Transfer_Learning
-
Resolution scaling governs DINOv3 transfer performance in chest radiograph classification
2025-10-08 16:25:04Soroosh Tayebi Arasteh, Mina Shaigan, Christiane Kuhl, Jakob Nikolas Kather, Sven Nebelung, Daniel TruhnAbstract
Self-supervised learning (SSL) has advanced visual representation learning, but its value in chest radiography, a high-volume imaging modality with fine-grained findings, remains unclear. Meta's DINOv3 extends earlier SSL models through Gram-anchored self-distillation. Whether these design choices improve transfer learning for chest radiography has not been systematically tested. We benchmarked DINOv3 against DINOv2 and ImageNet initialization across seven datasets (n>814,000). Two representative backbones were evaluated: ViT-B/16 and ConvNeXt-B. Images were analyzed at 224x224, 512x512, and 1024x1024 pixels. We additionally assessed frozen features from a 7B model. The primary outcome was mean AUROC across labels. At 224x224, DINOv3 and DINOv2 achieved comparable performance on adult datasets. Increasing resolution to 512x512 yielded consistent improvements for DINOv3 over both DINOv2 and ImageNet. In contrast, results in pediatric cohort showed no differences across initializations. Across all settings, ConvNeXt-B outperformed ViT-B/16. Models using frozen DINOv3-7B features underperformed relative to fully finetuned 86-89M-parameter backbones, highlighting the importance of domain adaptation. Scaling to 1024x1024 did not further improve accuracy. Resolution-related gains were most evident for boundary-dependent and small focal abnormalities. In chest radiography, higher input resolution is critical for leveraging the benefits of modern self-supervised models. 512x512 pixels represent a practical upper limit where DINOv3-initialized ConvNeXt-B networks provide the strongest performance, while larger inputs offer minimal return on cost. Clinically, these findings support use of finetuned, mid-sized backbones at 512x512 for chest radiograph interpretation, with the greatest gains expected in detecting subtle or boundary-centered lesions relevant to emergency and critical care settings.
Abstract (translated)
自监督学习（SSL）在视觉表征学习方面取得了进展，但在胸部X光摄影中的价值仍不清楚。胸部X光是一种高容量的影像模式，具有细微发现的特点。Meta的DINOv3通过Gram-锚定自我蒸馏扩展了早期的SSL模型。这些设计选择是否改善了胸部X光摄影中的迁移学习尚未被系统地测试过。我们在七个数据集（n>814,000）上对DINOv3与DINOv2和ImageNet初始化进行了基准比较。我们评估了两个代表性的骨干网络：ViT-B/16和ConvNeXt-B。图像在224x224、512x512和1024x1024像素下进行分析。我们还评估了来自7B模型的冻结特征。主要结果指标是标签的平均AUROC。 在224x224分辨率下，DINOv3与DINOv2在成人数据集上实现了相当的表现。增加到512x512像素后，DINOv3相对于DINOv2和ImageNet的一致改进变得更加明显。相比之下，在儿科群体中，不同初始化之间没有差异。在所有设置下，ConvNeXt-B优于ViT-B/16。使用冻结的DINOv3-7B特征的模型相对于完全微调后的86-89M参数骨干网络表现较弱，这突显了领域适应的重要性。分辨率增加到1024x1024后，并未进一步提高准确性。与边界相关的和小焦点异常的变化在分辨率相关增益中最明显。 在胸部X光摄影中，更高的输入分辨率对于利用现代自监督模型的好处至关重要。512x512像素代表了一个实际的上限，在这个分辨率下，由DINOv3初始化的ConvNeXt-B网络表现最佳，而更大的输入则无法提供额外的成本效益。临床上，这些发现支持使用在512x512上进行微调、中等大小骨干网络用于胸部X光片解释，并且预期在此类病变检测中的改进最为显著，特别是在急诊和重症监护环境中相关的细微或边界中心化病灶。
URL
https://arxiv.org/abs/2510.07191
PDF
-
TransFIRA: Transfer Learning for Face Image Recognizability Assessment
2025-10-07 18:16:21Allen Tu, Kartik Narayan, Joshua Gleason, Jennifer Xu, Matthew Meyn, Tom Goldstein, Vishal M. PatelAbstract
Face recognition in unconstrained environments such as surveillance, video, and web imagery must contend with extreme variation in pose, blur, illumination, and occlusion, where conventional visual quality metrics fail to predict whether inputs are truly recognizable to the deployed encoder. Existing FIQA methods typically rely on visual heuristics, curated annotations, or computationally intensive generative pipelines, leaving their predictions detached from the encoder's decision geometry. We introduce TransFIRA (Transfer Learning for Face Image Recognizability Assessment), a lightweight and annotation-free framework that grounds recognizability directly in embedding space. TransFIRA delivers three advances: (i) a definition of recognizability via class-center similarity (CCS) and class-center angular separation (CCAS), yielding the first natural, decision-boundary--aligned criterion for filtering and weighting; (ii) a recognizability-informed aggregation strategy that achieves state-of-the-art verification accuracy on BRIAR and IJB-C while nearly doubling correlation with true recognizability, all without external labels, heuristics, or backbone-specific training; and (iii) new extensions beyond faces, including encoder-grounded explainability that reveals how degradations and subject-specific factors affect recognizability, and the first recognizability-aware body recognition assessment. Experiments confirm state-of-the-art results on faces, strong performance on body recognition, and robustness under cross-dataset shifts. Together, these contributions establish TransFIRA as a unified, geometry-driven framework for recognizability assessment -- encoder-specific, accurate, interpretable, and extensible across modalities -- significantly advancing FIQA in accuracy, explainability, and scope.
