检索结果-内蒙古大学图书馆

39th Annual AAai Conference on Artificial Intelligence, AAai 2025

作者： Kong, Jiayi Song, Xurui Huai, Shuo Xu, Baixin Luo, Jun He, Ying S-Lab Nanyang Technological University Singapore College of Computing and Data Science Nanyang Technological University Singapore

ISBN: (纸本)157735897X

While 3D head reconstruction is widely used for modeling, existing neural reconstruction approaches rely on high-resolution multi-view images, posing notable privacy issues. Individuals are particularly sensitive to facial features, and facial image leakage can lead to many malicious activities, such as unauthorized tracking and deepfake. In contrast, geometric data is less susceptible to misuse due to its complex processing requirements, and absence of facial texture features. In this paper, we propose a novel two-stage 3D facial reconstruction method aimed at avoiding exposure to sensitive facial information while preserving detailed geometric accuracy. Our approach first uses non-sensitive rear-head images for initial geometry and then refines this geometry using processed privacy-removed gradient images. Extensive experiments show that the resulting geometry is comparable to methods using full images, while the process is resistant to DeepFake applications and facial recognition (FR) systems, thereby proving its effectiveness in privacy protection. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Geometry

来源：评论

学校读者我要写书评

暂无评论

Enhancing Protein Language Model With Feature Integration for Anticancer Peptide Prediction

Enhancing Protein Language Model With Feature Integration fo...

引用

2024 IEEE-EMBS International Conference on Biomedical and Health Informatics, BHI 2024

作者： Binte Sayuti, Tiara Natasha Cheng, Shen Ajith, Santhisenan Bin Abdul Samad, Abdul Hadi Rajapakse, Jagath C. College of Computing and Data Science Nanyang Technological University Health Informatics Lab Singapore College of Computing and Data Science Nanyang Technological University Singapore School of Electrical and Electronic Engineering Nanyang Technological University Singapore

ISBN: (纸本)9798350351552

In the fight against cancer, anticancer peptides (ACP) hold promising therapeutic potential due to their selective cytotoxicity and lower side effects compared to traditional treatments. However, identifying novel ACP is challenged by high costs and labor-intensive processes. Protein language models (PLMs), such as ESM-2 and ProtBERT, have revolutionized peptide prediction by leveraging vast datasets to capture complex biological patterns through pre-training. However, they often struggle to accurately model specific biochemical interactions. To address this limitation, we integrated four sequence-based features: amino acid composition (AAC), dipeptide composition (DPC), composition of k-spaced amino acid group pairs (CKS), and k-mer sparse matrix (k-mer) through a cross-attention mechanism. These features infuse biochemical insights that PLM alone may overlook, enabling a more detailed prediction of anti-cancer properties. This integration enhances biochemical insights, improving prediction accuracy by 15.8% for ProtBERT and 2.9% for ESM-2, with ESM-2 achieving the highest accuracy at 77.8%. SHapley Additive exPlanations (SHAP) analysis confirms the importance of these features, demonstrating that incorporating amino acid features into PLMs enhances ACP prediction. © 2024 IEEE.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Uncertainty-Calibrated Test-Time Model Adaptation without Forgetting

引用

IEEE Transactions on Pattern Analysis and Machine Intelligence 2025年 PP卷 1-14页

作者： Tan, Mingkui Chen, Guohao Wu, Jiaxiang Zhang, Yifan Chen, Yaofo Zhao, Peilin Niu, Shuaicheng South China University of Technology School of Software Engineering China Pazhou Laboratory Guangzhou China XVERSE China National University of Singapore School of Computing Singapore Tencent AI Lab China Nanyang Technological University College of Computing and Data Science Singapore

