检索结果-内蒙古大学图书馆

Depth Estimation using Weighted-loss and Transfer learning

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Hafeez, Muhammad Adeel Madden, Michael G. Sistu, Ganesh Ullah, Ihsan Machine Learning Research Group School of Computer Science University of Galway Ireland Insight SFI Research Centre for Data Analytics University of Galway Ireland Valeo Vision Systems Tuam Ireland

Depth estimation from 2D images is a common computer vision task that has applications in many fields including autonomous vehicles, scene understanding and robotics. The accuracy of a supervised depth estimation method mainly relies on the chosen loss function, the model architecture, quality of data and performance metrics. In this study, we propose a simplified and adaptable approach to improve depth estimation accuracy using transfer learning and an optimized loss function. The optimized loss function is a combination of weighted losses to which enhance robustness and generalization: Mean Absolute Error (MAE), Edge Loss and Structural Similarity Index (SSIM). We use a grid search and a random search method to find optimized weights for the losses, which leads to an improved model. We explore multiple encoder-decoder-based models including DenseNet121, DenseNet169, DenseNet201, and EfficientNet for the supervised depth estimation model on NYU Depth dataset v2. We observe that the EfficientNet model, pre-trained on ImageNet for classification when used as an encoder, with a simple upsampling decoder, gives the best results in terms of RSME, REL and log10: 0.386, 0.113 and 0.049, respectively. We also perform a qualitative analysis which illustrates that our model produces depth maps that closely resemble ground truth, even in cases where the ground truth is flawed. The results indicate significant improvements in accuracy and robustness, with EfficientNet being the most successful architecture. Copyright © 2024, The Authors. All rights reserved.

关键词： Signal encoding

Estimating the robustness of classification models by the structure of the learned feature-space

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Ho, Kalun Pfreundt, Franz-Josef Keuper, Janis Keuper, Margret CC-HPC Fraunhofer ITWM Fraunhofer-Platz 1 Kaiserslautern67663 Germany Institute for Machine Learning and Analytics Offenburg University Germany Data and Web Science Group University of Mannheim Germany

Over the last decade, the development of deep image classification networks has mostly been driven by the search for the best performance in terms of classification accuracy on standardized benchmarks like ImageNet. More recently, this focus has been expanded by the notion of model robustness, i.e. the generalization abilities of models towards previously unseen changes in the data distribution. While new benchmarks, like ImageNet-C, have been introduced to measure robustness properties, we argue that fixed testsets are only able to capture a small portion of possible data variations and are thus limited and prone to generate new overfitted solutions. To overcome these drawbacks, we suggest to estimate the robustness of a model directly from the structure of its learned feature-space. We introduce robustness indicators which are obtained via unsupervised clustering of latent representations from a trained classifier and show very high correlations to the model performance on corrupted test data. © 2021, CC BY.

关键词： Classification (of information)

learning Embeddings for Image Clustering: An Empirical Study of Triplet Loss Approaches

学校读者我要写书评

暂无评论

Learning Embeddings for Image Clustering: An Empirical Study...

International Conference on Pattern Recognition

作者： Kalun Ho Janis Keuper Franz-Josef Pfreundt Margret Keuper Data and Web Science Group University of Mannheim Germany Institute for Machine Learning and Analytics (IMLA) Offenburg University Germany Competence Center High Performance Computing Fraunhofer ITWM Kaiserslautern Germany

In this work, we evaluate two different image clustering objectives, k-means clustering and correlation clustering, in the context of Triplet Loss induced feature space embeddings. Specifically, we train a convolutional neural network to learn discriminative features by optimizing two popular versions of the Triplet Loss in order to study their clustering properties under the assumption of noisy labels. Additionally, we propose a new, simple Triplet Loss formulation, which shows desirable properties with respect to formal clustering objectives and outperforms the existing methods. We evaluate all three Triplet loss formulations for K-means and correlation clustering on the CIFAR-10 image classification dataset.

关键词： Correlation Pattern recognition Noise measurement Convolutional neural networks Image classification

Predicting Influential Higher-Order Patterns in Temporal Network data

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Gote, Christoph Perri, Vincenzo Scholtes, Ingo Data Analytics Group University of Zurich Zurich Switzerland Chair of Systems Design ETH Zurich Zurich Switzerland Chair of Machine Learning for Complex Networks University of Würzburg Würzburg Germany

Networks are frequently used to model complex systems comprised of interacting elements. While edges capture the topology of direct interactions, the true complexity of many systems originates from higher-order patterns in paths by which nodes can indirectly influence each other. Path data, representing ordered sequences of consecutive direct interactions, can be used to model these patterns. On the one hand, to avoid overfitting, such models should only consider those higher-order patterns for which the data provide sufficient statistical evidence. On the other hand, we hypothesise that network models, which capture only direct interactions, underfit higher-order patterns present in data. Consequently, both approaches are likely to misidentify influential nodes in complex networks. We contribute to this issue by proposing five centrality measures based on MOGen, a multi-order generative model that accounts for all indirect influences up to a maximum distance but disregards influences at higher distances. We compare MOGen-based centralities to equivalent measures for network models and path data in a prediction experiment where we aim to identify influential nodes in out-of-sample data. Our results show strong evidence supporting our hypothesis. MOGen consistently outperforms both the network model and path-based prediction. We further show that the performance difference between MOGen and the path-based approach disappears if we have sufficient observations, confirming that the error is due to overfitting. Copyright © 2021, The Authors. All rights reserved.