Abstract (translated)
在不受限制的环境中，如监控、视频和网络图像中的面部识别，必须应对姿态变化、模糊度、光照变化以及遮挡等极端情况。传统视觉质量指标无法预测输入是否对部署的编码器来说真正具有可识性。现有的FIQA（Face Image Recognizability Assessment）方法通常依赖于视觉直觉、手工标注或计算密集型生成流程，导致其预测结果与编码器决策几何无关。 我们提出了一种名为TransFIRA（Transfer Learning for Face Image Recognizability Assessment）的轻量级无标注框架，该框架直接在嵌入空间中定义可识性。TransFIRA实现了三项创新： (i) 通过类中心相似度（CCS）和类中心角度分离（CCAS），为过滤和加权提供了第一个自然且与决策边界对齐的标准。 (ii) 提出了一种基于可识别性的聚合策略，该策略在BRIAR和IJB-C数据集上实现了最先进的验证准确率，并将与真实可识性相关的相关性几乎翻倍。这些结果均无需外部标签、直觉或针对特定骨干网络的训练。 (iii) TransFIRA扩展了面部以外的应用，包括基于编码器解释性的方法，揭示了退化和主体特有因素如何影响可识别性，并提出了首个具有可识性意识的身体识别评估。 实验表明，在面部数据集中TransFIRA达到了最先进的结果，在身体识别方面表现出色，并且在跨数据集迁移时也表现稳健。这些贡献共同确立了TransFIRA作为统一、几何驱动的框架，用于可识别性评估：针对特定编码器，准确，可解释，且具有跨模式扩展的能力。这显著提高了FIQA的准确性、可解释性和应用范围。
URL
https://arxiv.org/abs/2510.06353
PDF
-
Scalable deep fusion of spaceborne lidar and synthetic aperture radar for global forest structural complexity mapping
2025-10-07 14:29:37Tiago de Conto, John Armston, Ralph DubayahAbstract
Forest structural complexity metrics integrate multiple canopy attributes into a single value that reflects habitat quality and ecosystem function. Spaceborne lidar from the Global Ecosystem Dynamics Investigation (GEDI) has enabled mapping of structural complexity in temperate and tropical forests, but its sparse sampling limits continuous high-resolution mapping. We present a scalable, deep learning framework fusing GEDI observations with multimodal Synthetic Aperture Radar (SAR) datasets to produce global, high-resolution (25 m) wall-to-wall maps of forest structural complexity. Our adapted EfficientNetV2 architecture, trained on over 130 million GEDI footprints, achieves high performance (global R2 = 0.82) with fewer than 400,000 parameters, making it an accessible tool that enables researchers to process datasets at any scale without requiring specialized computing infrastructure. The model produces accurate predictions with calibrated uncertainty estimates across biomes and time periods, preserving fine-scale spatial patterns. It has been used to generate a global, multi-temporal dataset of forest structural complexity from 2015 to 2022. Through transfer learning, this framework can be extended to predict additional forest structural variables with minimal computational cost. This approach supports continuous, multi-temporal monitoring of global forest structural dynamics and provides tools for biodiversity conservation and ecosystem management efforts in a changing climate.
Abstract (translated)
森林结构复杂性指标将多种冠层特征整合为单一数值，以反映栖息地质量和生态系统功能。来自全球生态动力学调查（GEDI）的卫星激光雷达使温带和热带森林中的结构复杂性的测绘成为可能，但其稀疏采样限制了连续高分辨率地图的生成。我们提出了一种可扩展的深度学习框架，该框架融合了GEDI观测数据与多模态合成孔径雷达（SAR）数据集，以生成全球范围内、分辨率为25米的森林结构复杂性的完整墙体到墙体的地图。 我们的改进版EfficientNetV2架构，在超过1.3亿个GEDI足迹上进行了训练，实现了卓越性能（全球R²=0.82），同时参数数量少于40万个。这使得该模型成为一种易于使用的工具，研究人员可以在此框架下处理任何规模的数据集，而无需特别的计算基础设施。 此模型能够生成跨生物群落和时间跨度准确预测，并提供校准后的不确定性估计，保留了细微的空间模式。它已被用来创建2015年至2022年间全球范围内的森林结构复杂性多时间序列数据集。 通过迁移学习，这一框架可以以最小的计算成本扩展到预测其他森林结构变量。这种方法支持对全球森林结构动态进行持续、多时段监测，并为气候变化条件下的生物多样性保护和生态系统管理提供了工具。
URL
https://arxiv.org/abs/2510.06299
PDF
-
BioAutoML-NAS: An End-to-End AutoML Framework for Multimodal Insect Classification via Neural Architecture Search on Large-Scale Biodiversity Data
2025-10-07 13:00:12Arefin Ittesafun Abian, Debopom Sutradhar, Md Rafi Ur Rashid, Reem E. Mohamed, Md Rafiqul Islam, Asif Karim, Kheng Cher Yeo, Sami AzamAbstract
Insect classification is important for agricultural management and ecological research, as it directly affects crop health and production. However, this task remains challenging due to the complex characteristics of insects, class imbalance, and large-scale datasets. To address these issues, we propose BioAutoML-NAS, the first BioAutoML model using multimodal data, including images, and metadata, which applies neural architecture search (NAS) for images to automatically learn the best operations for each connection within each cell. Multiple cells are stacked to form the full network, each extracting detailed image feature representations. A multimodal fusion module combines image embeddings with metadata, allowing the model to use both visual and categorical biological information to classify insects. An alternating bi-level optimization training strategy jointly updates network weights and architecture parameters, while zero operations remove less important connections, producing sparse, efficient, and high-performing architectures. Extensive evaluation on the BIOSCAN-5M dataset demonstrates that BioAutoML-NAS achieves 96.81% accuracy, 97.46% precision, 96.81% recall, and a 97.05% F1 score, outperforming state-of-the-art transfer learning, transformer, AutoML, and NAS methods by approximately 16%, 10%, and 8% respectively. Further validation on the Insects-1M dataset obtains 93.25% accuracy, 93.71% precision, 92.74% recall, and a 93.22% F1 score. These results demonstrate that BioAutoML-NAS provides accurate, confident insect classification that supports modern sustainable farming.
Abstract (translated)
昆虫分类对于农业管理和生态研究至关重要，因为它直接影响作物的健康和产量。然而，由于昆虫具有复杂的特征、类不平衡以及大规模的数据集等问题，这一任务仍然极具挑战性。为了解决这些问题，我们提出了BioAutoML-NAS，这是第一个使用多模态数据（包括图像和元数据）的BioAutoML模型，并应用神经架构搜索（NAS）对图像进行自动学习，以确定每个连接中的最佳操作。多个细胞被堆叠起来形成完整的网络，每个细胞提取详细的图像特征表示。一个多模态融合模块结合了图像嵌入与元数据，使模型能够利用视觉和分类生物信息来分类昆虫。一种交替的双层优化训练策略共同更新网络权重和架构参数，并通过去除不重要的连接以生成稀疏、高效且性能高的架构。 在BIOSCAN-5M数据集上的广泛评估表明，BioAutoML-NAS实现了96.81%的准确率、97.46%的精确度、96.81%的召回率和97.05%的F1值，分别比最先进的迁移学习、Transformer模型、AutoML方法及NAS方法高出约16%、10%和8%。在Insects-1M数据集上的进一步验证获得了93.25%的准确率、93.71%的精确度、92.74%的召回率以及93.22%的F1值。 这些结果表明，BioAutoML-NAS能够提供准确且可靠的昆虫分类，支持现代可持续农业。
URL
https://arxiv.org/abs/2510.05888
PDF
-
Fusion-Based Neural Generalization for Predicting Temperature Fields in Industrial PET Preform Heating
2025-10-06 21:38:37Ahmad Alsheikh, Andreas FischerAbstract
Accurate and efficient temperature prediction is critical for optimizing the preheating process of PET preforms in industrial microwave systems prior to blow molding. We propose a novel deep learning framework for generalized temperature prediction. Unlike traditional models that require extensive retraining for each material or design variation, our method introduces a data-efficient neural architecture that leverages transfer learning and model fusion to generalize across unseen scenarios. By pretraining specialized neural regressor on distinct conditions such as recycled PET heat capacities or varying preform geometries and integrating their representations into a unified global model, we create a system capable of learning shared thermal dynamics across heterogeneous inputs. The architecture incorporates skip connections to enhance stability and prediction accuracy. Our approach reduces the need for large simulation datasets while achieving superior performance compared to models trained from scratch. Experimental validation on two case studies material variability and geometric diversity demonstrates significant improvements in generalization, establishing a scalable ML-based solution for intelligent thermal control in manufacturing environments. Moreover, the approach highlights how data-efficient generalization strategies can extend to other industrial applications involving complex physical modeling with limited data.