Test-time adaptation (TTA) seeks to tackle potential distribution shifts between training and testing data by adapting a given model w.r.t. any testing sample. This task is particularly important when the test environment changes frequently. Although some recent attempts have been made to handle this task, we still face two key challenges: 1) prior methods have to perform backpropagation for each test sample, resulting in unbearable optimization costs to many applications;2) while existing TTA solutions can significantly improve the test performance on out-of-distribution data, they often suffer from severe performance degradation on in-distribution data after TTA (known as catastrophic forgetting). To this end, we have proposed an Efficient Anti-Forgetting Test-Time Adaptation (EATA) method which develops an active sample selection criterion to identify reliable and non-redundant samples for test-time entropy minimization. To alleviate forgetting, EATA introduces a Fisher regularizer estimated from test samples to constrain important model parameters from drastic changes. However, in EATA, the adopted entropy loss consistently assigns higher confidence to predictions even when the samples are underlying uncertain, leading to overconfident predictions that underestimate the data uncertainty. To tackle this, we further propose EATA with Calibration (EATA-C) to separately exploit the reducible model uncertainty and the inherent data uncertainty for calibrated TTA. Specifically, we compare the divergence between predictions from the full network and its sub-networks to measure the reducible model uncertainty, on which we propose a test-time uncertainty reduction strategy with divergence minimization loss to encourage consistent predictions instead of overconfident ones. To further re-calibrate predicting confidence on different samples, we utilize the disagreement among predicted labels as an indicator of the data uncertainty. Based on this, we devise a min-max entropy

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Historical Test-time Prompt Tuning for Vision Foundation Models 38

Historical Test-time Prompt Tuning for Vision Foundation Mod...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Zhang, Jingyi Huang, Jiaxing Zhang, Xiaoqin Shao, Ling Lu, Shijian College of Computing and Data Science Nanyang Technological University Singapore College of Computer Science and Technology Zhejiang University of Technology China UCAS-Terminus AI Lab University of Chinese Academy of Sciences China

Test-time prompt tuning, which learns prompts online with unlabelled test samples during the inference stage, has demonstrated great potential by learning effective prompts on-the-fly without requiring any task-specific annotations. However, its performance often degrades clearly along the tuning process when the prompts are continuously updated with the test data flow, and the degradation becomes more severe when the domain of test samples changes continuously. We propose HisTPT, a Historical Test-time Prompt Tuning technique that memorizes the useful knowledge of the learnt test samples and enables robust test-time prompt tuning with the memorized knowledge. HisTPT introduces three types of knowledge banks, namely, local knowledge bank, hard-sample knowledge bank, and global knowledge bank, each of which works with different mechanisms for effective knowledge memorization and test-time prompt optimization. In addition, HisTPT features an adaptive knowledge retrieval mechanism that regularizes the prediction of each test sample by adaptively retrieving the memorized knowledge. Extensive experiments show that HisTPT achieves superior prompt tuning performance consistently while handling different visual recognition tasks (e.g., image classification, semantic segmentation, and object detection) and test samples from continuously changing domains. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Finite Volume Features, Global Geometry Representations, and Residual Training for Deep Learning-based CFD Simulation 41

Finite Volume Features, Global Geometry Representations, and...

引用

41st International Conference on Machine Learning, ICML 2024

作者： Jessica, Loh Sher En Arafat, Naheed Anjum Lim, Wei Xian Chan, Wai Lee Kong, Adams Wai Kin Rolls-Royce@NTU Corporate Lab Nanyang Technological University Singapore College of Computing and Data Science Nanyang Technological University Singapore School of Mechanical & Aerospace Engineering Nanyang Technological University Singapore

Computational fluid dynamics (CFD) simulation is an irreplaceable modelling step in many engineering designs, but it is often computationally expensive. Some graph neural network (GNN)based CFD methods have been proposed. However, the current methods inherit the weakness of traditional numerical simulators, as well as ignore the cell characteristics in the mesh used in the finite volume method, a common method in practical CFD applications. Specifically, the input nodes in these GNN methods have very limited information about any object immersed in the simulation domain and its surrounding environment. Also, the cell characteristics of the mesh such as cell volume, face surface area, and face centroid are not included in the message-passing operations in the GNN methods. To address these weaknesses, this work proposes two novel geometric representations: Shortest Vector (SV) and Directional Integrated Distance (DID). Extracted from the mesh, the SV and DID provide global geometry perspective to each input node, thus removing the need to collect this information through message-passing. This work also introduces the use of Finite Volume Features (FVF) in the graph convolutions as node and edge attributes, enabling its message-passing operations to adjust to different nodes. Finally, this work extends the use of residual training to improve flow field prediction for a GNN scenario with immersed object, when low resolution data is available. Experimental results on two datasets with five different state-of-the-art GNN methods for CFD indicate that SV, DID, FVF and residual training can effectively reduce the predictive error of current GNN-based methods by as much as 41%. Our codes and datasets are available at https://***/toggled/FvFGeo. Copyright 2024 by the author(s)