关键词： Forecasting

Watch your Up-Convolution: CNN Based Generative Deep Neural Networks are Failing to Reproduce Spectral Distributions

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Durall, Ricard Keuper, Margret Keuper, Janis Competence Center High Performance Computing Fraunhofer ITWM Kaiserslautern Germany Data- and Webscience Group University Mannheim Germany IWR University of Heidelberg Germany Institute for Machine Learning and Analytics Offenburg University Germany

Generative convolutional deep neural networks, e.g. popular GAN architectures, are relying on convolution based up-sampling methods to produce non-scalar outputs like images or video sequences. In this paper, we show that common up-sampling methods, i.e. known as up-convolution or transposed convolution, are causing the inability of such models to reproduce spectral distributions of natural training data correctly. This effect is independent of the underlying architecture and we show that it can be used to easily detect generated data like deepfakes with up to 100% accuracy on public benchmarks. To overcome this drawback of current generative models, we propose to add a novel spectral regularization term to the training optimization objective. We show that this approach not only allows to train spectral consistent GANs that are avoiding high frequency errors. Also, we show that a correct approximation of the frequency spectrum has positive effects on the training stability and output quality of generative networks. Copyright © 2020, The Authors. All rights reserved.

关键词： Convolution

Predicting Influential Higher-Order Patterns in Temporal Network data

学校读者我要写书评

暂无评论

Predicting Influential Higher-Order Patterns in Temporal Net...

International Conference on Advances in Social Network Analysis and Mining, ASONAM

作者： Christoph Gote Vincenzo Perri Ingo Scholtes Data Analytics Group University of Zurich Zurich Switzerland Chair of Systems Design ETH Zurich Switzerland Chair of Machine Learning for Complex Networks Julius-Maximilians-Universität Würzburg Würzburg Germany

关键词： Social networking (online) Computational modeling Complex networks Predictive models Length measurement data models Topology

Watch Your Up-Convolution: CNN Based Generative Deep Neural Networks Are Failing to Reproduce Spectral Distributions

学校读者我要写书评

暂无评论

Watch Your Up-Convolution: CNN Based Generative Deep Neural ...

Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Ricard Durall Margret Keuper Janis Keuper Competence Center High Performance Computing Fraunhofer ITWM Kaiserslautern Germany IWR University of Heidelberg Germany Data and Web Science Group University of Mannheim Germany Institute for Machine Learning and Analytics Offenburg University Germany

ISBN: (数字)9781728171685

ISBN: (纸本)9781728171692

关键词： Gallium nitride Convolution Distortion Neural networks Training Generative adversarial networks

A Network Perspective on the Influence of Code Review Bots on the Structure of Developer Collaborations

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Röseler, Leonore Scholtes, Ingo Gote, Christoph Data Analytics Group University of Zurich Zurich Switzerland Social Computing Group University of Zurich Zurich Switzerland Machine Learning for Complex Networks Julius-Maximilians-Universität Würzburg Würzburg Germany Systems Design ETH Zurich Zurich Switzerland

Background: Despite a growing body of literature on the impact of software bots on open source software development teams, their effects on team communication, coordination, and collaboration practices are not well understood. Bots can have negative consequences, such as producing information overload or reducing interactions between developers. Objective: The objective of this study is to investigate the effect of specific GitHub Actions on the collaboration networks of Open Source Software teams, using a network-analytic approach. The study will focus on Code Review bots, one of the most frequently implemented types of bots. Method: Fine-grained, time-stamped data of co-editing networks, developer-file contribution networks, as well as workflow runs and git commit logs will be obtained from a large sample of GitHub repositories. This will allow us to study how bots affect the collaboration networks of developers over time. By using a more representative sample of GitHub repositories than previous studies, which includes projects whose sizes span the whole range of Open Source communities, this study will provide generalizable results and updated findings on the general usage and distribution of GitHub Actions. With this study, we aim to contribute to advancing our knowledge of human-bot interaction and the effects of support tools on software engineering teams. Copyright © 2023, The Authors. All rights reserved.

关键词： Open source software

Hybridizing Target- and SHAP-encoded Features for Algorithm Selection in Mixed-variable Black-box Optimization

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Dietrich, Konstantin Prager, Raphael Patrick Doerr, Carola Trautmann, Heike Big Data Analytics in Transportation TU Dresden Germany ScaDS.AI Dresden Germany Data Science: Statistics and Optimization University of Münster Germany Sorbonne Université CNRS LIP6 Paris France Machine Learning and Optimisation Paderborn University Germany Data Management and Biometrics Group University of Twente Netherlands

Exploratory landscape analysis (ELA) is a well-established tool to characterize optimization problems via numerical features. ELA is used for problem comprehension, algorithm design, and applications such as automated algorithm selection and configuration. Until recently, however, ELA was limited to search spaces with either continuous or discrete variables, neglecting problems with mixed variable types. This gap was addressed in a recent study that uses an approach based on target-encoding to compute exploratory landscape features for mixed-variable problems. In this work, we investigate an alternative encoding scheme based on SHAP values. While these features do not lead to better results in the algorithm selection setting considered in previous work, the two different encoding mechanisms exhibit complementary performance. Combining both feature sets into a hybrid approach outperforms each encoding mechanism individually. Finally, we experiment with two different ways of meta-selecting between the two feature sets. Both approaches are capable of taking advantage of the performance complementarity of the models trained on target-encoded and SHAP-encoded feature sets, respectively. Copyright © 2024, The Authors. All rights reserved.

关键词： Optimization