Abstract (translated)
准确且高效的温度预测对于在工业微波系统中优化聚对苯二甲酸乙二醇酯（PET）预成型品吹塑前的预热过程至关重要。我们提出了一种新的深度学习框架，用于通用温度预测。与传统的需要针对每种材料或设计变化进行大量重新训练的模型不同，我们的方法引入了一个数据效率高的神经架构，利用迁移学习和模型融合来适应未见过的情景。通过在不同的条件下对专业化的神经回归器进行预训练（例如回收PET的热容量或不同的预成型几何形状），并将它们的表现形式整合到一个统一的整体模型中，我们创建了一种能够在不同输入情况下学习共享热动力学的系统。该架构包括跳过连接以增强稳定性和预测准确性。 我们的方法在减少大型模拟数据集需求的同时，实现了比从头开始训练的模型更优的性能。通过对两种案例研究（材料变化和几何多样性）进行实验验证，结果表明，在泛化能力方面有显著改进，从而为制造环境中的智能热控制提供了一种可扩展的基于机器学习的解决方案。此外，该方法展示了如何将数据效率高的通用策略应用于其他涉及复杂物理建模且数据有限的工业应用中。
URL
https://arxiv.org/abs/2510.05394
PDF
-
Fine-Tuned CNN-Based Approach for Multi-Class Mango Leaf Disease Detection
2025-10-06 19:47:26Jalal Ahmmed, Faruk Ahmed, Rashedul Hasan Shohan, Md. Mahabub Rana, Mahdi HasanAbstract
Mango is an important fruit crop in South Asia, but its cultivation is frequently hampered by leaf diseases that greatly impact yield and quality. This research examines the performance of five pre-trained convolutional neural networks, DenseNet201, InceptionV3, ResNet152V2, SeResNet152, and Xception, for multi-class identification of mango leaf diseases across eight classes using a transfer learning strategy with fine-tuning. The models were assessed through standard evaluation metrics, such as accuracy, precision, recall, F1-score, and confusion matrices. Among the architectures tested, DenseNet201 delivered the best results, achieving 99.33% accuracy with consistently strong metrics for individual classes, particularly excelling in identifying Cutting Weevil and Bacterial Canker. Moreover, ResNet152V2 and SeResNet152 provided strong outcomes, whereas InceptionV3 and Xception exhibited lower performance in visually similar categories like Sooty Mould and Powdery Mildew. The training and validation plots demonstrated stable convergence for the highest-performing models. The capability of fine-tuned transfer learning models, for precise and dependable multi-class mango leaf disease detection in intelligent agricultural applications.
Abstract (translated)
芒果是南亚地区重要的水果作物，但其种植经常受到叶病的影响，这些疾病严重影响了产量和质量。这项研究评估了五种预训练的卷积神经网络（CNN）——DenseNet201、InceptionV3、ResNet152V2、SeResNet152 和 Xception，在使用迁移学习策略进行微调的情况下，这五种模型在八个类别的芒果叶病多分类识别中的表现。通过标准评估指标如准确率、精确度、召回率、F1值和混淆矩阵对这些模型进行了性能评估。 在所有测试的架构中，DenseNet201的表现最佳，在各分类上均取得了99.33%的准确性，并且在识别割叶甲（Cutting Weevil）和细菌性溃疡病（Bacterial Canker）方面尤为突出。此外，ResNet152V2 和 SeResNet152 也提供了强大的结果，而 InceptionV3 和 Xception 在视觉上相似类别如煤污病（Sooty Mould）和白粉病（Powdery Mildew）的识别中表现较差。 训练和验证图表明了最佳模型在收敛过程中的稳定性。这项研究表明，微调后的迁移学习模型具有在智能农业应用中进行精确且可靠的多分类芒果叶病检测的能力。
URL
https://arxiv.org/abs/2510.05326
PDF
-
Exploring the Efficacy of Modified Transfer Learning in Identifying Parkinson's Disease Through Drawn Image Patterns
2025-10-06 16:55:07Nabil Daiyan, Md Rakibul HaqueAbstract
Parkinson's disease (PD) is a progressive neurodegenerative condition characterized by the death of dopaminergic neurons, leading to various movement disorder symptoms. Early diagnosis of PD is crucial to prevent adverse effects, yet traditional diagnostic methods are often cumbersome and costly. In this study, a machine learning-based approach is proposed using hand-drawn spiral and wave images as potential biomarkers for PD detection. Our methodology leverages convolutional neural networks (CNNs), transfer learning, and attention mechanisms to improve model performance and resilience against overfitting. To enhance the diversity and richness of both spiral and wave categories, the training dataset undergoes augmentation to increase the number of images. The proposed architecture comprises three phases: utilizing pre-trained CNNs, incorporating custom convolutional layers, and ensemble voting. Employing hard voting further enhances performance by aggregating predictions from multiple models. Experimental results show promising accuracy rates. For spiral images, weighted average precision, recall, and F1-score are 90%, and for wave images, they are 96.67%. After combining the predictions through ensemble hard voting, the overall accuracy is 93.3%. These findings underscore the potential of machine learning in early PD diagnosis, offering a non-invasive and cost-effective solution to improve patient outcomes.