关键词： Mesh generation

来源：评论

学校读者我要写书评

暂无评论

Flow Snapshot Neurons in Action: Deep Neural Networks Generalize to Biological Motion Perception 38

Flow Snapshot Neurons in Action: Deep Neural Networks Genera...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Han, Shuangpeng Wang, Ziyu Zhang, Mengmi College of Computing and Data Science Nanyang Technological University Singapore Singapore Show Lab National University of Singapore Singapore

Biological motion perception (BMP) refers to humans' ability to perceive and recognize the actions of living beings solely from their motion patterns, sometimes as minimal as those depicted on point-light displays. While humans excel at these tasks without any prior training, current ai models struggle with poor generalization performance. To close this research gap, we propose the Motion Perceiver (MP). MP solely relies on patch-level optical flows from video clips as inputs. During training, it learns prototypical flow snapshots through a competitive binding mechanism and integrates invariant motion representations to predict action labels for the given video. During inference, we evaluate the generalization ability of all ai models and humans on 62,656 video stimuli spanning 24 BMP conditions using point-light displays in neuroscience. Remarkably, MP outperforms all existing ai models with a maximum improvement of 29% in top-1 action recognition accuracy on these conditions. Moreover, we benchmark all ai models in point-light displays of two standard video datasets in computer vision. MP also demonstrates superior performance in these cases. More interestingly, via psychophysics experiments, we found that MP recognizes biological movements in a way that aligns with human behaviors. Our data and code are available at link. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Test-Time Model Adaptation with Only Forward Passes 41

Test-Time Model Adaptation with Only Forward Passes

引用

41st International Conference on Machine Learning, ICML 2024

作者： Niu, Shuaicheng Miao, Chunyan Chen, Guohao Wu, Pengcheng Zhao, Peilin College of Computing and Data Science Nanyang Technological University Singapore Joint NTU-WeBank Research Centre on Fintech Singapore Singapore Tencent AI Lab Shenzhen China

Test-time adaptation has proven effective in adapting a given trained model to unseen test samples with potential distribution shifts. However, in real-world scenarios, models are usually deployed on resource-limited devices, e.g., FPGAs, and are often quantized and hard-coded with non-modifiable parameters for acceleration. In light of this, existing methods are often infeasible since they heavily depend on computation-intensive backpropagation for model updating that may be not supported. To address this, we propose a test-time Forward-Optimization Adaptation (FOA) method. In FOA, we seek to solely learn a newly added prompt (as model's input) via a derivative-free covariance matrix adaptation evolution strategy. To make this strategy work stably under our online unsupervised setting, we devise a novel fitness function by measuring test-training statistic discrepancy and model prediction entropy. Moreover, we design an activation shifting scheme that directly tunes the model activations for shifted test samples, making them align with the source training domain, thereby further enhancing adaptation performance. Without using any backpropagation and altering model weights, FOA runs on quantized 8-bit ViT outperforms gradient-based TENT on full-precision 32-bit ViT, while achieving an up to 24-fold memory reduction on ImageNet-C. Code: https://***/mr-eggplant/FOA. Copyright 2024 by the author(s)

关键词： Image coding

来源：评论

学校读者我要写书评

暂无评论

A THEORETICAL PERSPECTIVE: HOW TO PREVENT MODEL COLLAPSE IN SELF-CONSUMING TRaiNING LOOPS

arXiv

引用

arXiv 2025年

作者： Fu, Shi Wang, Yingjie Chen, Yuzhu Tian, Xinmei Tao, Dacheng Generative AI Lab College of Computing and Data Science Nanyang Technological University Singapore639798 Singapore University of Science and Technology of China Hefei China

