a continually updated hypernetwork is designed to manage the mapping between task identities and their associated model parameters, enabling the dynamic allocation of the model across clients. For the biased optimization, we introduce a novel adaptive model recalibration (AMR) to incorporate the candidate changes of historical models into current server updates, and assign weights to identical tasks across different time steps based on the similarity for continual optimization. Extensive experiments on the AMOS dataset demonstrate the superiority of our FedDAH to other FCL methods on sites with different task streams. The code is available:https://***/jinlab-imvr/FedDAH. © 2025, CC BY-SA.

关键词： Federated learning

Dynamic Allocation Hypernetwork with Adaptive Model Recalibration for Federated Continual Learning

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Qi, Xiaoming Zhang, Jingyang Fu, Huazhu Yang, Guanyu Li, Shuo Jin, Yueming Department of Biomedical Engineering Department of Electrical and Computer Engineering National University of Singapore Singapore Singapore Ministry of Education Nanjing China Institute of High Performance Computing A*STAR Singapore Departments Biomedical Engineering Computer and Data Science Case Western Reserve University Cleveland United States

关键词： Federated learning

Quantifying Strain Dependence of Multi-Frequency Shearwave Elasticity Imaging

学校读者我要写书评

暂无评论

Quantifying Strain Dependence of Multi-Frequency Shearwave E...

Western New York Image Processing Workshop (WNYIPW)

作者： Siladitya Khan Soumya Goswami Fan Feng Marvin M. Doyley Stephen A. McAleavey Departments of Biomedical & Computer Engineering University of Rochester Rochester NY USA Rochester Center for Biomedical Ultrasound University of Rochester Rochester NY USA Goergen Institute for Data Science University of Rochester Rochester NY USA Departments of Electrical & Computer Engineering University of Rochester Rochester NY USA

Nonlinear shear modulus (NLSM) can potentially differentiate benign and malignant pathologies. Previous studies demonstrated the discriminatory value of mapping quantitative estimates of material non-linearity, yet estimates remain biased due to unaccounted shear-wave dispersion within the tissue. We demonstrate a finite element simulation of an inversion technique that combines traditional quasi-static and dynamic elasticity imaging to generate a multi-frequency estimate of the third-order shear modulus A(x,z,ω) over a broad bandwidth (200-600 Hz). We show that our reconstruction results are superior to those obtained by traditional group speed by a factor of two in terms of contrast enhancement measures. Experimental ultrasound elastography on a custom oil-in-gelatin emulsion confirms the presence of dispersion in the NLSM parameter. Constitutive hyper-viscoelastic model responses of spectral NLSM are simulated based on matched linear shear modulus data of experimental phantoms. Hyper-elastic Ogden with a Voigt loss is shown to describe NLSM behavior at frequencies greater than 400 Hz. Reconstruction accuracy in the presence of simulated tracking and correlation noise demonstrates the robustness of our estimates.

关键词：

Detecting Eye Disease Using Vision Transformers Informed by Ophthalmology Resident Gaze data *

学校读者我要写书评

暂无评论

Detecting Eye Disease Using Vision Transformers Informed by ...

Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC)

作者： Shubham Kaushal Yifan Sun Ryan Zukerman Royce W.S. Chen Kaveri A. Thakoor Data Science Institute Columbia University New York NY USA Department of Industrial Engineering and Operations Research Columbia University New York NY USA Department of Ophthalmology Columbia University Irving Medical Center (CUIMC) New York NY USA Departments of Computer Science Biomedical Engineering and the Data Science Institute Columbia University New York NY USA

We showcase two proof-of-concept approaches for enhancing the Vision Transformer (ViT) model by integrating ophthalmology resident gaze data into its training. The resulting Fixation-Order-Informed ViT and Ophthalmologist-Gaze-Augmented ViT show greater accuracy and computational efficiency than ViT for detection of the eye disease, *** relevance— By enhancing glaucoma detection via our gaze-informed ViTs, we introduce a new paradigm for medical experts to directly interface with medical AI, leading the way for more accurate and interpretable AI ‘teammates’ in the ophthalmic clinic.

关键词：