Abstract (translated)
帕金森病（PD）是一种进行性的神经退行性疾病，其特点是多巴胺能神经元的死亡，导致各种运动障碍症状。早期诊断帕金森病对于防止不良后果至关重要，然而传统的诊断方法往往繁琐且昂贵。在这项研究中，提出了一种基于机器学习的方法，使用手绘螺旋和波形图像作为潜在生物标志物来检测PD。我们的方法利用卷积神经网络（CNN）、迁移学习以及注意力机制来提高模型性能并增强其对过拟合的抵抗能力。为了增加训练数据集的多样性和丰富性，通过数据扩充技术增加了图像的数量。 所提出的架构包括三个阶段：使用预训练的CNN、添加自定义卷积层和集成投票。利用硬投票进一步提高了性能，通过汇总多个模型的预测结果来实现。实验结果显示了令人鼓舞的准确率。对于螺旋图像，加权平均精确度、召回率和F1分数分别为90%，而对于波形图像则为96.67%。经过集成硬投票组合预测后，整体准确性达到了93.3%。 这些发现强调了机器学习在早期PD诊断中的潜力，提供了一种非侵入性且成本效益高的解决方案，以改善患者的预后情况。
URL
https://arxiv.org/abs/2510.05015
PDF
-
Comparative Analysis of YOLOv5, Faster R-CNN, SSD, and RetinaNet for Motorbike Detection in Kigali Autonomous Driving Context
2025-10-06 15:26:08Ngeyen Yinkfu, Sunday Nwovu, Jonathan Kayizzi, Angelique UwamahoroAbstract
In Kigali, Rwanda, motorcycle taxis are a primary mode of transportation, often navigating unpredictably and disregarding traffic rules, posing significant challenges for autonomous driving systems. This study compares four object detection models--YOLOv5, Faster R-CNN, SSD, and RetinaNet--for motorbike detection using a custom dataset of 198 images collected in Kigali. Implemented in PyTorch with transfer learning, the models were evaluated for accuracy, localization, and inference speed to assess their suitability for real-time navigation in resource-constrained settings. We identify implementation challenges, including dataset limitations and model complexities, and recommend simplified architectures for future work to enhance accessibility for autonomous systems in developing countries like Rwanda.
Abstract (translated)
在卢旺达的基加利，摩托车出租车是主要的交通工具之一，它们常常不遵守交通规则且行驶路径难以预测，这对自动驾驶系统构成了重大挑战。这项研究比较了四种物体检测模型——YOLOv5、Faster R-CNN、SSD 和 RetinaNet，在使用从基加利收集的包含198张图片的自定义数据集的情况下，用于摩托车检测的效果。这些模型在 PyTorch 中实现，并采用了迁移学习技术，根据准确性、定位能力和推理速度对它们进行了评估，以确定它们在资源受限环境中的实时导航适用性。我们识别了实施过程中的挑战，包括数据集限制和模型复杂度问题，并建议在未来的工作中采用简化架构，以便于在像卢旺达这样的发展中国家提高自动驾驶系统的可访问性。
URL
https://arxiv.org/abs/2510.04912
PDF
-
From Actions to Kinesics: Extracting Human Psychological States through Bodily Movements
2025-10-06 14:31:53Cheyu Lin, Katherine A. FlaniganAbstract
Understanding the dynamic relationship between humans and the built environment is a key challenge in disciplines ranging from environmental psychology to reinforcement learning (RL). A central obstacle in modeling these interactions is the inability to capture human psychological states in a way that is both generalizable and privacy preserving. Traditional methods rely on theoretical models or questionnaires, which are limited in scope, static, and labor intensive. We present a kinesics recognition framework that infers the communicative functions of human activity -- known as kinesics -- directly from 3D skeleton joint data. Combining a spatial-temporal graph convolutional network (ST-GCN) with a convolutional neural network (CNN), the framework leverages transfer learning to bypass the need for manually defined mappings between physical actions and psychological categories. The approach preserves user anonymity while uncovering latent structures in bodily movements that reflect cognitive and emotional states. Our results on the Dyadic User EngagemenT (DUET) dataset demonstrate that this method enables scalable, accurate, and human-centered modeling of behavior, offering a new pathway for enhancing RL-driven simulations of human-environment interaction.
Abstract (translated)
理解人类与人造环境之间的动态关系是包括环境心理学和强化学习（RL）在内的多个学科的关键挑战。在建模这些互动时，一个核心障碍在于无法以既具普适性又保护隐私的方式捕捉人类的心理状态。传统方法依赖于理论模型或问卷调查，这些方式在范围、静态性和劳动强度方面都有局限。 我们提出了一种基于动作识别的框架，该框架直接从三维骨骼关节数据中推断出人类活动的交流功能——即动情学（kinesics）。这一框架结合了空间-时间图卷积网络（ST-GCN）和卷积神经网络（CNN），利用迁移学习来规避手动定义的身体行动与心理类别之间映射的需求。该方法在保持用户匿名性的同时，揭示出反映认知和情绪状态的肢体运动中的潜在结构。 我们在二元用户参与数据集（DUET）上的实验结果表明，这种方法能够实现行为的大规模、准确且以人为中心建模，为强化学习驱动的人类-环境互动仿真提供了新的途径。
URL
https://arxiv.org/abs/2510.04844
PDF
-
A Comparative Study of Vision Transformers and CNNs for Few-Shot Rigid Transformation and Fundamental Matrix Estimation
2025-10-06 13:18:27Alon Kaya, Igal Bilik, Inna StainvasAbstract
Vision-transformers (ViTs) and large-scale convolution-neural-networks (CNNs) have reshaped computer vision through pretrained feature representations that enable strong transfer learning for diverse tasks. However, their efficiency as backbone architectures for geometric estimation tasks involving image deformations in low-data regimes remains an open question. This work considers two such tasks: 1) estimating 2D rigid transformations between pairs of images and 2) predicting the fundamental matrix for stereo image pairs, an important problem in various applications, such as autonomous mobility, robotics, and 3D scene reconstruction. Addressing this intriguing question, this work systematically compares large-scale CNNs (ResNet, EfficientNet, CLIP-ResNet) with ViT-based foundation models (CLIP-ViT variants and DINO) in various data size settings, including few-shot scenarios. These pretrained models are optimized for classification or contrastive learning, encouraging them to focus mostly on high-level semantics. The considered tasks require balancing local and global features differently, challenging the straightforward adoption of these models as the backbone. Empirical comparative analysis shows that, similar to training from scratch, ViTs outperform CNNs during refinement in large downstream-data scenarios. However, in small data scenarios, the inductive bias and smaller capacity of CNNs improve their performance, allowing them to match that of a ViT. Moreover, ViTs exhibit stronger generalization in cross-domain evaluation where the data distribution changes. These results emphasize the importance of carefully selecting model architectures for refinement, motivating future research towards hybrid architectures that balance local and global representations.