High-quality data is essential for training large generative models, yet the vast reservoir of real data available online has become nearly depleted. Consequently, models increasingly generate their own data for further training, forming Self-consuming Training Loops (STLs). However, the empirical results have been strikingly inconsistent: some models degrade or even collapse, while others successfully avoid these failures, leaving a significant gap in theoretical understanding to explain this discrepancy. This paper introduces the intriguing notion of recursive stability and presents the first theoretical generalization analysis, revealing how both model architecture and the proportion between real and synthetic data influence the success of STLs. We further extend this analysis to transformers in in-context learning, showing that even a constant-sized proportion of real data ensures convergence, while also providing insights into optimal synthetic data sizing. Copyright © 2025, The Authors. All rights reserved.

关键词： data assimilation

来源：评论

学校读者我要写书评

暂无评论

generative Adversarial Learning for Semi-supervised Retinal Layer Segmentation in OCT Images

Generative Adversarial Learning for Semi-supervised Retinal ...

引用

2024 IEEE-EMBS International Conference on Biomedical and Health Informatics, BHI 2024

作者： Lin Ong, Charlene Zhi Bakar Ali, Asad Abu Rajapakse, Jagath C Msd International GmbH Singapore Singapore College of Computing and Data Science Nanyang Technological University Health Informatics Lab Singapore Singapore

ISBN: (纸本)9798350351552

It is often challenging to obtain large number of labeled data for retinal layer segmentation in optical coherence tomography scans due to the need for expert ophthalmologists. On the other hand, huge quantities of unlabeled scans are often collected in medical centers. In this work, we propose a novel generative adversarial learning framework, GOctSeg, for semi-supervised retinal layer segmentation. GOctSeg consists of a U-Net generator, a discriminator, and a segmentor. The generation of synthetic images from synthetic labels is performed with the U - Net generator, which is used as input together with labeled B-scan patches into the Segmentor network for retinal layer boundary regression. We have also evaluated our methodology on two datasets, the data Resource for Multiple Sclerosis and Healthy Controls and the Duke university Diabetic Macular Edema dataset and demonstrated that our method outperformed or is comparable to other state-of-the-art methods with limited labels for boundary regression. Furthermore, we investigated the performance of GOctSeg on images with low signal-to-noise ratio or with blurred boundaries and showed that our methodology remained robust. Through ablation studies, we demonstrated the utility of synthetic labels for generative learning to guide in semi-supervised retinal layer segmentation. We envision that this methodology can be used to significantly reduce the effort required to obtain labels when there is label scarcity in clinical settings. © 2024 IEEE.

关键词： Ophthalmology

来源：评论

学校读者我要写书评

暂无评论

Self-Bootstrapping for Versatile Test-Time Adaptation

arXiv

引用

arXiv 2025年

作者： Niu, Shuaicheng Chen, Guohao Zhao, Peilin Wang, Tianyi Wu, Pengcheng Shen, Zhiqi College of Computing and Data Science Nanyang Technological University Singapore Tencent AI Lab Shenzhen China

In this paper, we seek to develop a versatile test-time adaptation (TTA) objective for a variety of tasks — classification and regression across image-, object-, and pixel-level predictions. We achieve this through a self-bootstrapping scheme that optimizes prediction consistency between the test image (as target) and its deteriorated view. The key challenge lies in devising effective augmentations/deteriorations that: i) preserve the image’s geometric information, e.g., object sizes and locations, which is crucial for TTA on object/pixel-level tasks, and ii) provide sufficient learning signals for TTA. To this end, we analyze how common distribution shifts affect the image’s information power across spatial frequencies in the Fourier domain, and reveal that low-frequency components carry high power and masking these components supplies more learning signals, while masking high-frequency components can not. In light of this, we randomly mask the low-frequency amplitude of an image in its Fourier domain for augmentation. Meanwhile, we also augment the image with noise injection to compensate for missing learning signals at high frequencies, by enhancing the information power there. Experiments show that, either independently or as a plug-and-play module, our method achieves superior results across classification, segmentation, and 3D monocular detection tasks with both transformer and CNN models. Code will be released. Copyright © 2025, The Authors. All rights reserved.

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：