Abstract (translated)
视觉变换器（ViTs）和大规模卷积神经网络（CNNs）通过预训练的特征表示，彻底改变了计算机视觉领域，为各种任务提供了强大的迁移学习能力。然而，在数据量较少的情况下，作为涉及图像变形的几何估计任务的骨干架构，它们的效率仍然是一个开放性问题。这项工作考虑了两个这样的任务：1) 估计图像对之间的二维刚体变换；2) 预测立体图像对的基本矩阵，这是一个在自动驾驶、机器人技术和三维场景重建等多种应用中都很重要的问题。为了解决这个引人入胜的问题，本研究系统地比较了大规模CNN（如ResNet、EfficientNet和CLIP-ResNet）与基于ViT的基础模型（如CLIP-ViT变体和DINO），在各种数据规模设置下进行对比分析，包括少量样本场景。这些预训练的模型是为分类或对比学习优化的，鼓励它们主要关注高层次语义信息。然而，所考虑的任务需要平衡局部特征和全局特征，这挑战了直接采用这些模型作为骨干架构的做法。 实验性的比较分析表明，在大量下游数据场景中进行微调时，ViT的表现优于CNN，这一点与从头开始训练的情况相似。但在小规模数据场景下，CNN的归纳偏置以及较小的容量提高了它们的表现，使其能够匹敌ViT的表现。此外，在跨领域评估中（即数据分布发生变化），ViTs表现出更强的泛化能力。这些结果强调了在微调时谨慎选择模型架构的重要性，并激励未来的研究向平衡局部和全局表示的混合架构发展。
URL
https://arxiv.org/abs/2510.04794
PDF
-
Beyond Appearance: Transformer-based Person Identification from Conversational Dynamics
2025-10-06 12:31:15Masoumeh Chapariniya, Teodora Vukovic, Sarah Ebling, Volker DellwoAbstract
This paper investigates the performance of transformer-based architectures for person identification in natural, face-to-face conversation scenario. We implement and evaluate a two-stream framework that separately models spatial configurations and temporal motion patterns of 133 COCO WholeBody keypoints, extracted from a subset of the CANDOR conversational corpus. Our experiments compare pre-trained and from-scratch training, investigate the use of velocity features, and introduce a multi-scale temporal transformer for hierarchical motion modeling. Results demonstrate that domain-specific training significantly outperforms transfer learning, and that spatial configurations carry more discriminative information than temporal dynamics. The spatial transformer achieves 95.74% accuracy, while the multi-scale temporal transformer achieves 93.90%. Feature-level fusion pushes performance to 98.03%, confirming that postural and dynamic information are complementary. These findings highlight the potential of transformer architectures for person identification in natural interactions and provide insights for future multimodal and cross-cultural studies.
Abstract (translated)
本文研究了基于Transformer架构的人识别性能，特别是在自然的面对面对话场景中。我们实现并评估了一个两流框架，该框架分别对从CANDOR对话语料库的一个子集中提取的133个COCO WholeBody关键点的空间配置和时间运动模式进行建模。我们的实验比较了预训练和从头开始训练的效果，探讨了使用速度特征的可能性，并引入了一种多尺度时间Transformer来进行层次化运动建模。实验结果表明，特定领域的训练显著优于迁移学习，并且空间配置比时间动态包含更多的判别性信息。在空间Transformer中实现了95.74%的准确率，而多尺度时间Transformer则达到了93.90%。特征级融合将性能提升至98.03%，证实了姿势和动态信息是互补的。这些发现突显了基于Transformer架构的人识别在自然互动中的潜力，并为未来跨模态和跨文化研究提供了见解。
URL
https://arxiv.org/abs/2510.04753
PDF
-
A Multilingual Framework for Dysarthria: Detection, Severity Classification, Speech-to-Text, and Clean Speech Generation
2025-10-05 00:52:04Ananya Raghu, Anisha Raghu, Nithika Vivek, Sofie Budman, Omar MansourAbstract
Dysarthria is a motor speech disorder that results in slow and often incomprehensible speech. Speech intelligibility significantly impacts communication, leading to barriers in social interactions. Dysarthria is often a characteristic of neurological diseases including Parkinson's and ALS, yet current tools lack generalizability across languages and levels of severity. In this study, we present a unified AI-based multilingual framework that addresses six key components: (1) binary dysarthria detection, (2) severity classification, (3) clean speech generation, (4) speech-to-text conversion, (5) emotion detection, and (6) voice cloning. We analyze datasets in English, Russian, and German, using spectrogram-based visualizations and acoustic feature extraction to inform model training. Our binary detection model achieved 97% accuracy across all three languages, demonstrating strong generalization across languages. The severity classification model also reached 97% test accuracy, with interpretable results showing model attention focused on lower harmonics. Our translation pipeline, trained on paired Russian dysarthric and clean speech, reconstructed intelligible outputs with low training (0.03) and test (0.06) L1 losses. Given the limited availability of English dysarthric-clean pairs, we fine-tuned the Russian model on English data and achieved improved losses of 0.02 (train) and 0.03 (test), highlighting the promise of cross-lingual transfer learning for low-resource settings. Our speech-to-text pipeline achieved a Word Error Rate of 0.1367 after three epochs, indicating accurate transcription on dysarthric speech and enabling downstream emotion recognition and voice cloning from transcribed speech. Overall, the results and products of this study can be used to diagnose dysarthria and improve communication and understanding for patients across different languages.
Abstract (translated)
构音障碍是一种运动性言语障碍，导致缓慢且常难以理解的说话。言语清晰度对沟通有着重大影响，可能导致社交互动中的障碍。构音障碍常常是帕金森病和肌萎缩侧索硬化症等神经疾病的一个特征，然而目前的工具在跨语言和不同严重程度上缺乏普适性。 在这项研究中，我们提出了一种基于人工智能的多语种统一框架，涵盖六个关键组成部分：（1）构音障碍二元检测，（2）严重度分类，（3）清晰语音生成，（4）语音转文字转换，（5）情感识别和（6）声音克隆。我们在英语、俄语和德语的数据集上进行了分析，并使用基于频谱图的可视化以及声学特征提取来指导模型训练。 我们的二元检测模型在所有三种语言中均达到了97%的准确率，显示出良好的跨语言泛化能力。严重度分类模型也实现了97%的测试准确率，解释性结果表明模型注意力集中在较低谐波上。我们基于俄语构音障碍和清晰语音成对训练的语言翻译管道，在低训练（0.03）和测试（0.06）L1损失的情况下生成了可理解的输出。鉴于英语构音障碍-清晰对的数量有限，我们在英语数据集上微调了俄罗斯模型，并取得了改进后的训练（0.02）和测试（0.03）损失结果，这突显了跨语言迁移学习在资源匮乏环境中的潜力。 我们的语音转文字管道在三个周期后达到了0.1367的单词错误率，表明对构音障碍言语实现了准确转录，并使下游的情感识别和声音克隆成为可能。总体而言，本研究的结果和产品可用于诊断构音障碍并改善跨语言患者的沟通与理解。 这项研究表明了多语种统一框架在解决构音障碍相关问题上的潜力，特别是在资源有限的语言环境中，为提高患者的生活质量提供了新的方向和技术支持。
URL
https://arxiv.org/abs/2510.03986
PDF
-
Multi-Modal Oral Cancer Detection Using Weighted Ensemble Convolutional Neural Networks
2025-10-04 17:06:46Ajo Babu George, Sreehari J R Ajo Babu George, Sreehari J R Ajo Babu George, Sreehari J RAbstract
Aims Late diagnosis of Oral Squamous Cell Carcinoma (OSCC) contributes significantly to its high global mortality rate, with over 50\% of cases detected at advanced stages and a 5-year survival rate below 50\% according to WHO statistics. This study aims to improve early detection of OSCC by developing a multimodal deep learning framework that integrates clinical, radiological, and histopathological images using a weighted ensemble of DenseNet-121 convolutional neural networks (CNNs). Material and Methods A retrospective study was conducted using publicly available datasets representing three distinct medical imaging modalities. Each modality-specific dataset was used to train a DenseNet-121 CNN via transfer learning. Augmentation and modality-specific preprocessing were applied to increase robustness. Predictions were fused using a validation-weighted ensemble strategy. Evaluation was performed using accuracy, precision, recall, F1-score. Results High validation accuracy was achieved for radiological (100\%) and histopathological (95.12\%) modalities, with clinical images performing lower (63.10\%) due to visual heterogeneity. The ensemble model demonstrated improved diagnostic robustness with an overall accuracy of 84.58\% on a multimodal validation dataset of 55 samples. Conclusion The multimodal ensemble framework bridges gaps in the current diagnostic workflow by offering a non-invasive, AI-assisted triage tool that enhances early identification of high-risk lesions. It supports clinicians in decision-making, aligning with global oncology guidelines to reduce diagnostic delays and improve patient outcomes.
Abstract (translated)
口腔鳞状细胞癌（OSCC）晚期诊断率高是其全球死亡率高的主要原因，据世界卫生组织统计，超过50%的病例在晚期被发现，五年生存率低于50%。本研究旨在通过开发一种多模态深度学习框架来提高OSCC早期检测能力，该框架整合了临床、放射学和病理图像，并使用DenseNet-121卷积神经网络（CNN）的加权集成。 **材料与方法** 进行了一项回顾性研究，利用公开可用的数据集代表三种不同的医学影像模态。每个特定模态的数据集通过迁移学习训练了一个DenseNet-121 CNN。为了提高鲁棒性，应用了数据增强和模态特异性预处理。预测结果采用验证加权集成策略进行融合。使用准确性、精确度、召回率、F1分数进行评估。 **结果** 放射学（100%）和病理图像（95.12%）的高验证准确率得到了实现，而临床图像的表现较低（63.10%），这是由于视觉异质性导致的。整个多模态验证数据集包含55个样本时，集成模型在诊断鲁棒性方面有所提高，总体准确率为84.58%。 **结论** 该多模态集成框架通过提供一种非侵入性的、人工智能辅助的初步筛查工具来弥合当前诊断工作流程中的差距，有助于早期识别高风险病灶。它支持临床医生决策，并符合全球肿瘤学指南，减少诊断延迟，改善患者预后。
URL
https://arxiv.org/abs/2510.03878
PDF
-
Searching for the Most Human-like Emergent Language
Abstract
In this paper, we design a signalling game-based emergent communication environment to generate state-of-the-art emergent languages in terms of similarity to human language. This is done with hyperparameter optimization, using XferBench as the objective function. XferBench quantifies the statistical similarity of emergent language to human language by measuring its suitability for deep transfer learning to human language. Additionally, we demonstrate the predictive power of entropy on the transfer learning performance of emergent language as well as corroborate previous results on the entropy-minimization properties of emergent communication systems. Finally, we report generalizations regarding what hyperparameters produce more realistic emergent languages, that is, ones which transfer better to human language.
Abstract (translated)
在这篇论文中，我们设计了一个基于信号博弈的新兴通信环境，以生成在与人类语言相似性方面处于行业前沿的新兴语言。这是通过超参数优化实现的，其中使用XferBench作为目标函数。XferBench 通过衡量新兴语言进行深度迁移学习到人类语言的适用性来量化其与人类语言的统计相似度。此外，我们展示了熵在新兴语言迁移学习性能中的预测能力，并且证实了关于新兴通信系统最小化熵特性的先前结果。最后，我们报告了哪些超参数可以生成更接近现实世界的新兴语言的一般结论，即那些更好地迁移到人类语言的语言。
URL
https://arxiv.org/abs/2510.03467
PDF
-
InsideOut: An EfficientNetV2-S Based Deep Learning Framework for Robust Multi-Class Facial Emotion Recognition
2025-10-03 14:53:47Ahsan Farabi, Israt Khandaker, Ibrahim Khalil Shanto, Md Abdul Ahad Minhaz, Tanisha ZamanAbstract
Facial Emotion Recognition (FER) is a key task in affective computing, enabling applications in human-computer interaction, e-learning, healthcare, and safety systems. Despite advances in deep learning, FER remains challenging due to occlusions, illumination and pose variations, subtle intra-class differences, and dataset imbalance that hinders recognition of minority emotions. We present InsideOut, a reproducible FER framework built on EfficientNetV2-S with transfer learning, strong data augmentation, and imbalance-aware optimization. The approach standardizes FER2013 images, applies stratified splitting and augmentation, and fine-tunes a lightweight classification head with class-weighted loss to address skewed distributions. InsideOut achieves 62.8% accuracy with a macro averaged F1 of 0.590 on FER2013, showing competitive results compared to conventional CNN baselines. The novelty lies in demonstrating that efficient architectures, combined with tailored imbalance handling, can provide practical, transparent, and reproducible FER solutions.
Abstract (translated)
面部情绪识别（FER）是情感计算中的一个关键任务，它能够支持人机交互、电子学习、医疗保健和安全系统等应用。尽管深度学习技术取得了进展，但由于遮挡、光照变化和姿势变化等因素的影响，以及细微的类内差异和数据集不平衡问题的存在，使得FER仍然具有挑战性，这些问题阻碍了少数情绪的识别。 我们提出了InsideOut框架，这是一个基于EfficientNetV2-S架构并通过迁移学习、强大的数据增强技术和针对不平衡处理优化方法构建的情绪识别可重复研究框架。该方法对FER2013数据集中的图像进行标准化处理，并采用了分层拆分和增强技术，同时通过对轻量级分类头采用类别加权损失进行微调来解决偏斜分布的问题。 InsideOut在FER2013数据集上实现了62.8%的准确率以及0.590的宏平均F1分数，与传统的卷积神经网络基准相比表现出了竞争力。其创新之处在于展示了高效架构结合专门处理不平衡的方法可以提供实用、透明且可重复的情绪识别解决方案。 该框架和方法对于在资源受限环境下提高FER性能具有重要意义，并为未来的研究提供了坚实的基础。
URL
https://arxiv.org/abs/2510.03066
PDF
-
From high-frequency sensors to noon reports: Using transfer learning for shaft power prediction in maritime
2025-10-03 13:47:16Akriti Sharma, Dogan Altan, Dusica Marijan, Arnbj{\o}rn MaressaAbstract
With the growth of global maritime transportation, energy optimization has become crucial for reducing costs and ensuring operational efficiency. Shaft power is the mechanical power transmitted from the engine to the shaft and directly impacts fuel consumption, making its accurate prediction a paramount step in optimizing vessel performance. Power consumption is highly correlated with ship parameters such as speed and shaft rotation per minute, as well as weather and sea conditions. Frequent access to this operational data can improve prediction accuracy. However, obtaining high-quality sensor data is often infeasible and costly, making alternative sources such as noon reports a viable option. In this paper, we propose a transfer learning-based approach for predicting vessels shaft power, where a model is initially trained on high-frequency data from a vessel and then fine-tuned with low-frequency daily noon reports from other vessels. We tested our approach on sister vessels (identical dimensions and configurations), a similar vessel (slightly larger with a different engine), and a different vessel (distinct dimensions and configurations). The experiments showed that the mean absolute percentage error decreased by 10.6 percent for sister vessels, 3.6 percent for a similar vessel, and 5.3 percent for a different vessel, compared to the model trained solely on noon report data.
Abstract (translated)
随着全球海上运输的增长，能源优化已成为降低运营成本和提高效率的关键。轴功率是从发动机传递到船轴的机械功率，直接关系到燃油消耗，因此准确预测轴功率是船舶性能优化的重要一步。能耗与船速、每分钟转数等船只参数以及天气和海况密切相关。频繁访问此类操作数据可以改善预测精度。然而，获取高质量的传感器数据往往难以实现且成本高昂，使得使用午报等替代来源成为一种可行的选择。 在本文中，我们提出了一种基于迁移学习的方法来预测船舶轴功率：首先在一个船只上用高频数据训练模型，然后利用其他船只每日午报中的低频数据对其进行微调。我们在姐妹船（相同尺寸和配置）、类似船（稍大且发动机不同）以及完全不同类型的船只上测试了这种方法。实验结果显示，在姐妹船上，平均绝对百分比误差减少了10.6%，在类似船舶上减少了3.6%，而在不同类型船舶上则减少了5.3%。相较于仅基于午报数据训练的模型而言，这一改进尤为显著。
URL
https://arxiv.org/abs/2510.03003
PDF
-
Adversarial Reinforcement Learning for Offensive and Defensive Agents in a Simulated Zero-Sum Network Environment
2025-10-03 05:53:51Abrar Shahid, Ibteeker Mahir Ishum, AKM Tahmidul Haque, M Sohel Rahman, A. B. M. Alim Al IslamAbstract
This paper presents a controlled study of adversarial reinforcement learning in network security through a custom OpenAI Gym environment that models brute-force attacks and reactive defenses on multi-port services. The environment captures realistic security trade-offs including background traffic noise, progressive exploitation mechanics, IP-based evasion tactics, honeypot traps, and multi-level rate-limiting defenses. Competing attacker and defender agents are trained using Deep Q-Networks (DQN) within a zero-sum reward framework, where successful exploits yield large terminal rewards while incremental actions incur small costs. Through systematic evaluation across multiple configurations (varying trap detection probabilities, exploitation difficulty thresholds, and training regimens), the results demonstrate that defender observability and trap effectiveness create substantial barriers to successful attacks. The experiments reveal that reward shaping and careful training scheduling are critical for learning stability in this adversarial setting. The defender consistently maintains strategic advantage across 50,000+ training episodes, with performance gains amplifying when exposed to complex defensive strategies including adaptive IP blocking and port-specific controls. Complete implementation details, reproducible hyperparameter configurations, and architectural guidelines are provided to support future research in adversarial RL for cybersecurity. The zero-sum formulation and realistic operational constraints make this environment suitable for studying autonomous defense systems, attacker-defender co-evolution, and transfer learning to real-world network security scenarios.
Abstract (translated)
本文通过一个自定义的OpenAI Gym环境，开展了一项关于网络安全中对抗强化学习的控制研究。该环境模拟了针对多端口服务的暴力破解攻击和响应式防御机制。环境中包含了现实中的安全权衡因素，包括背景流量噪音、逐步渗透机制、基于IP地址的规避战术、蜜罐陷阱以及多层次的速率限制防御措施。 利用深度Q网络（DQN）在零和奖励框架内对竞争双方——进攻者与防守者进行了训练。在这个框架中，成功的攻击会带来大量的最终奖励，而每次行动则会产生小额成本。通过系统性的评估，在不同的配置下（包括陷阱检测概率、攻击难度阈值以及不同培训方案），结果表明防守者的观察能力和蜜罐的有效性为成功攻击设置了巨大的障碍。 实验揭示了在这一对抗环境中，奖赏塑形和谨慎的训练调度对于学习稳定性至关重要。防守者在整个超过50,000次以上的训练过程中始终保持着战略优势，并且当暴露于包括自适应IP封锁和端口特定控制在内的复杂防御策略时，其性能显著提升。 本文提供了详尽的实施细节、可重复的超参数配置以及架构指南，以支持未来在网络安全对抗强化学习领域的研究工作。零和公式化和现实操作限制使得该环境非常适合用于研究自主防御系统、攻防共进化及向实际网络安全场景中的迁移学习。
URL
https://arxiv.org/abs/2510.05157
PDF
-
ExplainRec: Towards Explainable Multi-Modal Zero-Shot Recommendation with Preference Attribution and Large Language Models
2025-10-03 02:09:41Bo Ma, LuYao Liu, ZeHua Hu, Simon LauAbstract
Recent advances in Large Language Models (LLMs) have opened new possibilities for recommendation systems, though current approaches such as TALLRec face challenges in explainability and cold-start scenarios. We present ExplainRec, a framework that extends LLM-based recommendation capabilities through preference attribution, multi-modal fusion, and zero-shot transfer learning. The framework incorporates four technical contributions: preference attribution tuning for explainable recommendations, zero-shot preference transfer for cold-start users and items, multi-modal enhancement leveraging visual and textual content, and multi-task collaborative optimization. Experimental evaluation on MovieLens-25M and Amazon datasets shows that ExplainRec outperforms existing methods, achieving AUC improvements of 0.7\% on movie recommendation and 0.9\% on cross-domain tasks, while generating interpretable explanations and handling cold-start scenarios effectively.
Abstract (translated)
URL
https://arxiv.org/abs/2511.14770
PDF
-
Automatic Speech Recognition for African Low-Resource Languages: A Systematic Literature Review
2025-10-01 17:36:06Sukairaj Hafiz Imam, Tadesse Destaw Belay, Kedir Yassin Husse, Ibrahim Said Ahmad, Idris Abdulmumin, Hadiza Ali Umar, Muhammad Yahuza Bello, Joyce Nakatumba-Nabende, Seid Muhie Yimam, Shamsuddeen Hassan MuhammadAbstract
ASR has achieved remarkable global progress, yet African low-resource languages remain rigorously underrepresented, producing barriers to digital inclusion across the continent with more than +2000 languages. This systematic literature review (SLR) explores research on ASR for African languages with a focus on datasets, models and training methods, evaluation techniques, challenges, and recommends future directions. We employ the PRISMA 2020 procedures and search DBLP, ACM Digital Library, Google Scholar, Semantic Scholar, and arXiv for studies published between January 2020 and July 2025. We include studies related to ASR datasets, models or metrics for African languages, while excluding non-African, duplicates, and low-quality studies (score <3/5). We screen 71 out of 2,062 records and we record a total of 74 datasets across 111 languages, encompassing approximately 11,206 hours of speech. Fewer than 15% of research provided reproducible materials, and dataset licensing is not clear. Self-supervised and transfer learning techniques are promising, but are hindered by limited pre-training data, inadequate coverage of dialects, and the availability of resources. Most of the researchers use Word Error Rate (WER), with very minimal use of linguistically informed scores such as Character Error Rate (CER) or Diacritic Error Rate (DER), and thus with limited application in tonal and morphologically rich languages. The existing evidence on ASR systems is inconsistent, hindered by issues like dataset availability, poor annotations, licensing uncertainties, and limited benchmarking. Nevertheless, the rise of community-driven initiatives and methodological advancements indicates a pathway for improvement. Sustainable development for this area will also include stakeholder partnership, creation of ethically well-balanced datasets, use of lightweight modelling techniques, and active benchmarking.
Abstract (translated)
语音识别（ASR）在全球取得了显著进步，然而非洲低资源语言的代表性仍然严重不足，在拥有超过2000种语言的大洲上造成了数字包容性障碍。这项系统文献回顾（SLR）旨在探讨针对非洲语言的ASR研究，并重点关注数据集、模型和训练方法、评估技术、挑战以及未来方向的建议。我们采用PRISMA 2020程序，在DBLP、ACM Digital Library、Google Scholar、Semantic Scholar 和arXiv上搜索了从2020年1月至2025年7月间发表的研究。我们包括与非洲语言相关的ASR数据集、模型或指标的研究，同时排除非非洲的、重复的和低质量（评分<3/5）的研究。在2,062条记录中筛选出71篇符合要求的文章，并记录了涵盖11种不同语言共计约11,206小时语音的74个数据集。 研究发现，不到15%的研究提供了可重复使用的材料，且数据集许可情况不明朗。自我监督和迁移学习技术显示出潜力，但受限于预训练数据量不足、方言覆盖率低以及资源缺乏的问题。大多数研究人员使用单词错误率（WER），而较少采用语言学相关的指标如字符错误率（CER）或音调符号错误率（DER），因此在语调丰富和形态复杂的语言中应用有限。现有证据表明，ASR系统的性能受到数据集可用性差、注释质量低、许可不确定性以及基准测试不足等问题的限制。 尽管如此，社区驱动的倡议和技术方法的进步为改善现状提供了途径。这一领域的可持续发展还需包括利益相关者合作、建立伦理平衡的数据集、使用轻量级建模技术以及积极进行基准测试。
URL
https://arxiv.org/abs/2510.01145
PDF
-
Bridging the Gap Between Simulated and Real Network Data Using Transfer Learning
2025-10-01 14:29:47Carlos G\"uemes-Palau, Miquel Ferriol-Galm\'es, Jordi Paillisse-Vilanova, Albert L\'opez-Bresc\'o, Pere Barlet-Ros, Albert Cabellos-AparicioAbstract
Machine Learning (ML)-based network models provide fast and accurate predictions for complex network behaviors but require substantial training data. Collecting such data from real networks is often costly and limited, especially for critical scenarios like failures. As a result, researchers commonly rely on simulated data, which reduces accuracy when models are deployed in real environments. We propose a hybrid approach leveraging transfer learning to combine simulated and real-world data. Using RouteNet-Fermi, we show that fine-tuning a pre-trained model with a small real dataset significantly improves performance. Our experiments with OMNeT++ and a custom testbed reduce the Mean Absolute Percentage Error (MAPE) in packet delay prediction by up to 88%. With just 10 real scenarios, MAPE drops by 37%, and with 50 scenarios, by 48%.
Abstract (translated)
基于机器学习（ML）的网络模型能够提供复杂网络行为的快速且准确预测，但需要大量的训练数据。从实际网络中收集这些数据通常成本高昂并且受限，尤其是在像故障这样的关键场景下更是如此。因此，研究人员常常依赖于模拟数据，然而当这些模型部署到真实环境中时，其准确性会有所降低。我们提出了一种混合方法，利用迁移学习结合模拟和现实世界的数据。通过RouteNet-Fermi，我们展示了使用少量的真实数据集对预训练模型进行微调可以显著提高性能。我们在OMNeT++和自定义测试平台上进行的实验表明，在包延迟预测中平均绝对百分比误差（MAPE）最多可降低88%。仅用10个真实场景，MAPE就降低了37%，而使用50个场景则下降了48%。
URL
https://arxiv.org/abs/2510.00956
PDF
